ECOLOGY AND STRATEGY
i
ADVANCES IN STRATEGIC MANAGEMENT Series Editor: Joel A. C. Baum Recent Volumes Volume 15:
Disciplinary Roots of Strategic Management Research
Volume 16:
Population-Level Learning and Industry Change
Volume 17:
Economics Meets Sociology in Strategic Management
Volume 18:
Multiunit Organization and Multimarket Strategy
Volume 19:
The New Institutionalism in Strategic Management
Volume 20:
Geography and Strategy
Volume 21:
Business Strategy over the Industry Lifecycle
Volume 22:
Strategy Process
ii
ADVANCES IN STRATEGIC MANAGEMENT VOLUME 23
ECOLOGY AND STRATEGY EDITED BY
JOEL A. C. BAUM Rotman School of Management, University of Toronto, Canada
STANISLAV D. DOBREV Graduate School of Business, University of Chicago, USA
ARJEN VAN WITTELOOSTUIJN University of Groningen, The Netherlands
Amsterdam – Boston – Heidelberg – London – New York – Oxford Paris – San Diego – San Francisco – Singapore – Sydney – Tokyo JAI Press is an imprint of Elsevier
iii
JAI Press is an imprint of Elsevier The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands 525 B Street, Suite 1900, San Diego, CA 92101-4495, USA First edition 2006 Copyright r 2006 Elsevier Ltd. All rights reserved No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333; email:
[email protected]. Alternatively you can submit your request online by visiting the Elsevier web site at http://elsevier.com/locate/permissions, and selecting Obtaining permission to use Elsevier material Notice No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN-13: 978-0-7623-1338-9 ISBN-10: 0-7623-1338-2 ISSN: 0742-3322 (Series) For information on all JAI Press publications visit our website at books.elsevier.com Printed and bound in The Netherlands 06 07 08 09 10 10 9 8 7 6 5 4 3 2 1
iv
CONTENTS LIST OF CONTRIBUTORS
ix
INTRODUCTION: ECOLOGY VERSUS STRATEGY OR STRATEGY AND ECOLOGY? Stanislav D. Dobrev, Arjen van Witteloostuijn and Joel A. C. Baum
1
PART I: ENTERPRENEURSHIP BOOM AND BUST: THE EFFECT OF ENTREPRENEURIAL INERTIA ON ORGANIZATIONAL POPULATIONS Martin Ruef
29
OPTIMAL INERTIA: WHEN ORGANIZATIONS SHOULD FAIL Nick Dew, Brent Goldfarb and Saras Sarasvathy
73
PART II: TOP MANAGEMENT TEAMS TOP MANAGEMENT TEAM COMPOSITION AND ORGANIZATIONAL ECOLOGY: A NESTED HIERARCHICAL SELECTION THEORY OF TEAM REPRODUCTION AND ORGANIZATIONAL DIVERSITY Christophe Boone, Filippo C. Wezel and Arjen van Witteloostuijn
v
103
vi
CONTENTS
CEO TURNOVER IN THE NEW ERA: A DIALOGUE WITH THE FINANCIAL COMMUNITY Margarethe F. Wiersema and Thomas P. Moliterno
137
PART III: ORGANIZATIONAL CHANGE ECOLOGY, STRATEGY AND ORGANIZATIONAL CHANGE Jitendra V. Singh
177
THE BEST OF BOTH WORLDS: EXPLOITATION AND EXPLORATION IN SUCCESSFUL FAMILY BUSINESSES Danny Miller and Isabelle Le Breton-Miller
215
PART IV: ORGANIZATIONAL LEARNING IF IT DOESN’T KILL YOU: LEARNING FROM ECOLOGICAL COMPETITION Henrich R. Greve and Hayagreeva Rao
243
STRATEGIC RENEWAL AS IMPROVISATION: RECONCILING THE TENSION BETWEEN EXPLORATION AND EXPLOITATION Mary M. Crossan and David K. Hurst
273
PART V: TECHNOLOGY STRATEGY TECHNOLOGY CHOICE, TRANSACTION ALIGNMENT, AND SURVIVAL: THE IMPACT OF SUB-POPULATION ORGANIZATIONAL STRUCTURE Lyda S. Bigelow
301
vii
Contents
EXPLORING THE TAIL OF CREATIVITY: AN EVOLUTIONARY MODEL OF BREAKTHROUGH INVENTION Lee Fleming and Mark Szigety
335
PART VI: COMPETITIVE STRATEGY THE COMPETITIVE DYNAMICS OF VERTICAL INTEGRATION: EVIDENCE FROM U.S. MOTION PICTURE PRODUCERS, 1912–1970 Giacomo Negro and Olav Sorenson
363
DYNAMICS OF COMPETITIVE REPOSITIONING: A MULTIDIMENSIONAL APPROACH Javier Gimeno, Ming-Jer Chen and Jonghoon Bae
399
PART VII: COOPERATIVE STRATEGY FIGHTING A COMMON FOE: ENMITY, IDENTITY AND COLLECTIVE STRATEGY Jo-Ellen Pozner and Hayagreeva Rao WHEN DO NETWORKS MATTER? A STUDY OF TIE FORMATION AND DECAY Andrew V. Shipilov, Tim J. Rowley and Barak S. Aharonson
445
481
PART VIII: SCALE AND SCOPE SCALE AND SCOPE ECONOMIES IN THE BRITISH MOTORCYCLE INDUSTRY, 1899–1993 Filippo C. Wezel and Arjen van Witteloostuijn
523
viii
CONTENTS
DIVERSIFICATION TO ACHIEVE SCALE AND SCOPE: THE STRATEGIC IMPLICATIONS OF RESOURCE MANAGEMENT FOR VALUE CREATION Tim R. Holcomb, R. Michael Holmes Jr. and Michael A. Hitt
549
PART IX: INDUSTRY EVOLUTION DIRECT AND INDIRECT EFFECTS OF PRODUCT PORTFOLIO ON FIRM SURVIVAL IN THE WORLDWIDE OPTICAL DISK DRIVE INDUSTRY, 1983–1999 Olga M. Khessina
591
INDUSTRY PERFORMANCE AND CHANGES IN COMPETITOR CHARACTERISTICS: EVIDENCE ON ISOLATIONISM VERSUS MUTUAL FORBEARANCE Anita M. McGahan
631
LIST OF CONTRIBUTORS Barak S. Aharonson
University of Toronto, Toronto, Canada
Jonghoon Bae
Korea University, Seoul, Republic of Korea
Joel A. C. Baum
University of Toronto, Toronto, Canada
Lyda S. Bigelow
Washington University, St. Louis, MO, USA
Christophe Boone
University of Antwerpen, Antwerpen, Belgium
Ming-Jer Chen
University of Virginia, Charlottesville, VA, USA
Mary M. Crossan
University of Western Ontario, Ontario, Canada
Nick Dew
Naval Postgraduate School, Monterey, CA, USA
Stanislav D. Dobrev
University of Chicago, Chicago, IL, USA
Lee Fleming
Harvard University, Boston, MA, USA
Javier Gimeno
INSEAD, France
Brent Goldfarb
University of Maryland, College Park, MD, USA
Henrich R. Greve
Norwegian School of Management, Oslo, Norway
Michael A. Hitt
Texas A&M University, College Station, TX, USA
Tim R. Holcomb
Texas A&M University, College Station, TX, USA
R. Michael Holmes Jr.
Texas A&M University, College Station, TX, USA
David K. Hurst
University of Western Ontario, Ontario, Canada ix
x
LIST OF CONTRIBUTORS
Olga M. Khessina
Georgetown University, Washington, DC, USA
Isabelle Le Breton-Miller
University of Alberta, Edmonton, Alberta
Anita M. McGahan
Boston University, Boston, MA, USA
Danny Miller
HEC Montreal, Montreal, Canada
Thomas P. Moliterno
University of South Carolina, Columbia SC, USA
Giacomo Negro
Universita` L. Bocconi, Milano, Italy
Jo-Ellen Pozner
Northwestern University, Evanston, IL, USA
Hayagreeva Rao
Stanford University, Stanford, CA, USA
Tim J. Rowley
University of Toronto, Toronto, Canada
Martin Ruef
Princeton University, Princeton, NJ, USA
Saras Sarasvathy
University of Virginia, Charlottesville, VA, USA
Andrew V. Shipilov
INSEAD, France
Jitendra V. Singh
University of Pennsylvania, Philadelphia, PA, USA
Olav Sorenson
London Business School, London, UK
Mark Szigety
Harvard University, Boston, MA, USA
Filippo C. Wezel
Tilburg University, Tilburg, The Netherlands
Margarethe F. Wiersema
Rice University, Houston, TE, USA
Arjen van Witteloostuijn
University of Groningen, Groningen, The Netherlands
INTRODUCTION: ECOLOGY VERSUS STRATEGY OR STRATEGY AND ECOLOGY? Stanislav D. Dobrev, Arjen van Witteloostuijn and Joel A. C. Baum The motivating question in strategic management research is how firms can differentiate themselves from their competitors so that they can gain and sustain competitive advantages that earn them supernormal returns. Different theoretical frameworks offer varied guidance on how this task is to be accomplished. The influential five industry forces framework (Porter, 1980; McGahan & Porter, 1997), for example, posits that competitive advantage is endemic to industry structure and firms’ positioning. The resource-based perspective (Rumelt, 1984; Wernerfelt, 1984), in contrast, views differentiation based on unique access to scarce, valuable bundles of inimitable resources, rather than industry structure and positioning, as the chief source of competitive advantage. The core competence approach (Prahalad & Hamel, 1990; Dosi, Teece, & Winter, 1992) attributes competitive advantage to idiosyncratic technological and organizational capabilities that firms cultivate as their histories unfold. Alternatively, the game-theoretic strategic conflict approach (Shapiro, 1989; van Witteloostuijn, 2002) emphasizes firms’ anticipation of and responses to their rivals’ actions, as they try to outmaneuver each other.
Ecology and Strategy Advances in Strategic Management, Volume 23, 1–26 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23001-8
1
2
STANISLAV D. DOBREV ET AL.
Despite their differences, these and other strands of strategic management theory focus on the alignment (and realignment) of internal strengths and weaknesses with external opportunities and threats; organizational flexibility is key to sustaining a competitive advantage and deriving economic returns that exceed their total costs (including capital costs) despite the tendency for profits to be competed away. Given this orientation, and the attention to financial returns, strategy research tends to have a pronounced focus on application, emphasizing the development of concrete prescriptions for corporate leaders based typically on insights from successful firms; the most basic method in strategic management is to observe the winners and attempt to discern what makes them win. The goal is to explain the success of a strategy, before it has become standard operating procedure – when firms can still benefit from adopting it. By contrast, organizational ecology, a research paradigm within the broad field of organizational sociology, seeks to explain variance in behavior across diverse populations of organizations and over time, typically using multivariate models that control for diverse potential causes. Ecological theory emphasizes adaptation through environmental selection at the population or industry level, rather than by individual organizations. These evolutionary dynamics are seen as favoring structurally inert organizations. Inertia is not only a survival-enhancing feature, but also a by-product of prior success and a consequence of selection. Organizational change that disrupts inertia is expected to diminish organizational performance. Thus, ecological accounts of strategy associate organizational inertia with superior performance, and are concerned with explaining the contexts in which one strategy or another becomes defined as effective and diffuses across the organizational landscape. Despite the fact that the fields of strategy and ecology share a focus on organizational performance, dialogue between them has been relatively rare. In part, this is because the theoretical and empirical foundations and focuses of the two groups remain at a distance. Occasionally (and in our view unfairly), the focus of ecological theory on explaining how context and history contribute to strategic management trends (oftentimes after those trends have not only come, but also gone) has lead strategic management theorists to criticize the ideas of ecologists as being ‘‘esoteric’’ or even ‘‘irrelevant.’’ We believe, however, that the rewards of ecological researchers’ adherence to the original paradigm are now ripe for harvesting. Theoretical insights developed in organizational ecology over the past 30 years offer a cumulative and integrated body of knowledge about organizational success and survival ready to be fruitfully extended to other areas
3
Introduction
of organizational studies – particularly the field of strategic management. Some scholars from each group have begun to recognize and exploit these possibilities. By presenting some of the very best of this work – from both ecology and strategy sides of the fence – we hope to fortify these linkages and bridges, and foster a more constructive dialogue and collaboration.
STRUCTURE OF THE VOLUME At its core, this volume tackles the contradictory views of the performanceenhancing effects of organizational flexibility and inertia head on, and in doing so, contributes to the development of theory and empirical evidence at the interface of strategic management and organizational ecology. In addition to the inertia–flexibility nexus, the volume explores a wide range of additional connections between these two perspectives across nine topical areas that both ecological and strategic management researchers have examined: (1) Entrepreneurship, (2) Top Management Teams, (3) Organizational Change, (4) Organizational Learning, (5) Technology Strategy, (6) Competitive Strategy, (7) Cooperative Strategy, (8) Scale and Scope, and (9) Industry Evolution. In each topical area, we present a pair of articles, one authored by researchers working primarily within the strategic management tradition, and the other by researchers whose work is influenced more strongly by an ecological orientation. In each pairing, readers can make a head-to-head comparison of the two approaches. Side by side, the paired articles show that strategy and ecology share a number of common subjects, and while distinct, often complementary ways of approaching them. To further illuminate differences and complementarities in approach, we initiated an exchange of papers among authors, and the published papers include their reflections on the relationship between the strategic and ecological perspectives in general and the paired papers in particular. To be sure, some of the contentiousness and rivalry between the fields shows up. But the authors also recognize the limitations of their own approach, the strengths of the other, and all favor a more productive dialogue. This volume, thus, aims to bring strategic and ecological perspectives closer together to join forces for further theory development in areas of mutual interest. By creating a forum for discussing key issues at the hearts of both strategy and ecology, we provide a compendium that contributes to the cross-fertilization of these perspectives and, we hope, acts as a catalyst for future research at this important interface. In our view, such a synthesis may
4
STANISLAV D. DOBREV ET AL.
not only increase the explanatory power of models seeking to understand variance in performance among organizations, but also to generate new ideas in the recombinant manner that underlies all learning processes. Our emphasis in this introduction is on potential cross-pollination from ecology to strategy. This is not because we think that crossing the border from the other side – from strategy to ecology – is less promising. Certainly it is not (Boone, Carroll, & van Witteloostuijn, 2004). Rather, it is because we believe that ecology to strategy cross-pollination is relatively unexplored. We devote the remainder of the introduction to two tasks. First, to set the scene, we illustrate some of the fundamental differences between ecology and strategy traditions in greater detail. Second, we set the stage for the chapter-pairs that follow by briefly summarizing the state of the art in several of the well-developed areas in organizational ecology to illustrate how their insights can inform arguments that are central in the strategic management literature and to highlight advantages that could be derived from further exploring the ecology–strategy interface. Note that we decided not to introduce each and every chapter-pair in detail. Rather, we refer to chapters along the way, linking specific chapters to some of the key areas of cross-pollination that we consider promising. In the chapter-pairs themselves, many more interesting cross-fertilization examples are revealed.
CONTRASTING ECOLOGY AND STRATEGY More often than not, ecology and strategy are introduced as contrasting perspectives. Particularly in the strategy field, there is a tendency to set ecology apart as an odd, specialist field not worthy of too much attention. Table 1 contrasts ecology and strategy in the reductionist manner implied by this treatment of ecology in the strategy field. The central alleged contrast is adaptation versus selection, from which many of the other contrasts follow. Here, the argument goes, ecology’s focus on population-level selection processes through indirect competition is very different from strategy’s, because it defines away the role of organizationlevel adaptation by adopting the assumption of (relative or structural) organizational inertia. In so doing, ecology takes the organization to be a black box, studying the survival rates of different types of organizations over the very long run in search for universal laws driving industry evolution. As by-products, ecology is seen as a monodisciplinary approach, heavily rooted in bio-ecological Darwinian sociology, with little – if any – managerial relevance.
5
Introduction
Table 1. Ecology Selection Inertia Black box Population level Indirect competition Survival performance Universal laws Monodisciplinary Longer run Managerial irrelevance
Ecology versus Strategy. Stratergy Adaptation Flexibility White box Firm level Direct competition Financial performance Contingency conditions Multidisciplinary Shorter run Managerial relevance
None of these characterizations finds its mark, with the exception, perhaps, of the issue of managerial irrelevance. Ecology’s inertia is a relative concept (Hannan & Freeman, 1984; Ruef, 1997); adaptation and learning are not excluded – just not taken for granted (e.g., Barnett & Carroll, 1995; Greve & Rao, this volume; Singh, this volume). The black box of the organization is opened by considering a variety of organizational features (e.g., Pennings, Lee, & van Witteloostuijn, 1998; Dobrev, Kim, & Carroll, 2003). Organization-level change studies abound (e.g., Singh, House, & Tucker, 1986; Baum & Singh, 1996; Dobrev, 1999; Swaminathan & Dowell, 2000; Barnett & Freeman, 2001; Sorenson, 2003). Direct competition is incorporated in niche overlap and localized competition models (e.g., Baum & Mezias, 1992; Baum & Singh, 1994a, 1994b, 1996; Baum & Haveman, 1997; Dobrev & Kim, 2006). Financial performance can easily be incorporated (e.g., Henderson, 1999; Boone, Carroll, & van Witteloostuijn, 2004). Many of ecology’s ‘‘universal’’ laws are conditional (e.g., Barron, 1999; Dobrev, 2000; Negro & Sorenson, this volume; Ruef, this volume; Wezel & van Witteloostuijn, this volume). Insights from other disciplines are frequently incorporated (e.g., Hannan, Po´los, & Carroll, 2006; Bigelow, this volume; Pozner & Rao, this volume) and ecological processes materialize in the short run as well as long run (e.g., Dobrev, 2001). To date, however, ecologists have, by and large, refrained from translating their arguments and results in ways that facilitate managerial applicability. Ecologists see inertial constraints on the capacity of individuals to change existing organizations successfully. Individuals cannot always (or often) determine in advance which variations will succeed. And even if they do, they will have difficulty changing their organizations’ strategies and structures quickly enough to keep pace with the demands of an uncertain,
6
STANISLAV D. DOBREV ET AL.
changing environment. The existence of these constraints does not mean that individuals are irrelevant to processes of organizational change, however. Rather, it means the capacity of individuals to change their organizations successfully is of great importance (see Boone et al., this volume; Wiersema & Moliterno, this volume). Indeed, it is in the absence of such challenges that individuals would not matter. It is time for the field of strategic management to move beyond ‘‘strawman’’ characterizations of three decades of sustained ecological research, and to begin to explore how ecological arguments can be turned into insights with high managerial relevance. It is time for ecologists to pay increased attention to translating their arguments and results in ways that facilitate managerial applicability. We begin the process by briefly discussing three areas of ecological research ripe for cross-pollination with strategic management: (1) demography of competition, (2) positional advantages, and (3) inertia and change. In subsequent chapter-pairs, these and other examples of cross-fertilization are explored in much more detail in each of the nine topical areas.1
DEMOGRAPHY OF COMPETITION Ecological theory is rich in ideas about how the demographic makeup of an industry in terms of its constituent organizations affects these organizations’ chances of success and survival. In particular, models of organizational age and size specify a variety of mechanisms that can either build or destroy a sustainable competitive advantage (e.g., Carroll, Bigelow, Seidel, & Tsai, 1996; Barnett, 1997; Dobrev, Kim, & Solari, 2004).
Firm Age: Liabilities of Newness, Senescence, Obsolescence and Adolescence Organizational age is not a central issue in strategy research. However, directly related to the capability-based view of competitive advantage are four ‘‘liabilities of aging’’ articulated in organizational ecology. The first is the liability of newness (Stinchcombe, 1965; Freeman, Carroll, & Hannan, 1983): young organizations lack the routines and experience necessary to underpin the development of capabilities. As firms age, they produce track records that serve as guideposts for external constituents to judge prior performance. Internally, the passage of time generates trust among internal
Introduction
7
members, which reduces conflict and thus improves the routinization of vital activities. Newness may also negatively affect an organization’s competitive standing in industries where status and reputation are important sources of positional advantage because the accrual of favorable relative standing visa`-vis competitors requires a track record as basis for such claims. A second organizational age-based prediction is the liability of senescence (Ranger-Moore, 1997). It posits that older organizations suffer deteriorated performance due to the ossifying effect of growing bureaucratization. The proliferation of structure by way of expanding architecture or overlay of new and existing policies and work processes constrains the potential for collective action that is needed to successfully implement an organization’s strategy. Akin to processes of sclerosis in animals, the organizational space that is needed for learning and innovation gets crowded out by demands from the center – slack resources that maintain the variance of outcomes become inefficiencies. The mechanism behind the liability of senescence underpins the erosion of capabilities among dominant incumbents often observed by management researchers. A third argument that also suggests the possibility of performance decline with age is the liability of obsolescence (Carroll, 1983; Baum, 1989). Unlike liability of senescence, which pertains to intra-organizational processes that develop with time, liability of obsolescence asserts that an organization’s capability or core competence faces increasing prospects of becoming outdated, or obsolete, as time passes and demands change since the capability was developed. Although essentially an argument about misalignment between organizational capabilities and environmental demands, it does not simply assume organizations’ inability to maintain sufficient malleability to recover expeditiously when the vital fit between organization and environment is broken. Instead, the liability of obsolescence story relies on the mechanism of imprinting (Stinchcombe, 1965), in which prevalent environmental conditions at the time of an organization’s founding get markedly imprinted in its capabilities. With time, as environmental conditions change, the organization-environment fit deteriorates because of the tendency of organizational capabilities to persist and continuously reproduce. The final liability of aging – the liability of adolescence – relates variance in organizational success and survival to the initial stock of resources and goodwill that new entrants possess (Bruderl & Schussler, 1990). The main prediction is that organizations’ chances of success are relatively high in the early years, while they can still rely on their initial endowments. Once such initial endowments are exhausted, organizations’ vulnerabilities are exposed and their risk of failure increases at adolescence. Access to critical resources
8
STANISLAV D. DOBREV ET AL.
in the early years when organizations are recombining activities and ideas in search of the right mix of capabilities that will eventually come to constitute their competitive advantage is paramount. Combining the liability of adolescence with the effect of imprinting, the density-delay model (Carroll & Hannan, 1989) in organizational ecology predicts that organizations founded into crowded markets where resource availability is tight develop routines for weathering scarcity rather than for managing growth, and such capabilities persist over time even if they later encounter greater resource abundance. As this brief review of liabilities of aging suggests, their combined application to a specific case is difficult not only because of the extensive datasets needed to test their predictions, but also because their predictions appear largely incompatible – is an organization more likely to perform better when it is young, old, or adolescent? To address this question, ecologists have pursued a deliberate strategy of integration, which has yielded empirical studies identifying contingencies on the liabilities particular organizations are most likely to face (e.g., Hannan, Carroll, Dobrev, Han, & Torres, 1998; Henderson, 1999) as well as logical formalizations that rely on non-monotonic logic to reconcile and integrate these predictions (Hannan et al., 2006, ch. 12). These approaches treat the four liabilities of aging as complementary rather than as competing organizational processes (Baum, 1996). Given the clear relevance of age-dependence theories in ecology for strategic management research on sources and sustainability of competitive advantage (Henderson, 1999; Khessina, this volume), we think an excellent opportunity for theoretical import exists.
Firm Size: Absolute and Relative Effects Empirical analyses of competing organizations suggest strongly that no single factor exerts a more pervasive impact than that of organizational size. Size plays a major role in most empirical research on organizations in strategic management – from studies of executive compensation (Main, O’Reilly, & Wade, 1995) to internal organizational demography (Finkelstein & Hambrick, 1990) to multimarket rivalry (Baum & Korn, 1999) to strategic alliances (Stuart, 2000). Size effects are so strong and ubiquitous that researchers studying a wide range of organizational outcomes (e.g., performance, innovation, growth, demography, and change) commonly regard size as a mandatory covariate for any theoretical argument. This approach reflects a long-standing view that ‘‘scale’’ is conducive to the development of
Introduction
9
superior capabilities and market positions (see Holcombe et al., this volume; Wezel & van Witteloostuijn, this volume). Despite this widespread recognition of its importance, little contemporary research in strategic management focuses attention on size; typically, size serves only as a control. In contrast, ecological research has examined a range of far more nuanced and complex size effects (e.g., Barnett, 1997). Theoretical arguments about size in organizational ecology can be divided into two types, those about absolute effects and those about effects relative to other relevant organizations (Hannan et al., 1998). An example of the first type is the claim that large organizations gain efficiencies from an expanded division of labor and unit specialization. As Schumpeter (1950) asserts, large firms excel at producing new knowledge because they are better fit to routinize innovation internally. So too is Levinthal’s (1991) claim that small organizations operate near an ‘‘extinction boundary,’’ whereby a random shock that merely inconveniences a large organization can destroy a small one. In economic terms, the extinction boundary can be thought of as the minimum efficient scale. An example of the second type is Pfeffer and Salancik’s (1978) resource dependence theory, which sees larger organizations as possessing leverage over smaller parties due to size asymmetry. Similar logic underlies Porter’s (1980) five industry forces framework, as well as many other studies in the strategic management literature. Consider the possible effects of relative size within a specific industry. Advantages of several kinds likely accrue from an organization’s dominant position in the size distribution of an industry at any given point in time. For example, the largest organizations may receive favorable treatment from regulators and other officials. These actions may well be justified in terms of welfare effects arising from the employment bases of very large firms. Conventional industrial-organization economics holds that scale economies give organizations operating above minimum efficient scale a potentially strong structural advantage (producing at lower average cost) over smaller ones. Scale advantages in cost might arise from many sources, including internal efficiency as well as leverage in supplier and buyer markets, especially talented labor where employees may be attracted to the higher pay and security of larger firms. Larger firms might also benefit from scope economies, gaining from synergies leading to new value creation and resulting from integration of certain activities within the organization. So, for both sociopolitical and economic reasons, relative size confers a positional competitive advantage (see Holcombe et al., this volume; Wezel & van Witteloostuijn, this volume). Increasing returns to scale within an industry imply that, in the long run, small organizations cannot win head-to-head competition with larger ones.
10
STANISLAV D. DOBREV ET AL.
But, in most markets and industries not all competition is confined to scale. Strategy theory and research suggests that smaller organizations often find ways to avoid the pressures of deleterious scale competition (Porter, 1980). One way is by exploiting market segments that are too small to be exploited profitably by very large organizations (Dobrev, 2000). Another way comes from specializing so as to produce goods and services whose appeal comes from socially constructed images related to perceived producer status or authenticity (Carroll & Swaminathan, 2000). A third comes from focusing on a particular customer segment and tailoring products to them, including through customized production (Boone, Broecheler, & Carroll, 2000). The viability of any of these strategies depends on the structure of the market and the organizational community. Translating and integrating these arguments into empirical models, organizational ecologists have shown that the effects of absolute and relative size are independent of one another, and also vary with organizational age (Hannan et al., 1998). These specifications are complicated, but choosing from the various arguments on an ad hoc basis is unlikely to produce an accurate explanation. Importantly, ecological theories of relative size effects specify both the set of comparison organizations and the mechanisms underscoring the importance of relations between the organizations within the set. For example, the size-localized competition model (Hannan & Freeman, 1977; Baum & Mezias, 1992) posits that organizations compete most intensely with organizations of similar size. Size-localized competition is a theoretical mechanism that provides detailed guidance about how relative size effects should be formulated or weighted in a model. The mechanism assumes that the strongest competition occurs in tightly packed regions of the size distribution. Relatedly, the scale-based selection model (Dobrev & Carroll, 2003) assesses organization-specific competitive intensity by examining the size structure of the competitive environment faced by each organization at any point in time. In particular, the scale-based competitive pressure faced by an organization depends both on the number of larger competitors it faces (who each hold a scale advantage over the focal organization) as well as on the distance of each larger competitor from the focal organization on the scale dimension (with distance reflecting the extent of the advantage). By specifying a mechanism through which size effects obtain and by indicating a comparison set, size-localized and scale-based competition models provide a rigorous way to deal with the potentially complex and dynamic effects of organizational size (Baum, 1995; McGahan, this volume). Given the pervasive impact of organizational size recognized in strategic management research, we think ecological models of size-localized and
11
Introduction
scale-based competition offer a great opportunity to develop a more central theoretical role for size within any major strategic management perspective – positioning, resource-based, capabilities, or conflict. These models, which emphasize the dynamic component of direct competition, indicating how an organization’s position vis-a`-vis all other organizations in the population co-determines performance, also hold great potential for extending and rendering more dynamic-related models of strategy management, such as Porter’s (1980) five forces framework (see Boone, Carroll, & van Witteloostuijn, 2004; Boone & van Witteloostuijn, 2004).
POSITIONAL ADVANTAGES Ecological theory is also rich in ideas about how ecological processes, more generally, affect an organization’s positional (dis)advantages. The key argument is that organizations are located in a larger, multidimensional demographic space. For example, organizations can also be characterized in terms of their products or technologies (e.g., Khessina, this volume; Wezel & van Witteloostuijn, this volume). This relates to the concept of the niche. A niche is a segment in a population’s resource space. A population features a specific, often multidimensional, niche structure. A key question is how the organizations’ demographic space maps onto this niche structure, and how this impacts upon organizational performance. Within this niche structure, an organization is characterized by its position vis-a`-vis all of its competitors. Related concepts in this context are niche (non)overlap and niche width. Using this set of related concepts – niche width, niche (non)overlap and niche position – ecology analyzes the advantages or disadvantages of an organization’s position in the broader competitive environment (for a review, see Dobrev, 2005). Niche Width Though the terms organizational niche and firm scope are largely synonymous, a subtle but important difference remains: the term niche, borrowed from bio-ecology, articulates the causal primacy of environmental resources on which organizations depend over other strategic choices made by organizations in positioning themselves in order to secure those resources. Niche width refers to an organization’s variance in resource utilization (Hannan & Freeman, 1989). This representation proves useful for classifying organizations in terms of wide-niche generalists (or K-strategists) and narrow-niche specialists (or r-strategists).
12
STANISLAV D. DOBREV ET AL.
Much strategic analysis builds on the simple notion that for firms to grow, expand, and increase profitability, they must diversify (Chandler, 1962; Steiner, 1964; Holcombe et al., this volume). Entry into new markets, products, or services not only allows the firm to tap potentially unexplored resources, but also reduces risk by spreading it over several operational domains. Additionally, diversification generates economies of scale and scope across similar functions in the separate lines of business. These advantages of a generalist strategy (i.e., broad scope) articulated in strategic management tend to overlook the factors including timing in the industry evolution, fluctuation in resource flows, market concentration, population-level diversity, and shifts in the organizational age and size distributions that organizational ecologists have examined (e.g., Dobrev, Kim, & Hannan, 2001). Environmental processes have been emphasized in work on organizational niche theory, which specifies the configuration of environmental conditions under which the generalist strategy conveys advantage. In particular, niche theory (Freeman & Hannan, 1983) posits that generalists are favored when the pattern of environmental change exhibits high variation and coarse grain, but not when the pattern is high variation and fine grain. Variation refers to environmental fluctuations around the mean over time. Grain refers to the patchiness of these fluctuations with frequent variations termed fine-grained and periodic ones termed coarse-grained. So, although both strategy and ecology research concludes that a firm’s performance generally increases with the broadening of its scope, ecological analyses elaborate the mechanisms behind this prediction in more detail, and also specify the cases where scope is not beneficial (i.e., high variation, fine grain). These basic ideas have enabled organizational ecologists to go on to develop more comprehensive models of industry evolution including resource partitioning theory (Carroll, 1985) and niche evolution theory (Dobrev, Kim, & Carroll, 2002). Consider resource partitioning theory (Carroll, 1985). By the logic of many strategic management theories, the dominance of large organizations in an industry should hinder the emergence and operation of small specialist organizations. Yet, in modern economies, a variety of industries display simultaneous trends of increased concentration and specialist proliferation. Resource partitioning theory views these trends as interdependent. The theory holds that under certain environmental and organizational conditions, the increased dominance of large organizations in an industry will enhance the life chances of specialist organizations. The resource partitioning model predicts that small entrants populate the market periphery following the failure of organizations in the market center. Since surviving large-scale
13
Introduction
generalist competitors take over some but not all of the market space vacated by exiting competitors, small specialist organizations can thrive on these unexhausted resource patches that open up once consolidation in the market center occurs. Four different mechanisms can produce resource partitioning: market location, customization, anti-mass-production sentiment, and conspicuous status consumption (Carroll, Dobrev, & Swaminathan, 2002). In resource partitioning theory, the effect of organization’s scope on its performance thus varies as the market evolves and concentrates. The theory’s insight derives from comparing the environmental resource space available for specialists as overall market concentration rises. Almost all variation in market concentration derives from generalist crowding and consolidation. As the competitive struggle for scale among generalists proceeds the size and target breadth of the survivors increase, but the combined resources held by generalist organizations declines, increasing the space available for specialists and their viability. Niche evolution theory (Dobrev et al., 2002) thus predicts that the overall positive effect of firm scope on performance is reversed in concentrated markets. So, niche width and resource partitioning theory suggest how an organization’s position in resource space may be associated with performance. On the one hand, niche width theory emphasizes the role of resource space change, specifying the change conditions that favor generalists or specialists. In strategy terms, this relates to the positional advantages of the extent of product differentiation. On the other hand, resource partitioning theory focuses on the impact of the shape of the resource space. In markets with a peaked resource space, with a viable mass-abundant center and resourcescarce periphery, generalist concentration is positively associated with specialist performance. Here, generalists are defined by their broader niche and location in the center, and specialists by their smaller niche and position in the periphery. Hence, resource partitioning theory adds to strategy the insight that under particular conditions concentration offers opportunities for niche entry or strategic repositioning.
Niche Overlap and Non-Overlap Another aspect of an organization’s position, next to niche width, is the extent to which its niche span overlaps, or does not so, with its competitors’. This is captured by the niche overlap construct. In strategy terms, a crucial consideration when assessing the relationship between scope and performance pertains to the strong connection between an organization’s niche
14
STANISLAV D. DOBREV ET AL.
width and the degree to which its niche overlaps the niches of other organizations. Niche overlap theory carefully assesses the advantages of niche overlap and niche non-overlap, arguing how the balance between the pros and cons of a broad or a narrow scope work out under particular competitive circumstances (see Gimeno et al., this volume; Negro & Sorenson, this volume; Pozner & Rao, this volume; Shipilov et al., this volume; Wezel & van Witteloostuijn, this volume). On the one hand, when an organization broadens its niche, it cannot decrease overlaps and it generally increases them. So a complete analysis of the consequences of niche width must take account of the excess overlap entailed by broad-niche strategies. Arguments about the pattern of competitive intensity triggered by niche crowding build on ideas about localized competition (Baum & Singh, 1994a, b, 1996; Podolny, Stuart, & Hannan, 1996) and density-dependent competition (Hannan & Carroll, 1992; Dobrev et al., 2001). Whereas niche overlap reflects the potential for competition, the extent of non-overlap captures the degree to which other organizations’ resource requirements do not overlap with those of the focal organization, and so are potentially complementary (Baum & Singh, 1994a, b, 1996). Generally, the intensity of competition experienced by an organization in a market increases with the extent of its niche overlaps, while its non-overlaps may produce mutualistic benefits arising from complementarities like mutual forbearance (Baum & Korn, 1996). The concept of niche overlap is helpful in theorizing a wide range of competitive and mutualistic processes arising from both symbiotic and complementary organizational interdependence. Such insights have many interesting implications for theories of strategic (re)positioning, vertical integration, diversification, organizational learning, and multipoint competition (see Gimeno et al., this volume; Greve & Rao, this volume; McGahan, this volume; Negro & Sorenson, this volume). Moreover, since firms are prone to maneuver for market position in search of improved access to resources and lower competitive intensity, ecological models based on niche theory can inform predictions about state effects (overall appeal of origin and destination states) in studies of organizational transformation (Dobrev & Kim, 2006).
INERTIA AND CHANGE At the heart of the ecology–strategy contrast is the debate surrounding the performance impact of flexibility (strategy) versus inertia (ecology). It is in this respect that ecology has been misinterpreted most. And, as a direct
15
Introduction
result, that ecology’s rich theoretical and deep empirical insights into the conditional performance consequences of organizational change have yet to influence mainstream strategy research.
Relative or Structural Inertia Without doubt, from a strategic management perspective, the most controversial element of ecological theory is the concept of inertia. In ecology, selection and adaptation models of organizational change are often considered in terms of the differential consequences brought about by changes in the organizational core versus the periphery. Because reliability and accountability (the two most broadly sought after organizational capabilities according to ecological theory) emerge from the reproducibility of core structures, inertial forces (along with the selection advantage derived from reliability and accountability) emanate from core features of organizations. Hannan and Freeman’s (1984) original definition gives a hierarchical list of four core organizational features, including an organization’s mission, its authority structure, its technology, and its marketing strategy. Empirical applications of the inertia theory using this definition of organizational core are not unequivocal in their interpretations of core structures. A typical research design of this sort usually begins by defining what organizational features constitute the core based on Hannan and Freeman’s original definition, and then formulates predictions about transformations in those features. Nevertheless, many demographic studies do find that core structural change is a precarious process; it leads to an elevated probability of performance decline, at least temporarily, even if the desired end state is on target. While organizations in these studies do not necessarily fail as a result of their efforts to change, they do not necessarily improve their organizational survival chances either. Changes affecting the non-core or periphery structure do not produce the same outcome; they might even lead to improved performance (for reviews, see Barnett & Carroll, 1995; Baum & Shipilov, 2006). Recent refinements of inertia theory propose that the deleterious process effects depend on the centrality and connectedness of the organizational units subjected to transformation attempts within the overall organizational structure (for a review, see Po´los & van Witteloostuijn, 2006). By this imagery, the degree of internal and external misalignment that occurs during the process of change depends on the location of change within the organization. Because intended change in centrally located organizational units
16
STANISLAV D. DOBREV ET AL.
triggers unintended change in units to which they are connected, the duration of transformation also increases. So consequently do the expected negative effects of the change process. This refinement reconceptualizes core features and change based on the insight that the adverse impact of transformation arises from its unintended effects. Because the unanticipated consequences of organizational change are a direct function of the extensiveness of change, core transformation is defined in terms of the additional subsequent unplanned changes that need to be implemented as a result of the initial change attempt (Dobrev et al., 2003). It is such cascades of change throughout the organization that largely account for the indirect and opportunity costs associated with the transition between two states (Hannan, Po´los, & Carroll, 2003a, b, 2004). Ecology’s inertia theory offers a useful antidote against uncritical proflexibility arguments often informing strategy research (see Crossan & Hurst, this volume; Miller & Le Bretton Miller, this volume; Singh, this volume). Not only does it provide an account of why organizations tend to be inert in practice, but it also argues that inertia is a consequence rather than the antecedent of population-level selection processes (for provocative analyses of the role of inertia in entrepreneurship, see the pair of papers in this volume by Dew et al. and Ruef). Building on these insights, ecology moved on to study the likely implications of inertia and change for organizational performance.
Organizational Size and the Complexity of Core Changes The extensiveness of change generally depends on the size of the organization, which is an important source of structural complexity and variability. Ecological research often links structural complexity with inertia, suggesting that complex organizations are inherently less capable of initiating and surviving fundamental change. Hannan and Freeman argue ‘‘the level of structural inertia increases with size for each form of organization’’ (1989, p. 82). The relationship between scale, complexity, and inertia implies that large organizations with dense and saturated structures are hard and slow to change. In short, the larger an organization, the more extensive the changes in structures and processes required throughout the rest of the production process to complete a given change. The inertia of large organizations runs counter to predictions that they have a greater margin of error that allows them to buffer themselves from negative repercussions of core transformation (Levinthal, 1991). This
17
Introduction
advantage attributable to large size emanates from the superior resources that large organizations command. Indeed, critics of structural inertia theory are quick to point out that powerful organizations in established, economically central industries possess the capacity to not only withstand internal change, but also to impact the course of industry development (e.g., Holcombe et al., this volume). In this way, exogenous changes in the environment that often instigate internal transformation attempts for most organizations effectively become endogenous to very large organizations. Recent attempts to reconcile these arguments suggest that the latter logic applies only to the very largest organizations, for which the effect of superior access to resources overrides the deleterious effect of process change in the core on organizational performance. Inertia considerations, however, prevail in predictions of performance implications for the remaining subset of scale competitors subjected to core transformation. Empirical analyses have shown that the detrimental effect of change processes on organizational performance initially increases and then decreases with size (Carroll & Teo, 1996; Dobrev et al., 2003). In so doing, ecology contributes to a better understanding of how organizational size interacts with organizational change, and of the impact of change on performance, clearly key issues in strategy.
Organizational Niche Width and Core Change Are all inert or all non-inert organizations equally likely to fail? A general answer is that some organizational characteristics might serve as buffers against inertial forces – but which characteristics? As organizations age and grow, they become senile, rigidly bureaucratized, and burdened by obsolete blueprints. Yet at the same time, past experiences and exposure to different contexts also provide learning opportunities for cumulating and interpreting knowledge – the so-called tacit, organization-specific capital, cognitively stored in the collective memory of the organization. Organizations that actively promote learning and exploration might also have a greater chance of weathering selection pressures (see Crossan & Hurst, this volume; Fleming & Szigety, this volume; Greve & Rao, this volume). Theories of organizational learning posit that a principal mechanism by which organizations learn from their experiences involves continually investing resources to support organizational search for new and better routines and solutions (March, 1991; Levitt & March, 1988). The success of this strategy depends on the extent to which an organization can surmount
18
STANISLAV D. DOBREV ET AL.
pressures for efficiency and slack-cutting, and commit to building and sustaining variation of outcomes to be considered as viable alternatives to existing routines and practices (e.g., Barnett & Hansen, 1996; Barnett & Sorenson, 2002). In sorting out differences among firms in their ability to learn, ecological theory suggests the organizational niche may play a central role. Specifically, the variation necessary for learning is more likely to be produced and maintained by broad-niche organizations (generalists) that operate across multiple environmental states and encompass large operational domains. Additionally, broad-niche firms possess experience with transferring resources between operational units, experience that may be drawn upon in case the organization undergoes a market repositioning and needs to move resources from the origin to the destination state of the transition. Indeed, the purported advantage of generalists lies in their hedging strategy (Freeman & Hannan, 1983) – they spread their bets across several alternatives with uneven payoff opportunities. Specialists, by contrast, bet their success on exploiting a single narrow niche. This focused strategy inevitably results in a trade-off between static efficiency and dynamic adaptability because it deprives the organization of the opportunity to develop a broad set of competencies that can be applied to multiple market domains, and to gain experience in transferring capabilities and resources across these domains (Baum, Li, & Usher, 2000). Broad-niche firms also benefit from internal variation-selection-based learning (Weick, 1969; Burgelman, 1991; Audia, Sorenson, & Hage, 2001; but see Ingram & Baum, 1997a). So, the generalist advantage manifested in a greater flexibility and higher capacity for adaptation serves as a buffer against the negative impact of the process of change. It follows that the deleterious process effect of change ought to vary by niche width, and that performance decline due to change should be smaller for broad-scope competitors.
The Perils of Learning It is widely believed that experience impacts the propensity of an organization to implement change. As it gains experience in undertaking a particular type of change, an organization becomes more likely to attempt a similar transformation in the future (Amburgey & Miner, 1992; Amburgey, Kelly, & Barnett, 1993). Experience with a certain type of change also tends to constrain future transformations: the organization becomes less likely to experiment with different types of transformation. The extreme case of this
Introduction
19
constraint is what Levitt and March (1988) call a ‘‘competency trap,’’ whereby organizations proceed down known courses of action even when inappropriate. So not only can change be derailed by inertia, but inertia can emerge as an outcome of change. In other words, learning and inertia both emerge as a function of past experiences and experiential learning invariably entails both opportunities and constraints (Ingram & Baum, 1997a, b; Sørensen & Stuart, 2000). Of course, answers to questions of why organizations change also require consideration of both environmental and organizational characteristics, as well as the interplay between them and experience. The interaction is particularly important because it reveals the internal mechanisms through which organizations interpret and react to their external context. Recent integration between ecology and learning theory follows Levitt and March (1988) and attends to three aspects of experiential learning in organizations: (1) interpretation of experience, (2) complexity of experience, and (3) ambiguity of success. First, because an organization’s environment consists mostly of other organizations, its propensity to change is modified by its relationship with other organizations (Miner, Amburgey, & Stearns, 1990; Haveman, 1993a; Greve, 1995; Baum & Singh, 1996). Because firms constantly strive to outmaneuver or respond to the actions of their competitors, one can conceive of the organizational environment in terms of the structure of the market, defined by the market positions of incumbents. Changes in the market structure triggered by the collective action (e.g., position moves) of competitors are likely to trigger a response by other incumbents (e.g., Chen, 1996). A major issue of organizational learning concerns how to interpret experience (Levitt & March, 1988). In a market, the volatility of the setting matters both for understanding what happened and in applying this knowledge to possible future action. Organizational intuition to engage recurrently in a behavior that has produced positive results in the past may interfere with its capacity to interpret market signals objectively (e.g., Miller, 1990). If this is correct, then the acquired propensity to change because of prior success may override the visibly better alternative to not change when the market remains stable. When environments demand realignment and repeating a change is an adequate strategy, experienced organizations are in a superior position because they are more likely to initiate that same or similar change. But when environments are stable, these organizations are at a disadvantage. This is because relative to firms without prior change experience, they are more
20
STANISLAV D. DOBREV ET AL.
likely to follow the ‘‘wisdom’’ of their experience, are more prone to misinterpret or underestimate exogenous conditions, and are thus more likely to change. Because changes in market position are path-dependent over time, experienced organizations may initiate change even when this leads to a misalignment with stable environments. Second, features of an organization’s design also impact its propensity to change (Zajac & Shortell, 1989; Amburgey & Miner, 1992). Research indicates that the likelihood to initiate transformation generally declines with size (e.g., Haveman, 1993b). Substantively, the explanation for the negative relationship between size and transformation centers on arguments about the complex and bureaucratic nature of large organizations. One interpretation sees complex and bureaucratic organizations as slow and clumsy, facing greater demands for balancing incentives and coordination among a growing number of internal constituents. Large organizations also likely face difficulty in interpreting their experiences, however. As the number of subunits and persons increases within an organization, the number of possible relations among them increases geometrically. The sheer amount of information required to store all behavior of a large organization strain decision makers’ cognitive limits. And, both bureaucracy and institutionalization make it difficult to understand what happened and why because some (often not easily identifiable) actions are routine responses and others are not. Such complexity of experience makes it harder to act in the future, because it is unclear what outcome might be expected and there are ample rationales available to those who would resist the action. Contained in this interpretation is a paradox: organizations that can collect a lot of information often have organizational structures that are difficult to change, while those that can change the fastest often have trouble collecting relevant information to guide changes (Baum & Singh, 1996; see also Miller & Le Bretton Miller, this volume). Third, organizational growth also often breeds expansion in the organizational niche, thus constraining exploration to occur mostly within the scope of the firm. When the range of resources that constitute organizational inputs is high, an organization is less pressed to explore beyond its boundaries. The variation of outcomes needed to support experimentation and learning can be produced within its niche. So even though high variance in resource utilization facilitates exploration, organizational learning is likely to occur locally (in terms of market position), within the domain of the broad-niche firm. Organizations with broad niche width also face problems of interpreting their experiences that are likely to increase inertia. Specifically, a generalist
21
Introduction
organization participating in a broad array of market segments encounters a problem of ambiguity regarding success and failure (Ingram & Baum, 1997a; Levitt & March, 1988; Greve & Baum, 2001). That is, the fate of the combined entity can be plausibly attributed to a variety of activities or segments by internal actors, whether they are truly related or not. As with complexity, the resulting plethora of rationales likely inhibits future action. Thus, a broad niche width appears to be a double-edged sword when it comes to learning: on the one hand it enables internal variation-selectionbased learning; on the other it creates problems of greater variation in competitive environments, which creates a risk of transferring routines from segments in which they are beneficial to segments where they are harmful. Organizational ecology offers a rich theory of organizational change, and its impact on performance. By helping to specify how different types of changes, types of organizations, and environmental conditions are likely to influence the performance consequences of change, ecological theory has much to contribute to advancing strategic management theory and research on organizational change.
CONCLUSION We have barely scratched the tip of the iceberg here. But it is time to turn you over to the volume’s contributors. As you will see, and we have tried briefly to illustrate, while strategists and ecologists frequently address the same broad phenomena, different disciplinary customs and preferences distinguish their work. Despite the differences in their approaches, they have often arrived at complementary insights. Each perspective, in our view, offers a partial view of strategy and organizations. After reading the paired contributions that follow, we hope you will see that together the two perspectives provide a more comprehensive and encompassing view of the wide range of strategic and organizational phenomena addressed in the chapters. We also hope that by interpreting ecological arguments and findings with a strategy lens, the managerial relevance of ecological insights becomes clearer, for example, for diagnosing the likely impact of organizational change or designing entry strategies. We bring strategy and ecology together in the belief that they have much to learn from one another, and that exchange will lead each field to develop richer and more compelling understandings of organizational strategy. The two groups have traded punches at times, but we hope in presenting the paired articles published here to help to open a broader, ongoing dialogue.
22
STANISLAV D. DOBREV ET AL.
NOTES 1. Because subsequent chapters introduce a range of relevant ecological theory, we do not provide a general overview of organizational ecology. For recent reviews, interested readers are referred to Baum and Shipilov (2006) and Carroll and Hannan (2000).
REFERENCES Amburgey, T. L., Kelly, D., & Barnett, W. P. (1993). Resetting the clock: The dynamics of organizational change and failure. Administrative Science Quarterly, 38, 51–73. Amburgey, T. L., & Miner, A. S. (1992). Strategic momentum: The effects of repetitive, positional, and contextual momentum on merger activity. Strategic Management Journal, 13, 335–348. Audia, P. G., Sorenson, O., & Hage, J. (2001). Tradeoffs in the organization of production: Multiunit firms, geographic dispersion and organizational learning. In: J. A. C. Baum & H. R. Greve (Eds), Multiunit organization and multimarket strategy – advances in strategic management (Vol. 18, pp. 75–105), Oxford, UK: JAI/Elsevier. Barnett, W. P. (1997). The dynamics of competitive intensity. Administrative Science Quarterly, 42, 128–160. Barnett, W. P., & Carroll, G. R. (1995). Modeling internal organizational change. Annual Review of Sociology, 21, 217–236. Barnett, W. P., & Freeman, J. (2001). Too much of a good thing? Product proliferation and organizational failure. Organization Science, 12, 539–558. Barnett, W. P., & Hansen, M. T. (1996). The Red Queen in organizational evolution. Strategic Management Journal, 17, 139–157. Barnett, W. P., & Sorenson, O. (2002). The Red Queen in organizational creation and development. Industrial and Corporate Change, 11, 289–325. Barron, D. N. (1999). The structuring of organizational populations. American Sociological Review, 64, 421–445. Baum, J. A. C. (1989). Liabilities of newness, adolescence, and obsolescence: Exploring age dependence in the dissolution of organizational relationships and organizations. Proceedings of the Administrative Sciences Association of Canada, 10(5), 1–10. Baum, J. A. C. (1995). The changing basis of competition in organizational populations: The Manhattan hotel industry, 1898–1990. Social Forces, 74, 177–205. Baum, J. A. C. (1996). Organizational ecology. In: S. R. Clegg, C. Hardy & W. R. Nord (Eds), Handbook of organization studies (pp. 77–114). London: Sage. Baum, J. A. C., & Haveman, H. A. (1997). Love thy neighbor? Differentiation and agglomeration in the Manhattan hotel industry, 1898–1990. Administrative Science Quarterly, 42, 304–338. Baum, J. A. C., & Korn, H. J. (1996). Competitive dynamics of interfirm rivalry. Academy of Management Journal, 39, 255–291. Baum, J. A. C., & Korn, H. (1999). Dynamics of dyadic competition. Strategic Management Journal, 20, 251–278. Baum, J. A. C., Li, S. X., & Usher, J. (2000). Making the next move: How experiential and vicarious learning shape the locations of chains’ acquisition. Administrative Science Quarterly, 45(4), 766–801.
Introduction
23
Baum, J. A. C., & Mezias, S. J. (1992). Localized competition and organizational failure in the Manhattan hotel industry, 1898–1990. Administrative Science Quarterly, 37, 580–604. Baum, J. A. C., & Shipilov, A. V. (2006). Ecological approaches to organizations. In: S. R. Clegg, C. Hardy, W. R. Nord & T. Lawrence (Eds), Handbook of organization studies (2nd ed., pp. 55–110). London: Sage. Baum, J. A. C., & Singh, J. V. (1994a). Organizational niche overlap and the dynamics of organizational founding. Organization Science, 5, 483–501. Baum, J. A. C., & Singh, J. V. (1994b). Organizational niche overlap and the dynamics of organizational mortality. American Journal of Sociology, 100, 346–380. Baum, J. A. C., & Singh, J. V. (1996). Dynamics of organizational responses to competition. Social Forces, 74, 1261–1297. Boone, C., Broecheler, V., & Carroll, G. R. (2000). Custom service: Application and tests of resource partitioning among Dutch auditing firms from 1880 to 1982. Organization Studies, 21, 355–381. Boone, C., Carroll, G. R., & van Witteloostuijn, A. (2004). Size, differentiation and the performance of Dutch daily newspapers. Industrial and Corporate Change, 13, 117–148. Boone, C., & van Witteloostuijn, A. (2004). A unified theory of market partitioning: An integration of resource-partitioning and sunk cost theories. Industrial and Corporate Change, 13, 701–725. Bruderl, J., & Schussler, R. (1990). Organizational mortality: The liabilities of newness and adolescence. Administrative Science Quarterly, 39, 530–547. Burgelman, R. A. (1991). Intraorganizational ecology of strategy-making and organizational adaptation: Theory and field research. Organization Science, 2, 239–262. Carroll, G. R. (1983). A stochastic model of organizational mortality: Review and reanalysis. Social Science Research, 12, 303–329. Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in populations of organizations. American Journal of Sociology, 90, 1262–1283. Carroll, G. R., Bigelow, L., Seidel, M.-D., & Tsai, L. (1996). The fates of de novo and de alio producers in the American automobile industry, 1885–1982. Strategic Management Journal, 17, 117–137. Carroll, G. R., Dobrev, S. D., & Swaminathan, A. (2002). Organizational processes of resource partitioning. Research in Organizational Behavior, 24, 1–40. Carroll, G. R., & Hannan, M. T. (1989). Density delay in the evolution of populations of newspaper organizations: A model and five empirical tests. Administrative Science Quarterly, 34, 411–430. Carroll, G. R., & Hannan, M. T. (2000). The demography of corporations and industries. Princeton, NJ: Princeton University Press. Carroll, G. R., & Swaminathan, A. (2000). Why the microbrewery movement? Organizational dynamics of resource partitioning in the US brewing industry. American Journal of Sociology, 106, 715–762. Carroll, G. R., & Teo, A. C. (1996). Creative self-destruction among organizations: An empirical study of technical innovation and organizational failure in the American automobile industry, 1885–1981. Industrial and Corporate Change, 5, 619–644. Chandler, A. D. (1962). Strategy and structure: Chapters in the history of the industrial enterprise. Cambridge, MA: MIT Press. Chen, M.-J. (1996). Competitor analysis and interfirm rivalry: Toward a theoretical integration. Academy of Management Review, 21, 100–134.
24
STANISLAV D. DOBREV ET AL.
Dobrev, S. D. (1999). The dynamics of the Bulgarian newspaper industry in a period of transition: Organizational adaptation, structural inertia, and political change. Industrial and Corporate Change, 8, 573–605. Dobrev, S. D. (2000). Decreasing concentration and reversibility of the resource partitioning process: Supply shortages and deregulation in the Bulgarian newspaper industry, 1987– 1992. Organization Studies, 21, 383–404. Dobrev, S. D. (2001). Revisiting organizational legitimation: Cognitive diffusion and sociopolitical factors in the evolution of Bulgarian newspaper enterprises, 1846–1992. Organization Studies, 22, 419–444. Dobrev, S. D. (2005). Niches. In: J. Beckert & M. Zafirovski (Eds), International encyclopedia of economic sociology. UK: Routledge. Dobrev, S. D., & Carroll, G. R. (2003). Size (and competition) among organizations: Modeling scale-based selection among automobile producers in four major countries. Strategic Management Journal, 24, 541–558. Dobrev, S. D., & Kim, T.-Y. (2006). Positioning among organizations in a population: Moves between market segments and the evolution of industry structure. Administrative Science Quarterly, 51(2). Dobrev, S. D., Kim, T.-Y., & Carroll, G. R. (2002). The evolution of organizational niches. Administrative Science Quarterly, 47, 233–264. Dobrev, S. D., Kim, T.-Y., & Carroll, G. R. (2003). Shifting gears, shifting niches: Organizational inertia and change in the evolution of the U.S. automobile industry, 1885–1981. Organization Science, 14, 264–282. Dobrev, S. D., Kim, T.-Y., & Hannan, M. T. (2001). Dynamics of niche width and resource partitioning. American Journal of Sociology, 106, 1299–1337. Dobrev, S. D., Kim, T.-Y., & Solari, L. (2004). The two sides of the coin: Core competence as capability and obsolescence. In: J. A. C. Baum A. M. McGahan (Eds), Business strategy over the industry life cycle – advances in strategic management (Vol. 21, pp. 255–285). Oxford, UK: JAI/Elsevier. Dosi, G., Teece, D. J., & Winter, S. G. (1992). Toward a theory of corporate coherence: Preliminary remarks. In: G. Dosi, R. Giannetti & P. M. Toninelli (Eds), Technology and enterprise in a historical perspective (pp. 184–211). Oxford: Clareden Press. Finkelstein, S., & Hambrick, D. C. (1990). Top-management-team tenure and organizational outcomes: The moderating role of managerial discretion. Administrative Science Quarterly, 35, 484–503. Freeman, J. H., Carroll, G. R., & Hannan, M. T. (1983). The liability of newness: Age dependence in organizational death rates. American Sociological Review, 48, 692–710. Freeman, J. H., & Hannan, M. T. (1983). Niche width and the dynamics of organizational populations. American Journal of Sociology, 88, 1116–1145. Greve, H. R. (1995). Jumping ship: The diffusion of strategy abandonment. Administrative Science Quarterly, 40(September), 444–473. Greve, H. R., & Baum, J. A. C. (2001). Introduction: A multiunit, multimarket world. In: J. A. C. Baum H. R. Greve (Eds), Multiunit organization and multimarket strategy: Advances in strategic management (Vol. 18, pp. 1–28). Oxford: Elsevier. Hannan, M. T., & Carroll, G. R. (1992). Dynamics of organizational populations: Density, competition, and legitimation. New York: Oxford University Press.
Introduction
25
Hannan, M. T., Carroll, G. R., Dobrev, S. D., Han, J., & Torres, J. C. (1998). Organizational mortality in European and American automobile industries, Part I: Revising age dependence. European Sociological Review, 14, 279–302. Hannan, M. T., & Freeman, J. H. (1977). The population ecology of organizations. American Journal of Sociology, 83, 929–984. Hannan, M. T., & Freeman, J. H. (1984). Structural inertia and organizational change. American Sociological Review, 49, 149–164. Hannan, M. T., & Freeman, J. H. (1989). Organizational ecology. Cambridge, MA: Harvard University Press. Hannan, M. T., Po´los, L., & Carroll, G. R. (2003a). The fog of change: Opacity and asperity in organizations. Administrative Science Quarterly, 48, 399–432. Hannan, M. T., Po´los, L., & Carroll, G. R. (2003b). Cascading organizational change. Organization Science, 14, 463–482. Hannan, M. T., Po´los, L., & Carroll, G. R. (2004). The evolution of inertia. Industial and Corporate Change, 13, 213–214. Hannan, M. T., Po´los, L., & Carroll, G. R. (2006). Social codes and ecologies: Logics of organization theory. Princeton, NJ: Princeton University Press. Haveman, H. A. (1993a). Follow the leader: Mimetic isomorphism and entry into new markets. Administrative Science Quarterly, 38, 593–627. Haveman, H. A. (1993b). Organizational size and change: Diversification in the savings and loan industry after deregulation. Administrative Science Quarterly, 38, 20–50. Henderson, A. (1999). Firm strategy and age dependence: A contingent view of the liabilities of newness, adolescence, and obsolescence. Administrative Science Quarterly, 44, 281–324. Ingram, P., & Baum, J. A. C. (1997a). Opportunity and constraint: Organizations’ learning from the operating and competitive experience of industries. Strategic Management Journal, 18(Summer), 75–98. Ingram, P., & Baum, J. A. C. (1997b). Chain affiliation and the failure of Manhattan hotels, 1898–1980. Administrative Science Quarterly, 42, 68–102. Levinthal, D. A. (1991). Random walks and organizational mortality. Administrative Science Quarterly, 36, 397–420. Levitt, B., & March, J. G. (1988). Organizational learning. Annual Review of Sociology, 14, 319– 340. Main, B. G. M., O’Reilly, C. A., III., & Wade, J. B. (1995). The CEO, the board of directors and executive compensation: Economic and psychological perspectives. Industrial and Corporate Change, 4, 293–332. March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2, 71–87. McGahan, A. M., & Porter, M. E. (1997). How much does industry matter, really. Strategic Management Journal, 18, 15–30. Miller, D. (1990). The Icarus Paradox: How exceptional companies bring about their own downfall. New York: HarperCollins. Miner, A. S., Amburgey, T. L., & Stearns, T. (1990). Interorganizational linkages and population dynamics: Buffering and transformational shields. Administrative Science Quarterly, 35, 689–713. Pennings, J., Lee, K., & van Witteloostuijn, A. (1998). Human capital, social capital, and firm dissolution. Academy of Management Journal, 41, 425–440.
26
STANISLAV D. DOBREV ET AL.
Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations. New York: Harper & Row. Podolny, J. M., Stuart, T. E., & Hannan, M. T. (1996). Networks, knowledge, and niches: Competition in the worldwide semiconductor industry, 1984–1991. American Journal of Sociology, 102, 659–689. Po´los, L., & van Witteloostuijn, A. (2006). Who should dare to change?: The theory of organizational inertia. Netherlands Journal of Social Sciences, 40, 247–269. Porter, M. E. (1980). Competitive strategy. New York: The Free Press. Prahalad, C. K., & Hamel, G. (1990). The core competence of the corporation. Harvard Business Review, 68(3), 79–91. Ranger-Moore, J. (1997). Bigger may be better, but is older wiser? Organizational age and size in the New York life insurance industry. American Sociological Review, 62, 903–920. Ruef, M. (1997). Assessing organizational fitness on a dynamic landscape: An empirical test of the relative inertia thesis. Strategic Management Journal, 18, 837–853. Rumelt, R. (1984). Towards a strategic theory of the firm. Englewood Cliffs, NJ: Prentice-Hall. Schumpeter, J. A. (1950). Capitalism, socialism and democracy (3rd ed.). New York: HarperCollins. Shapiro, C. (1989). Theories of oligopoly behavior. In: R. Schmalensee & R. D. Willig (Eds), Handbook of industrial organization (pp. 329–414). Amsterdam: North-Holland. Singh, J. V., House, R. J., & Tucker, D. J. (1986). Organizational change and organizational mortality. Administrative Science Quarterly, 31, 587–611. Sørensen, J. B., & Stuart, T. E. (2000). Aging, obsolescence, and organizational innovation. Administrative Science Quarterly, 45, 81–112. Sorenson, O. (2003). Interdependence and adaptability: Organizational learning and the longterm effect of integration. Management Science, 49, 446–463. Steiner, G. A. (1964). Why and how to diversify. California Management Review, 4, 11–17. Stinchcombe, A. L. (1965). Social structure and organizations. In: J. G. March (Ed.), Handbook of organizations (pp. 142–193). Chicago: Rand McNally. Stuart, T. E. (2000). Interorganizational alliances and the performance of firms: A study of growth and innovation rates in a high-technology industry. Strategic Management Journal, 21, 791–811. Swaminathan, A., & Dowell, G. (2000). Racing and back-pedaling into the future: New product introduction and organizational mortality in the U. S. Bicycle Industry, 1890–1918. Organization Studies, 21, 405–431. van Witteloostuijn, A. (2002). Interorganizational Economics. In: J. A. C. Baum (Ed.), Companion to organizations (pp. 686–712). Oxford: Blackwell Publishers. Weick, K. E. (1969). The social psychology of organizing. Reading, MA: Addison-Wesley. Wernerfelt, B. (1984). A resource-based view of the firm. Strategic Management Journal, 5, 171– 181. Zajac, E. J., & Shortell, S. M. (1989). Changing generic strategies: Likelihood, direction, and performance implications. Strategic Management Journal, 10, 413–430.
PART I: ENTERPRENEURSHIP
27
28
BOOM AND BUST: THE EFFECT OF ENTREPRENEURIAL INERTIA ON ORGANIZATIONAL POPULATIONS$ Martin Ruef ABSTRACT Although recent public attention has focused on boom-and-bust cycles in industries and financial markets, organizational theorists have made only limited contributions to our understanding of this issue. In this chapter, I argue that a distinctive strategic insight into the mechanisms generating boom-and-bust cycles arises from a focus on entrepreneurial inertia – the lag time exhibited by organizational founders or investors entering a market niche. While popular perceptions of boom-and-bust cycles emphasize the deleterious effect of hasty entrants or overvaluation, I suggest $
This chapter has benefited from feedback in presentations at Cornell University, the University of Arizona, the Stanford Graduate School of Business, and the 2004 American Sociological Association Meetings in San Francisco. In particular, I would like to thank Bill Barnett, Jim Baron, Ron Breiger, Joel Baum, Glenn Carroll, Stanislav Dobrev, Joe Galaskiewicz, John Hall, Michael Hannan, Omar Lizardo, Misiek Piskorski, Charles Ragin, Wesley Sine, Pam Tolbert, and Arjen van Witteloostuijn for their suggestions. Naturally, any opinions and findings expressed are those of the author and do not necessarily reflect the views of other commentators.
Ecology and Strategy Advances in Strategic Management, Volume 23, 29–72 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23002-X
29
30
MARTIN RUEF
instead that slow, methodical entries into an organizational population or market may pose far greater threats to niche stability. This proposition is explored analytically, considering the development of U.S. medical schools since the mid-18th century.
INTRODUCTION What mechanisms explain the occurrence of boom-and-bust cycles in the development of markets and industries? Economic historians have argued that speculative bubbles will arise periodically in financial markets due to overconfidence in profits and subsequent bandwagon effects (Kindleberger, 1996; Galbraith, 1997). Organizational ecologists have found that many industries – such as automobile manufacturing (Hannan, 1997), brewing (Carroll & Wade, 1991), telephony (Barnett, 1997), and credit unions (Barron 1999) – have been subject to similar boom-and-bust cycles during their histories, characterized by rapid increases in the number of entrants followed in short order by abrupt consolidation. Their explanations for this phenomenon, however, emphasize structural constraints rather than irrationality. For instance, Carroll and Hannan’s (1989) density delay argument suggests that organizations founded under conditions of intense competition are likely to suffer a subsequent liability of scarcity. This implies that boom periods in the development of organizational populations, featuring large numbers of entrants scrambling for scarce resources, often lay the foundation for subsequent shake-outs of frail organizations (see Ruef, 2004, for a review of other ecological mechanisms). More recently, management scholars have noted that the introduction of new types of services and products – e-commerce, video games, or integrated circuits, to name a few – are especially susceptible to boom-and-bust cycles (Paich & Sterman, 1993). They attribute these cycles to a combination of structural constraint and bounded rationality, examining the poor strategic decision-making of entrepreneurs who enter complex and uncertain niches.1 Naturally, misperceptions of feedback can also generate market instability for products that are well understood, as Sterman’s (1989) experiments in supply chain management demonstrate. A common thread across this diverse research in sociology, economics, and strategic management is that boom-and-bust cycles arise when investors, managers, or organizations engage in behavior that is either overly confident about future outcomes (as in managerial accounts of faulty decision-making
Boom and Bust
31
on the part of entrepreneurs) or strongly imprinted by previous experience (as in ecological arguments about density delay). Both conditions imply a strategic failure on the part of the economic actors involved, who are either undersocialized forecasters or oversocialized victims of social context (Granovetter, 1985). In this chapter, I argue to the contrary that neither assumption is required to generate a boom-and-bust cycle. In fact, I will suggest that it is the most methodical – and in certain respects ‘‘rational’’ – market entrants and entrepreneurs who may pose the greatest threat to the stability of markets and organizational populations. By slowly assembling resources and planning market entry, these actors help to ensure their own success (cf. Dixit, 1992), while contributing to collectively irrational outcomes. This chapter explores the effects of this micro–macro disconnect on boom-and-bust cycles in organizational populations. Boom-and-bust cycles are conceptualized in terms of the changing prevalence – or density – of organizations serving a specific market niche. I begin with a theoretical framework that underscores the importance of lags in the initial stage of market entry, a mechanism that I refer to as entrepreneurial inertia (see Lomi, Larsen, & Freeman (2005), for a parallel discussion of delayed adjustment within a simulation framework). The effects of this mechanism are then assessed empirically for the population of U.S. medical schools, observed from their origins in 1765. By the late 19th century, medical schools were being founded at an unprecedented rate, spurred by the legitimacy offered through licensing laws and the quest for profits among entrepreneurs creating proprietary institutions (Starr, 1982, p. 112). Between 1870 and 1900, the number of American medical schools nearly doubled (peaking at 174 in 1898), before going into an equally dramatic tail-spin over the following 30 years. By the beginning of the Great Depression, the schools again numbered fewer than they had during the Civil War. To account for this boom-and-bust cycle, I argue that existing explanations, emphasizing either exogenous influences (e.g. the impact of the famous Flexner Report), ecological influences (intense intra-industry competition at the end of the 19th century), or irrational contagion, must be qualified by attention to the founding process of individual organizations. I show that the mechanism of entrepreneurial inertia improves upon existing explanations at an aggregate level and provides further insights when the process of market entry is studied for individual medical schools. The chapter concludes by considering how micro-level studies of entrepreneurial strategy can be integrated with macro-level studies of dynamics in organizational populations and markets.
32
MARTIN RUEF
THEORETICAL AND SUBSTANTIVE BACKGROUND Since the inception of ecological analyses of organizational populations, two ideas have played a central role in explaining trajectories of population development. One is that formal organizations tend to be characterized by structural inertia (Hannan & Freeman, 1977, 1984). Owing to internal rigidity – resulting from sunk costs in specific resources, bounded rationality, political resistance to change, and habituation – most organizations are unlikely to engage in major transformation. Moreover, those organizations that do change core features often face selection pressures, at least in the short term, as stakeholders question their reliability or accountability (Barnett & Carroll, 1995; Pe´li, Bruggeman, Masuch, & Nualla´in, 1994). Resistance to change is exacerbated by external conditions, such as market uncertainty and divergent metrics applied by different stakeholders. For organizational theorists, the analytical consequence is that much is to be gained from focusing on processes of entry into (and exit from) organizational populations, as opposed to adaptation on the part of incumbent participants. A second idea follows from this insight. If most of the meaningful variance in population development can be accounted for in terms of entry and exit processes, then organizational scholars ought to study how the intrinsic rates of these processes vary by the demographic composition – size and distribution – of organizational populations (Carroll & Hannan, 2000). In principle, demographic composition should explain both secular trends in the growth and decline of organizational populations, as well as short-term fluctuations, such as boom-and-bust cycles. A number of extensions to organizational ecology have appeared that may accommodate such fluctuations, of which density delay (Carroll & Hannan, 1989) is one of the most prominent. One irony in research on the development of organizational populations is that the idea of structural inertia, which largely gave rise to the demographic perspective on population trajectories, is analytically absent from population-level modeling. A number of studies have examined the emergence of inertia in young ventures (e.g. Baron, Hannan, & Burton, 1999) and the dangers that radical or frequent change holds for established organizations (Ruef, 1997; Amburgey, Kelly, & Barnett, 1993). But these empirical variations at the level of individual organizations have not been traced back to the population level. To some extent this reflects a more general theoretical ambiguity as to whether structural inertia is a feature of specific organizations or a property that accumulates in organizational
Boom and Bust
33
populations (Pe´li, Po´los & Hannan, 2000). Moreover, structural inertia seems a convenient background assumption in explaining away microlevel mechanisms leading to boom-and-bust cycles. If speculative bubbles are created by entrepreneurs who rush foolishly to enter new market niches or to adopt novel innovations (Aldrich & Fiol, 1994), then selection pressures will quickly eliminate these actors for their lack of reliability and accountability, at least in certain logical formulations of structural inertia theory (e.g. Pe´li et al., 1994). As a result, organizational ecology suggests that sustained boom-and-bust cycles cannot arise from the actions of individual entrepreneurs or firms, but must arise due to properties of industries. In contrast to the prevailing interpretation, this chapter proposes that inertia on the part of individual entrepreneurs or incumbent organizations can be a principal cause of boom-and-bust cycles. This is especially true when new ventures evidence ‘‘entrepreneurial inertia’’ – long lags between the time when entrepreneurs become committed to the idea of founding an organization and its operational startup. Entrepreneurial inertia has opposing implications for the focal venture being founded and the population whose niche it seeks to occupy. For the focal venture, the delay between preoperational and operational startup allows founders to carefully assemble resources, study competitors, and plan their entry into a market. Provided that the nascent venture survives this preoperational phase, it will emerge as a stronger competitor than organizations that rush into an emerging niche. At the population level, on the other hand, the prevalence of entrepreneurial inertia can destabilize an industry, for two reasons. First, entrepreneurs and incumbents tend to form expectations about the competitive pressures they see arising from other operational entities. When a large number of preoperational ventures are unobserved, this translates into poor forecasts regarding the evolution of an industry over time (Lomi et al., 2005). Second, when ventures with long periods of gestation do become operational, they reflect an unusual competitive intensity, as noted above. Again, entrepreneurs and incumbents fail to accurately forecast the threat posed by these (ostensibly) fledgling organizations. I now proceed to ground these intuitions concerning entrepreneurial strategy in an analysis of U.S. medical schools. A short case study of the founding process of the Johns Hopkins Medical School proves instructive in understanding the implications of entrepreneurial lags on population dynamics. This case study is followed by quantitative examinations of the development of the medical school population.
34
MARTIN RUEF
Case Study: The Johns Hopkins Medical School The topic of a medical school at Johns Hopkins was already raised in June of 1874, when Charles Eliot, president of Harvard, came to Baltimore to confer with the trustees of the newly chartered University. Eliot emphasized that: y your opportunity y for the founding of a proper medical school in this community is a very precious one. I hope it will be one of the earliest departments of the University organized; and that it will be founded in such a way as to greatly influence for good, medical education in this country. No medical school already organized has anything like your opportunity; for medical schools everywhere are exceedingly poor. (French, 1946, p. 107)
The ‘‘opportunity’’ that Eliot referred to involved the sizable endowment that the institution’s namesake, Johns Hopkins, had bequeathed in his will prior to his death. Pursuing this idea, the University’s first dean, Daniel Gilman, studied European medical schools in a trip abroad the following year, where he sought to recruit faculty and analyzed the organizational forms of foreign universities (Hawkins, 1960). When Gilman came back to Baltimore, he echoed Eliot’s sentiments and, during his inaugural address, declared plans to found a department of medicine. Gilman’s announcement, in February of 1876, was followed in short order by several efforts at resource mobilization. In March, the University’s board of trustees accepted an endowment for a medical professorship and, in May, it purchased land for the school. These activities proved premature. The income generated by Hopkins endowment could not yet satisfy Gilman’s ambitions and the first medical professor (William Welch) was not recruited until 1885 (Hawkins, 1960, pp. 146–147). When Welch began instruction in the 1886–1887 term, his course was limited to existing practitioners, rather than medical students. In a subsequent 1889 report, Gilman was forced to acknowledge that, although the University and its newly created hospital could offer some courses to professionals who had already received their medical degrees, a major financial endowment was required to create a proper medical school (French, 1946, p. 111). It was not until December of 1892 that the requisite endowment of half-amillion dollars had been secured from philanthropists. In October of the following year, the school opened with 18 first-year students, four faculty, and William Welch as dean. Announcing the opening of the school, Welch stressed the small number and high quality of graduates it would turn out. His prediction could not be evaluated until 1897, when the school finally had its first ‘‘product,’’ 15 graduates with doctoral medical degrees.
Boom and Bust
35
The graduates were indeed imposing – four went on to become medical faculty at Johns Hopkins and one was a pioneering female physician (French, 1946, pp. 425–426); the medical school’s alacrity, on the other hand, seemed less impressive – it had taken over 20 years to get from Gilman’s initial founding announcement to the first graduating class. Two points about the founding of the Johns Hopkins Medical School are worth emphasizing. First, entrepreneurs often make strong commitments to a nascent organization long before it is able to function as an operational venture. Some of these commitments involve material investments, such as the land that the Johns Hopkins trustees procured after Dean Gilman publicly announced his plans to create a medical school. Other commitments are cognitive in character, such as the European models that Gilman studied in 1875 and adopted as the organizational form of the new university in Baltimore. When the lag between the initiation of an organization and operational startup is substantial, there is no guarantee that the environmental factors driving those commitments will be consistent with the environment that the organization ultimately competes in. This does not imply, however, that entrepreneurial inertia necessarily handicaps a nascent organization. Admittedly, the long incubation period of Johns Hopkins was driven partially by mishaps in fundraising and strategic miscalculation on the part of Gilman and the trustees. But much of the delay was also due to the founders’ ambition in institution-building, as they sought to conform to the standards of an emergent ‘‘dominant design’’ (Anderson & Tushman, 1990) for medical education in the United States. The result seemed to justify the delay – Abraham Flexner’s (1910) influential report on medical education identified Hopkins in the top echelon of medical schools. As Starr (1982) noted, ‘‘here were the glimmerings of the great university-dominated medical centers of the next century’’ (p. 116). In the two decades following the report, the population of U.S. medical schools exhibited severe decline, as other schools failed to meet up to the dominant design. This suggests a paradox in entrepreneurial inertia. For a startup, a prolonged founding process can allow for more extensive resource mobilization, organizational design, and role socialization among members – factors that are likely to improve the survival chances of the venture once (and if) it becomes operational. At the same time, the delay tends to deceive other entrepreneurs as to the true nature of competition in an ecological niche. Indeed, American medical schools were being founded at an unprecedented rate in the 1880s, based on a competitive landscape that did not yet include Johns Hopkins.
36
MARTIN RUEF
Theoretical Formalization: Population-Level To further understand the implications of structural inertia for boom-andbust cycles, we can consider a set of highly simplified models of development in organizational populations. These models are not intended to be realistic, but to allow the impact of inertia on population stability to be diagnosed in a fairly transparent fashion. In particular, these models rely on the assumptions that: (a) all organizations in a population exhibit an identical level of structural inertia; (b) the organizations are identical in all other salient aspects (i.e. there is no unobserved heterogeneity); and (c) there is a fixed carrying capacity for the population (i.e. the number of organizations that can be supported by the environment is constant over time). The latter assumption is particularly helpful in eliminating exogenous explanations for boom-and-bust cycles, as well as those linked to the psychological valuations imposed by entrepreneurs, managers, or investors. Employing these assumptions, the classic model of growth in organizational populations is Hannan and Freeman’s (1977, p. 941) logistic model: dNðtÞ NðtÞ ¼ rNðtÞ 1 (1) dt K where N(t) is the number of organizations in a population at time t, K the (fixed) capacity of the environment to support the population, and r the intrinsic growth rate of the population in the absence of resource or institutional constraints. The functional form of the model is motivated substantively by the insight that the cognitive legitimacy (and thus growth) of an organizational population tends to increase linearly with existing population density, but that competition attenuates this growth as the population approaches carrying capacity. Notably, the model does not take structural inertia into account – organizational startups and exits are, more or less, instantaneous reactions to the level of legitimation and competition in a population. Abundant evidence indicates support for this model, both in its original population-level form (e.g. Nielsen & Hannan, 1977) and in subsequent specifications that disaggregate the startup and exit processes contributing to population growth (Carroll & Hannan, 2000, Chapter 10). However, the variance explained can be quite limited when a population exhibits a boomand-bust-cycle (Ruef, 2004). Consider the density of U.S. medical schools, as shown in Fig. 1. Fitting the logistic model to the population trajectory of all schools between 1766 and 1999, we find that it explains only 1.8% the variation in annual rates of net growth (see Table 1).2 This suggests that at
37
Boom and Bust
All Schools
Sectarian Schools
Year
Fig. 1.
Number of Medical Schools in the United States (1765–1999).
least some of the simplifying assumptions made in the model may be untenable. One simple extension to the logistic model incorporates structural inertia for entrepreneurs wishing to enter and incumbents seeking to abandon an organizational population. In both cases, internal and external constraints contribute to delays in entry and exit processes. For instance, organizational founding often proceeds through a series of stages, including resource mobilization, legal establishment, social organization, and operational startup, that can take years to complete (Hannan & Freeman, 1989; Ruef, 2005). Even in high-tech sectors, such as semi-conductor production, where time-to-market speed is seen as critical, the waiting time until first product
38
MARTIN RUEF
Table 1. Estimates of Growth in the Population of U.S. Medical Schools, without Organizational Heterogeneity (1766–1999). Parameter
Classic Logistic Model
Delayed Logistic Model
Hybrid Delayed Model
Carrying capacity (K) Intrinsic growth rate (r) Startup rate (r1) Exit rate (r2)
150.4 0.030 – –
137.5 0.043 – –
129.1 – 0.215 0.180
Structural inertia (t)* R2 Number of observations
n/a 0.018 234
3.0 0.064 234
3.0 0.193 234
Note: Asterices (*) denote fixed parameters.
delivery can approach 2 years for a startup (Schoonhoven, Eisenhardt, & Lyman, 1990). The entrepreneurial lag can be even more pronounced for organizational forms such as medical schools (Kimberley, 1979), as the case of Johns Hopkins illustrates. For this population, the delay may also be exacerbated by institutional barriers to entry, such as medical boards and societies seeking to license the activities of schools (Starr, 1982). Though less obvious, it is also possible that exit processes may be subject to similar delays. Students of strategic bankruptcy (Delaney, 1992) and failing organizations (Meyer & Zucker, 1989) find that the organizational dissolution is often not a rapid reaction to adverse market conditions, but a carefully planned process in which managers seek to appease external stakeholders, minimize personal losses, and leverage legal protections. Similarly, economists describe a process of ‘‘optimal’’ disbanding, in which the exit of an incumbent firm is not viewed as a sudden failure but, rather, as a strategic, profit-maximizing response in the face of new challengers or technologies (see Dew, Goldfarb, & Sarasvathy (2006), in this volume). As with startup events, delays may also increase when outside authorities impose barriers to exit (as in the case of state legislators who believe that regional medical schools are desirable, despite an absence of local market demand). When processes contributing to growth or decline in an organizational population exhibit temporal lags, a modification of conventional ecological models is required. The simplest specification advanced for this purpose is the delayed logistic model (Turchin, 2003, pp. 56–57). It assumes that the effects of population density are subject to a constant fixed delay t. When this delay affects both legitimation and competition, the growth of an
39
Boom and Bust
organizational population can be determined as dNðtÞ Nðt tÞ ¼ rNðt tÞ 1 dt K
(2)
Aside from incorporating a term to represent the effect of structural inertia, the delayed logistic model also has the analytical advantage that it permits qualitative differences in trajectories as an organizational population approaches its carrying capacity. These qualitative differences are illustrated in Fig. 2, as: (a) a trajectory in which a population approaches its carrying capacity without significant fluctuations (stable steady state); (b) a trajectory in which a population approaches its carrying capacity with boom-and-bust oscillations (stable steady state with cycles); and (c) a trajectory in which a population never converges to its carrying capacity (unstable steady state, exhibiting ongoing boom-and-bust cycles).3 It is worth emphasizing that, in the absence of inertia affecting startup or exit processes, the trajectory for the model necessarily falls in the first category – i.e. stable steady state without oscillations. 200
Carrying Capacity (K = 150)
Density
150
100
Stable Steady State (Tau = 1)
50
Stable with Cycles (Tau = 5)
0
Fig. 2.
Unstable (Tau = 8) 0
20
40
60
80 100 Year
120
140
160
Trajectories of Organizational Populations under Different Parameterizations of Delayed Logistic Model.
40
MARTIN RUEF
How well does the delayed logistic model perform when applied to the trajectory of the medical school population? Before any given medical school graduates a cohort of students, outsiders have limited information as to the quality of the school or its competitive intensity. To estimate the model, I therefore calculated the average lag time between the initial organization of U.S. medical schools and the dates when they had their first graduating classes. The delay between preoperational startup and operational startup, averaging almost 2 years, serves as a basic indicator of entrepreneurial inertia. Since entrepreneurs also typically have imperfect information on other organizations being initiated at the same time as their venture (and which organizations are simultaneously failing), the indicator was lagged by one additional year. Therefore, inertia (t)E3.0. As seen in the second column of Table 1, the delayed logistic model explains far more variance (about 6.4% of the total) than the logistic model. The revised estimates for the model parameters suggest a lower carrying capacity for medical schools (about 137 for the entire United States), but a higher intrinsic growth rate (with the organizational population increasing by 4.3% of its lagged density each year, in the absence of resource or institutional constraints). Considering the hypothetical population trajectories plotted above, it is also of interest to consider how well the model does qualitatively in accounting for the boom-and-bust cycle among U.S. medical schools. A diagnostic procedure allows us to determine the stability of the model on the basis of the product of the constant term for structural inertia (t) and the intrinsic rate of population growth (r). Although the mathematics of this procedure are somewhat involved (see Nisbet and Gurney (1982) and appendix for computational details), the intuition behind it can be conveyed in a simple verbal fashion. Fluctuations arise in the delayed density model when there are temporal lags in density dependence and when the intrinsic rate of population growth is high. In organizational populations marked by especially fast growth (e.g. electronic commerce during the 1990s), even modest delays in entrepreneurial activity or the dissolution of failing incumbents may be sufficient to produce a boom-and-bust cycle. Conversely, organizational populations characterized by slow growth (such as American medical schools) are only likely to witness boom-and-bust cycles when structural inertia is high, at least within the delayed density specification. Using the terminology advanced by Hannan and Freeman (1984), the crucial determinant of instability in a population is therefore not structural inertia, but relative inertia, the extent to which organizational activities evidence delays relative to the intrinsic rate of population change.
Boom and Bust
41
Further insights into population development can be gained by diagnosing the joint occurrence of inertia and high intrinsic growth rates for any given population. A stability analysis of the delayed logistic model reveals that the steady state of any population trajectory can be categorized based on relative inertia, assessed via the product r t. The model has a stable steady state without oscillations when 0or toe1, a stable steady state with oscillations when e1or top/2, and an unstable steady state with oscillations when r t4p/2 (appendix; see Tuma & Hannan (1984) for a general discussion of stability analysis with deterministic non-linear models). For the population of U.S. medical schools, r t ¼ 0.043 3E0.13, which falls well below e1 (0.368). Given the assumptions of the delayed logistic model, the stability analysis provides no support for a boom-andbust cycle among U.S. medical schools. Several critiques might be raised to account for this result. One is that the underlying assumptions of the population-level model (e.g. identical structural inertia) cannot be maintained. I return to this concern shortly. Another critique is that the particular functional form of the density delay model is wrong. In particular, many organizational theorists may argue that inertia is especially prevalent for the entrepreneurial process – as exemplified by the case of Johns Hopkins medical school – but that selection pressures have a far more immediate impact on organizational survival. This argument echoes a common consensus in bioecology, which contends that only ‘‘birth’’ processes are plausibly linked to delayed density dependence (Turchin, 2003, p. 58). The solution is a hybrid specification that represents entry processes in terms of a delayed logistic model and exit processes in terms of a classic logistic model: dNðtÞ Nðt tÞ NðtÞ ¼ r1 Nðt tÞ 1 r2 NðtÞ 1 (3) dt K K where r1 is the intrinsic startup rate, r2 the intrinsic exit rate, and t the delay that applies specifically to startup activities (i.e. entrepreneurial inertia). Estimating this model for the evolution of U.S. medical schools yields the best model fit thus far, accounting for almost 20% of the variance in aggregate patterns of population growth and decline (see Table 1, third column). This suggests that inertia may operate asymmetrically for processes of organizational startup and exit. Further diagnosis of the model requires that we consider population stability over the response surface of all parameter combinations. As suggested by previous work in mathematical ecology (Nisbet & Gurney, 1982), two indicators of relative inertia are crucial in this respect – inertia relative to the intrinsic startup rate in an organizational
42
MARTIN RUEF
population (r1t); and inertia relative to the intrinsic exit rate (r2t). Although no analytical solution is available to characterize the response surface for model [3], Monte Carlo simulation allows us to explore population stability numerically over a large number of these parameter combinations (see Fig. 3). The qualitative behavior of the hybrid model with entrepreneurial inertia exhibits all three patterns observed in the simpler delayed logistic model. Thus, there are parameter combinations leading to a stable steady state with no oscillations (lower-left corner), a stable steady state with boom-and-bust cycles (middle-left), and an unstable steady state with cyclical behavior (upper-middle-left). In addition, large parts of the response surface reveal the possibility of extinction for the entire organizational population, given sufficiently high levels of inertia, startup rates, and/or exit rates. Based on the results presented in Table 1, we can locate U.S. medical schools on the Y-axis at log10(0.215 3)E0.190 and the X-axis at
Inertia × Startup Rate (Log)
1.00
0.50 Unstable (Cycles)
Stable (Cycles)
0.00
U.S. Medical Schools
-0.50
Stable (No Cycles)
Unstable (Extinction)
-1.00 -2.00
-1.00
0.00
1.00
Inertia × Exit Rate (Log)
Fig. 3.
Steady States of Organizational Populations in Model of Entrepreneurial Inertia.
43
Boom and Bust
log10(0.180 3)E0.268. Consistent with the stability analysis reported for the delayed logistic model, this suggests that we should not expect a boomand-bust cycle for this organizational population, based on these structural parameters alone. However, as the figure makes clear, the population is located in a rather tenuous position of the phase space. Even small deviations from the idealized assumptions noted at the beginning of this section may be sufficient to generate boom-and-bust cycles around a stable steady state or contribute to the extinction of the schools altogether.
Theoretical Formalization: Organization-Level If entrepreneurial inertia (t) is a crucial driver of instability in organizational populations, as the previous analyses suggest, then it is worth considering what consequences arise if there is heterogeneity in inertia and what the causes of such heterogeneity may be. As shown in Fig. 4, the assumption of a fixed lag time between preoperational and operational startup is not warranted for U.S. medical schools. The histogram is based on 309 medical schools that became operational before 1930, the nadir of the ‘‘bust’’ in school density (see JAMA 1901–1930, esp. 1908, 51(7), 594–602). The lag is calculated as the time between the date when the school was officially organized and when it graduated its first class of students. Seventy-six medical schools with missing information on entrepreneurial lag times are excluded. While the modal school graduates its first class only 1 year after being organized, some require as much as 26 years and others require only a matter of months. The figure excludes the lag time for right-censored cases (N ¼ 2 at the end of the observation window) and schools that never graduated a class of medical students (N ¼ 85). Notably, some of the inoperational schools exhibit extremely high levels of entrepreneurial inertia, with one organization reporting a 44-year lag between preoperational startup and dissolution. The heterogeneity in entrepreneurial lags may contribute to population dynamics through three mechanisms: (a) the rate of preoperational startups among new entrants; (b) the delay (and, therefore, disbanding risk) experienced by preoperational ventures before they achieve operational status; and (c) the exit rate of incumbent organizations. Preoperational Startup With respect to the first mechanism, our previous population-level models assumed that density-dependent effects evidence a lag in their influence on
44
MARTIN RUEF 200
175
150
Mean = 2.1 years SD = 2.9 years
125
100
75
50
25
0 0
Fig. 4.
5
10 15 20 Entrepreneurial Inertia
25
Distribution of Entrepreneurial Inertia among U.S. Medical Schools (1765–1930).
the startup rate in an organizational population. This lag reflects the fact that entrepreneurs tend to judge the attractiveness of a niche based on operational incumbents within it, rather than the total density of operational and non-operational organizations (see Lant & Baum (1995) for a discussion of biases in competitor definition). Partially, this results from simple measurement error – just as researchers experience great difficulty in enumerating nascent organizations that may not yet produce output, have an established address, and/or hire permanent employees (Aldrich, 1999; Kalleberg, Marsden, Aldrich, & Cassell, 1990), so do other entrepreneurs. The problem is compounded by the fact that some founders tend to be fairly secretive about their activities at this nascent stage. Another reason why the attractiveness of an organizational population is judged based on operational incumbents is that the survival of preoperational ventures can be highly precarious. Moore and Cain (2004) have
Boom and Bust
45
suggested that the strategic myopia of some entrepreneurs is explained by their neglect of fellow entrants. As organizational ecologists point out, this myopic logic also leads investors, regulators, and incumbents to discount the preoperational ventures, in favor of more tangible evidence regarding competition and legitimation within an arena of organizational activity (Carroll & Hannan, Chapter 15). As a result: Hypothesis 1. The density of operational organizations in a population has a stronger impact on the rate of entry than does total density (preoperational and operational). Two caveats concerning this hypothesis need to be considered. The first is that entries into a market niche are often subject to bandwagon effects. Indeed, much of the literature on financial boom-and-bust cycles emphasizes this aspect of investor psychology (e.g. Kindleberger, 1996). Among organizational founders, similar bandwagon effects may arise with respect to the activities of other preoperational startups with which an entrepreneur is acquainted. In some cases, knowledge of these startups is transmitted by interpersonal networks among entrepreneurs in a region (Sorenson & Audia, 2000). In other cases, a single entrepreneur (or entrepreneurial team) simultaneously initiates multiple startups of a particular organizational form. For instance, Johann Malok, a fraudulent medical practitioner in Chicago, personally organized six medical colleges in 1891 and 1892. Clearly, we want to separate such instances of local contagion in startup activity from the aggregate effect of entrepreneurial inertia. A second caveat pertains to competition with other entrepreneurs for resources when the founding rate is especially high. If investors, philanthropists, or regulators have recently backed the founding of a number of medical schools, can new entrepreneurs expect to approach those stakeholders again with requests for further endowments? Research in organizational ecology suggests that high founding rates in a given year depress founding rates in the following year, even when those entries involve preoperational ventures (Carroll & Hannan, 2000). Again, it is important to separate this process (emphasizing intense competition among new entrepreneurs) from the lagged density dependence implied by Hypothesis 1. Entrepreneurial Inertia and Organizational Form Boom-and-bust cycles become more likely when there is historical variation in the match between the operational density of incumbents and the total density of incumbents and nascent ventures. One factor that may predict this historical variation is the extent to which entrepreneurs are organizing
46
MARTIN RUEF
around a dominant design (Anderson & Tushman, 1990; Tushman & Murmann, 1998). The existence of a dominant design has two, somewhat opposing implications for entrepreneurial inertia. On the one hand, the existence of a dominant design means that entrepreneurs have to ‘‘get it right,’’ in the sense of organizing according to the normative standards of an externally prescribed architecture. Baron (2004) describes such organizational forms as involving sharp identities, insofar as they impose a large number of prescriptions that delineate them from other organizational forms. Entrepreneurs adopting these identities for their startups must engage in the time-consuming process of satisfying technical, structural, and personnel constraints, increasing startup lag times in the process. The dominant design that emerged among 19th century medical schools was that of the university-based, regular school. Many entrepreneurs hoping to start these medical schools felt that they had to invest time in setting up clinical laboratories, adopting prescribed curriculum standards, forming linkages with universities, establishing affiliated hospitals, and developing rigorous entrance exams (see Bevan (1908) on the features of this dominant design). The emergence of a dominant design in the field thus created a cultural barrier to entry for orthodox entrants. Founders who organized on the basis of less sharp criteria, such as those creating non-university based or sectarian medical colleges, had greater flexibility in choosing what administrative elements to incorporate and which ones to drop. This suggests that: Hypothesis 2. Entrepreneurial inertia increases when a venture is organized on the basis of a dominant design (e.g. university-based, regular schools). An opposing implication of the existence of a dominant design is that it often (though not always) implies that there are a large number of convenient organizational exemplars from which nascent entrepreneurs can draw ideas and inspiration. Mimetic processes may then simplify the adoption of organizational features and encourage conformity to prescribed features of the design (DiMaggio & Powell, 1983). As a consequence, when a normative organizational design also becomes numerically prevalent – and resonates with the identities of potential participants and stakeholders (Baron, 2004) – entrepreneurial inertia should decrease. Again, an example from the medical school domain serves to illustrate this mechanism. When Daniel Gilman was conceiving his plans for Johns Hopkins medical school, he had to travel to Europe to synthesize features that he saw in a variety of German universities. A few decades later, newly founded medical schools such as Duke, Rochester, and Vanderbilt could
Boom and Bust
47
readily emulate the (far more proximate) Johns Hopkins, as well as each other. By the same token, founders of eclectic institutions in the late 19th century could engage in selective imitation of regular schools, while also attending to the substantial number of sectarian organizations that blossomed during the period. In summary: Hypothesis 3. Entrepreneurial inertia decreases when entrepreneurs are able to draw on a large number of operational exemplars featuring a dominant design. Taken together, Hypotheses 2 and 3 imply that boom-and-bust cycles tend to appear just as organizational populations are shifting to a dominant form, but before that form is sufficiently widespread to alleviate entrepreneurial inertia. This corresponds to the state of American medical education at the end of the 19th century, where new prototypes such as Johns Hopkins were admired in some professional circles (e.g. the American Medical Association (AMA)), but a large number of sectarian medical colleges – including homeopathic, eclectic, physiomedical, and other alternative forms – continued to thrive. Organizational Exits Even as entrepreneurs shifted their efforts to the founding of universitybased, regular schools, it might be argued that another mechanism accelerated the bust that was already evident at the beginning of the 20th century. Specifically, when a large number of entrepreneurs initiated founding efforts employing this dominant organizational form, they competed intensively over resources – financing, philanthropy, staff, public goodwill – that could support the creation of any medical school at early stages of development. A likely consequence was attrition among schools that had yet to become operational; or, stated more generally: Hypothesis 4. Preoperational failure increases when there are a large number of organizations entering a population concurrently. The challenges of initial entry, entrepreneurial inertia, and preoperational failure are likely to filter many would-be organizations from ever appearing on the landscape of active participants in an organizational population. For those nascent ventures that do make it to the operational stage, though, our earlier discussion suggests that there may be benefits to an extended period of gestation. Some of these benefits may appear in tangible features of the organization’s form – such as the extent to which it corresponds to the dominant design in an industry – or in the scale of the newly operational
48
MARTIN RUEF
enterprise – predicated on its ability to attract a large staff (or, in the case of medical schools, a large number of potential students) during its period of incubation. As IO economists have pointed out, delays in committing sunk costs to a venture can also allow investors to obtain more information in uncertain environments (leading to investment theories of ‘‘optimal inertia’’ (Dixit, 1992)). Finally, the argument for the organizational benefits of entrepreneurial inertia extends to less tangible capabilities, such as the reliability and accountability developed by a nascent organization during a prolonged founding process (see Hannan & Freeman, 1984). Net of organizational form and scale, I propose that: Hypothesis 5. Operational organizations with long periods of preoperational gestation (i.e. entrepreneurial inertia) increase their likelihood of survival. This empirical claim may be challenged on the grounds that some unmeasured properties of entrepreneurial ability or resources may be correlated with both the duration of the preoperational lag and subsequent organizational survival, leading to a spurious association. However, omitted variables will typically contribute to underestimation of the effect of entrepreneurial inertia. On average, the unmeasured human capital of a medical school’s founding team, for instance, is likely to reduce the preoperational lag time, but improve the survival of the operational school. By the same token, the personal funds and philanthropic donations raised by a school ex ante (prior to preoperational initiation) will usually reduce the time needed to achieve operational startup, while improving subsequent survival chances. Given these assumptions, empirical tests of the effect of entrepreneurial inertia for individual organizations tend to be conservative. A second critique of the hypothesis may arise from its superficial similarity to the liability of newness argument, which argues that the disbanding risk of an organization declines monotonically with age (see Carroll & Hannan (2000) (Chapter 13) and Ruef (2002) for reviews of this literature). How can the effect of entrepreneurial inertia be disentangled from agedependence? As shown in Fig. 5, the predictions arising from the two mechanisms are actually quite distinctive. For the sake of simplicity, we will assume that the liability of newness is evidenced as the same monotonic decline across the organizational lifecourse – i.e. whether an organization is at a preoperational or operational stage. Consider the disbanding risks, then, of two organizations, where the second evidences a longer period of preoperational gestation (B4A). Upon becoming operational, the second has a lower disbanding risk than the first did when it became operational,
49
Disbanding Risk
Boom and Bust
A
B f (B – A)
Organizational Age
Fig. 5.
The Liability of Newness and Benefits of Entrepreneurial Inertia.
merely as a consequence of being further down the liability of newness curve. In addition, however, Hypothesis 5 posits that the second organization will enjoy further survival benefits upon becoming operational, as a discrete function of the differences in preoperational gestation f(B—A). Moreover, these benefits subsequently accrue to the second organization independently of the age-dependence curve. Stated another way, the impact of age dependence is contemporaneous, affecting organizations immediately at certain points in their lifecycle. Entrepreneurial inertia, by contrast, represents an imprinted effect, influencing organizational viability long after the startup cycle has been completed. A final critique of the entrepreneurial inertia thesis warrants attention. Perhaps the most important counter-argument in the strategy literature concerns markets with first-mover advantages (Lieberman & Montgomery, 1998). If there are considerable benefits for an organization that first enters a market, how should this be weighed against the putative advantages to entrepreneurial inertia? In the case of medical schools, in particular, firstmover benefits were likely to accrue when a school was the first in a regional market. Historically, legislatures were eager to provide funding and legitimacy to the first medical schools that opened their doors within state
50
MARTIN RUEF
boundaries. First-mover advantages may have also accrued geographically with respect to local philanthropists, faculty, and students. Consequently, my analyses of organizational viability control for potential benefits from early movement into unserved markets, while also addressing the conflicting demand for a careful and deliberative startup process, as postulated by Hypothesis 5.
DATA, MEASURES, AND METHOD Data My sample includes all medical schools in the United States, in existence between 1765 and 1930. The ending year for the analysis is based on two substantive considerations. First, in terms of numbers, 1930 represents the nadir of the medical schools, following the upsurge of the late 19th century. Analyses conducted through this year thus capture the full boom-and-bust cycle in the early history of American medical education. Second, it could be argued that the viability of medical schools after 1930 becomes increasingly subject to exogenous conditions, including the Great Depression, the development of regional health insurance markets, state funding for training programs, and NIH and private (foundation) funding for research (see Starr, 1982, pp. 352–359). Given the lack of comparability between the resource environment of early medical schools and the postwar years in particular, it seems appropriate to end analysis at this stage. Data were collected from several archives, including William Norwood’s (1944) Medical Education in the United States Before the Civil War, Nathan Davis’ (1877) Contributions to the History of Medical Education and Medical Institutions in the United States, John Rauch’s (1891) Report on Medical Education, JAMA’s (1901–1930) listings of contemporary and extinct medical schools in their annual series on medical education, and Frederick Waite’s (1946) history of sectarian medical schools. Supplemental information on the environment of medical schools was obtained from the Historical Statistics of the United States (1975). I have summarized the descriptive statistics for all operational schools and their environment in Table 2. Note that the units of analysis here are organizational spells (periods of observation with no change in covariates) and, consequently, these statistics may differ from those that are assessed cross-sectionally (cf. Fig. 4).
51
Boom and Bust
Table 2.
Descriptive Statistics for Population of Medical Schools (Total N ¼ 4,690 Organizational Spells).
Variable
Mean
S.D.
Minimum
Maximum
Organizational Age (years) Size (students) Regular schooly Eclectic schooly University-based Equal-status merger Inertia (years)
32.37 159.08 0.84 0.05 0.50 0.08 2.43
30.19 140.76 – – – – 3.28
0 0 0 0 0 0 0.5z
147 1055 1 1 1 1 26
Environmental First-mover Total density (lagged) Operational density (lagged) War years Medical Practice Act Examining board
0.20 123.58 107.66 0.06 0.84 0.73
– 39.81 36.84 – – –
0 1 0 0 0 0
1 174 154 1 1 1
–
0
1
Organizational mortality
0.04
Organizational statistics are restricted to 321 operational schools. y
Homeopathic and other sectarian schools (e.g. botanical, physiomedical) represent the omitted category. z Nineteen schools achieving operational startup in the same calendar year as their founding were coded as having a six-month lag time.
Measures Entrepreneurial inertia. One key dependent variable of interest is entrepreneurial inertia, or the lag time between the date when a public announcement is made that a school is being organized and the date when it is recognized as an operational enterprise (see Fig. 4). Since students are perhaps the most recognizable ‘‘output’’ from medical schools, I code operational startup as the year when the first students graduate. Initial public announcements regarding the organization of a medical school are also coded by year. Preoperational entry and exit events. Entries by preoperational medical schools are recorded when the organization of a new school is first announced, either as a de novo startup or as an equal-status merger of two or more pre-existing schools. Starting events were not registered when a school resumed operation after an extended hiatus (e.g. following a war).4
52
MARTIN RUEF
Exploratory analyses suggest that the startup process tends to occur more expeditiously for schools based on a prior educational infrastructure, with a significant negative correlation between equal-status mergers and entrepreneurial lag times (r ¼ 0.11, po0.05). Consequently, I control for mode of entry in models of both entrepreneurial inertia and exits among operational schools. Organizational exits subsume a variety of events, including voluntary dissolution, bankruptcy, removal of a legal charter, acquisition by another school, and declaration that a school is fraudulent by a state board of health. Equal-status mergers are excluded under this definition. For schools that never achieved operational startup, voluntary dissolutions and bankruptcies are the most common outcome, although a substantial number of these cases (40%) were declared to be fraudulent by state boards of health or other regulatory bodies. Organizational form. The form of early medical schools can be differentiated according to two major characteristics: (a) autonomy (i.e. whether the school was an independent institution or if it was affiliated with a college or university); and (b) the type of medical education being offered (i.e. whether the organization is a ‘‘regular’’ school, offering training in allopathic medicine, or a ‘‘sectarian’’ school, offering training in eclectic, homeopathic, physiomedical, botanical, or other form of medical practice). As suggested above, the dominant design that emerged among 19th century medical schools was that of the university-based, regular school. The eclectic schools represented one significant deviation from this model. Although two of these schools – the Lincoln Medical College of Cotner University and the University of Nebraska Department of Eclectic Medicine – had external affiliations, most were free-standing organizations. Eclectic schools promoted empirical training that sought to combine the best practices of other clinical approaches. They accepted much of allopathic medicine, but fought against the practice of drug-intensive treatment, particularly when it relied on non-naturally occurring substances. Homeopathic schools represented a more radical alternative to the dominant model of regular medical education. The homeopathic educators advanced a different philosophy of clinical treatment (the ‘‘Law of Similars’’), emphasizing the uniqueness of each patient (the principle of individualization) and the use of serially diluted drug doses (‘‘potentization’’) (Ullman, 1988). Despite the recurrent conflict that flared between the homeopaths and regular practitioners throughout the 19th century, homeopathic organizations enjoyed considerable legitimacy, with some half-dozen schools featuring university affiliations. Consequently, I code the homeopathic schools
Boom and Bust
53
as an organizational form that is distinct from both the regular and eclectic organizations.5 Other organizational characteristics. The analyses of organizational failure control for the size and age of the schools. Size is assessed via the number of enrolled medical students at each school. Age corresponds to the number of years that have passed since operational startup. I also control for the first-mover status of each medical school. A school is defined as a ‘‘first-mover’’ if it was the first to become operational in a given state. In one state (Illinois), two schools simultaneously graduated the first cohort of medical students from the state. Since neither school had clear precedence in this case, neither was coded as a first-mover. Environment. My analyses incorporate three contextual variables, addressing the influence of war, state-level licensing of physicians, and industry-specific period effects. State-level licensing presents one viable exogenous explanation for the boom-and-bust cycle among medical schools. Between the 1870s and early 1900s, all states witnessed the passage of medical practice acts that allowed licensure of physicians only when they presented medical school diplomas and/or were examined by independent state boards (Starr, 1982). The establishment of the independent state boards presented the greatest dilemma to marginal or unorthodox medical schools, since their graduates sometimes lacked training in a core curriculum of surgery, physiology, pathology, and the like; these school’s diplomas could even be rejected outright by state board examiners. To account for this contextual factor, I collected information on all state-level medical practice acts passed in the United States (see JAMA, 1901; Wilder, 1901, pp. 775– 835; Rauch, 1891) and distinguished those that created independent boards of examiners. A second possible exogenous explanation for the severe decline in U.S. medical schools was Abraham Flexner’s (1910) report on Medical Education in the United States and Canada, which condemned the large number of organizations that failed to conform to the university-affiliated, regular school form. In the aftermath of the report, Flexner (1960) himself suggested that ‘‘schools collapsed to the right and left, usually without a murmur’’. Medical historians have been less willing to ascribe many of these failures to the report. Based on self-identified reasons for school closure, Hiatt and Stockton (2003) suggest that only 7% of the schools evaluated may have closed as a direct result of the report; the reasons for merger or dissolution are unclear in another 35% of cases. To examine this impact more definitively, I include historical period effects in my models, distinguishing between (a) the era before the professionalization of medical practitioners in
54
MARTIN RUEF
the United States (through 1846); (b) the era of early professionalization, beginning with the formation of the AMA (1847–1910); and (c) the consolidation of professional authority for regular physicians, initiated by the Flexner Report (1911–1930). A number of other contextual variables were considered but are not featured in the following analyses, either because they provided very limited improvement in model fit or introduced problems of multicollinearity. To control for broad business cycles, I introduced a dummy variable for economic depressions. As suggested by a previous analysis (Ruef, 2004), the effects of this covariate fail to reach statistical significance in any of my models. I also examined a per-capita measure of physician density as a demand-side variable. Not surprisingly, this variable was highly correlated with the density of operational medical schools and was therefore dropped from further analysis. Missing Data Problems with missing data primarily affect the entrepreneurial inertia metric, which is unavailable for 76 schools that ultimately became operational. In analyses where the metric serves as an independent variable (i.e. as a predictor of medical school survival), there is no threat to internal validity (Berk, 1983). Nevertheless, I want to minimize attrition from the sample that could affect estimates of other covariates and, consequently, use imputation procedures to replace these missing cases (Little, 1992). The issue is more worrisome for the analyses predicting transitions to operational startup (per Hypotheses 2 and 3). Here, the cases have been dropped based on the dependent variable, raising serious concerns about selection bias. To assess the degree of potential bias, I ran a Heckman (1979) sample selection model. I reasoned that I was more likely to observe the startup lags for medical schools that (a) subsequently achieved large student enrollments; (b) were regular schools with university affiliations; and (c) existed (or were founded) after 1900, when the AMA instituted more systematic data collection efforts. Using logit selection, I found that there was strong evidence for (a) and (c) – large and recent medical schools were far more likely (po0.001) to have data on startup lags. The evidence for (b), on the other hand, was modest – regular schools were slightly more likely than sectarian schools to have lag data (po0.05), but there was no substantive difference between schools that were affiliated with universities and those that were not.
55
Boom and Bust
Since selection is driven primarily by variables (school size and recency) that are not featured as predictors of startup lag times, these results provide some preliminary support for the intuition that sample selection bias is not a severe problem in the following analyses. Further evidence was gleaned by comparing the results of OLS regression on the observed portion of the sample, with and without corrections for selection. The findings that include statistical corrections for selection effects are virtually identical to the standard OLS model (with one minor difference that is noted below in the discussion of findings). As a result, we have some confidence that the models which feature entrepreneurial inertia as a dependent variable do not suffer from significant selection bias. Statistical Methodology The statistical model of organizational founding and failure involves four distinct stages (see Fig. 6). In the first stage (A), potential founders decide whether or not they wish to initiate the founding of a new medical school. Since the risk set of potential founders is unobserved, this is modeled as an entry process generating annual event counts at the level of Potential Founders Hypothesis 1
A Pre-Operational Entry
Entrepreneurial Inertia
B
C
Hypothesis 4
Hypotheses 2 and 3 Operational Startup
Disbanding or Acquisition
D Hypothesis 5 Right-Censored Cases
Fig. 6.
Model of Organizational Startup and Exit Processes.
56
MARTIN RUEF
the entire organizational population. The count variable is overdispersed (i.e. its variance exceeds it means), reflecting the possibility of contagion in entries in particular years. Accordingly, the following negative binomial model (Cameron & Trivedi, 1986) is applied to predict preoperational entries: yt lðtÞut lðtÞut PðY t ¼ yt Þ ¼ e (4) yt ! where Yt is the estimated entry count in year t, yt the corresponding observed entry count, l(t) a linear function of the independent variables, and ut reflects specification error or contagion. Following preoperational entry, schools may either transition to operational startup (B) or preoperational disbanding (C). Both processes reflect entrepreneurial inertia, although the latter can be characterized as a right-censored case of it. A competing risks model (Blossfeld & Rohwer, 1995) is used to represent the rate with which preoperational medical schools transition into one of these alternative destination states. Because preoperational exits, in particular, are duration-dependent (with exit rates being especially high prior to the second year of an entrepreneurial effort), a piecewise exponential model was applied to capture this temporal variation: rðtÞ ¼ expðgn þ A0 X Þ if t 2 n
(5)
where r(t) is the transition rate for the medical schools; t indexes the entrepreneurial lag time; n indexes time periods corresponding to early stage (o 2 years) and late-stage (2+ years) entrepreneurial efforts; and gn is a constant coefficient associated with the nth time period. Other independent variables of interest (e.g. organizational form) are represented within the X matrix. For those medical schools that manage to graduate students, a final process of interest concerns the risk of disbanding, conditional on having achieved operational startup (D). Exploratory analyses reveal a classic liability of newness pattern in this risk over the lifetime of a medical school, with risks being highest during the first three decades of operation, declining during the following three decades, and dropping to very low levels thereafter. As in the case of preoperational exits, a piecewise exponential model can model this temporal variation, while accounting for age-independent effects arising from entrepreneurial inertia.
57
Boom and Bust
RESULTS Preoperational Entry Table 3 summarizes factors influencing preoperational entry among American medical schools through 1930. The first specification (see Model 1) corresponds to a classic density-dependence model in organizational ecology, in which the decision of entrepreneurs to initiate a school is driven by both operational incumbents in the niche, as well as schools that have yet to graduate students. Consistent with ecological theory, the log-linear term is positive and highly significant. The total density of schools initially serves to legitimate these organizations as a preferred form of medical education, as opposed to such earlier alternatives as medical apprenticeship or reliance on foreign medical graduates from Europe (Kaufman, 1976). The squared density term, on the other hand, is statistically insignificant. While ecological theory proposes that competition tends to increase geometrically, deterring entry at high levels of organizational density (Carroll & Hannan,
Table 3.
Negative Binomial Models Predicting Pre-Operational Entries among U.S. Medical Schools (1766–1930).
Variable
Model 1
Model 2
Model 3
Constant Log (total density) Total density squared Log (operational density) Operational density squared War year Medical Practice Actsy Period 2 (1847–1910) Period 3 (1911–1930) Prior entries Prior entries squared Overdispersion
2.439 (0.402) 0.908 (0.139) 0.017 (0.010) – – 0.246 (0.188) 0.097 (0.032) 0.064 (0.287) 2.167 (0.475) – – 0.158 (0.063)
2.379 (0.399) – – 0.971 (0.151) 0.037 (0.014) 0.239 (0.179) 0.092 (0.032) 0.020 (0.284) 2.276 (0.483) – – 0.166 (0.065)
2.341 (0.389) – – 0.864 (0.142) 0.037 (0.012) 0.268 (0.180) 0.058 (0.032) 0.225 (0.290) 1.951 (0.451) 0.193 (0.054) 0.009 (0.003) 0.094 (0.056)
Number of observations Condition index 2 log likelihood (df)
165 11.375 534.13 (7)
165 11.093 535.28 (7)
165 14.239 520.33 (9)
po0.05. po0.01 (two-tailed tests). y
Combined number of new state acts passed in current and previous year.
58
MARTIN RUEF
2000), the model appears misspecified when preoperational ventures are included in the density variable. Following the logic of Hypothesis 1, an alternative model predicts startup events exclusively on the basis of operational incumbents, the most visible participants in an organizational population (Model 2). Using this measure, the density-dependence account proves to be more satisfactory, with the startup rate increasing as a log-linear function of density (the legitimation effect in ecological theory, po0.01) and decreasing as a quadratic function of density (the competition effect, po0.01). These ecological effects are accompanied by two notable institutional processes: (a) the medical practice acts passed by most states in the late 19th century tended to generate sociopolitical legitimation for medical schools, spurring their formation; and (b) the curricular reforms called for by the Flexner Report served to depress medical school foundings after 1910. As noted in my theoretical discussion, the impact of density dependence on entries should not be conflated with either the bandwagon effect surrounding recent foundings nor with the competition these foundings may unleash for resources supporting preoperational ventures. To guard against this conflation, a final model includes covariates for the number of entries preceding any given year (Model 3). As identified in some previous ecological analyses (see Aldrich, 1999, pp. 268–270, for a review), local contagion contributes to positive autocorrelation up to a point, but large numbers of entries may consume many of the resources available for launching new ventures, deterring subsequent entrepreneurs. My findings concerning density dependence (based on operational ventures) continue to be robust in light of these processes. Entrepreneurial Inertia Table 4 provides estimates from a competing risk model, predicting the transition rate of newly founded medical schools to either operational startup (Model 1) or preoperational exit (Model 2). I find support for the idea that entrepreneurial inertia increases when founders adopt a dominant design – in this case, that of the university-affiliated, regular medical school (Hypothesis 2). Transition rates to operational startup decrease by roughly one-third for those ventures seeking linkages with a university and another third for ventures adopting a curriculum rooted in allopathic medicine. The findings thus support the intuition that dominant designs carry many normative elements which entrepreneurs find difficult to develop (e.g. clinical laboratories, in the case of the idealized medical school form at the end of
59
Boom and Bust
Table 4. Competing Risk Models Predicting Transition Rate to Operational Startup among U.S. Medical Schools (1766–1930). Variable
Operational Startup(Model 1)
Pre-Operational Exit(Model 2)
Duration (0–1 year) Duration (2+ Years)
0.823 (0.230) 0.551 (0.246)
2.852 (0.580) 3.931 (0.651)
Organizational form Regular schooly Eclectic schooly University based Equal-status merger
0.487 0.179 0.348 0.707
Organizational ecology Concurrent entries Density (regular schools)z
0.001 (0.016) 0.013 (0.004)
Environment War Year(s) Medical Practice Act Examining Board Period 2 (1847–1910) Period 3 (1911–1930)
0.017 0.402 0.176 0.520 0.191
Number of Events Condition index 2 log likelihood (df)
(0.177) (0.261) (0.145) (0.252)
0.557 0.756 0.803 1.423
(0.285) (0.474) (0.324) (1.020)
0.089 (0.023) 0.006 (0.007)
(0.189) (0.221) (0.196) (0.234) (0.406)
0.173 0.314 0.491 0.477 0.454
309
(0.411) (0.377) (0.323) (0.632) (1.178)
84 13.997 1633.37 (13)
po0.05. po0.01 (one-tailed tests for hypothesized effects, two tailed otherwise). y
Homeopathic and other sectarian schools (e.g. botanical, physiomedical) represent the omitted category. z Limited to density of operational schools.
the 19th century). Among those schools that achieve operational startup, the entrepreneurial lag is reduced slightly in organizational forms that deviate from the dominant design (e.g. eclectic schools) and reduced even further in forms that develop with little regard for the norms of the dominant design (homeopathic, botanical, and physiomedical schools). It is worth noting, however, that the standard error for the regular school dummy variable may be slightly underestimated, given sample selection effects. More specifically, results from a Heckman model suggest that the t-ratio decreases by approximately 15% once corrections for sample selection are included. Nevertheless, this still places the significance of the effect at the 0.01 level (one-tailed test).
60
MARTIN RUEF
As suggested by Hypothesis 3, entrepreneurial inertia is reduced when a population has a large number of organizational exemplars that are both operational and which conform to the dominant design. This effect is pronounced for medical schools transitioning to operational startup, but not quite significant for those that experience preoperational exits.6 A more decisive influence on exits involves concurrent entries, which tend to drain resources available for startup activities, and thereby increase the risk of preoperational failure (Hypothesis 4). Among environmental conditions, those related to the professionalization of the medical occupation have the most notable impact on entrepreneurial inertia. For the transition to operational startup, rates decrease significantly after the formation of the AMA (and other professional societies) around the middle of the 19th century. The passage of medical practice acts in various states also decrease startup rates, but this estimate is only marginally significant (po0.10). Arguably, both institutional activities generated new barriers to entry that had to be crossed for organizations seeking to define themselves as medical schools. Exits among Operational Organizations The preceding analyses have addressed how entrepreneurial inertia arises and how it may distort entrepreneurs’ perceptions of an organizational population by camouflaging large numbers of preoperational ventures. My final analysis addresses the effects of entrepreneurial inertia for those medical schools that experience operational startup (Table 5). As shown in the baseline model, a number of medical school characteristics are linked to exit events along lines consistent with previous organizational research (Model 1). Thus, medical schools exhibit a ‘‘liability of newness,’’ with recently operational schools exiting at a rate 5.89 times higher than more established schools (with 60+ years of operation). Medical school exits decrease significantly as a function of school size (as measured by the number of enrolled students) and affiliation with established universities. I also find that schools adopting the hybrid, ‘‘eclectic’’ form have a higher rate of exit (e0.57 ¼ 1.77) than either orthodox ‘‘regular’’ schools or their oppositional counterparts (e.g. homeopathic, botanical, and physiomedical schools). This result dovetails with research on organizational identities, which suggests that organizations that fail to conform to standard categorical expectations are likely to suffer from illegitimacy (Zuckerman, 1999; Po´los, Hannan, & Carroll, 2002).
61
Boom and Bust
Table 5.
Piecewise Exponential Models Predicting Exit Rate among Operational U.S. Medical Schools (1766—1930).
Variable Age (0–29 years since startup) Age (30–59 years) Age (60+ years) Organizational form Regular schooly Eclectic schooly University-based Equal-status merger Organizational features Size (# of students)z Inertia (Years)z
Model 1
Model 2
Model 3
3.064 (0.419) 3.491 (0.460) 4.529 (0.659)
2.689 (0.446) 3.125 (0.484) 4.132 (0.677)
2.300 (0.962) 2.266 (0.990) 2.397 (1.135)
0.041 0.570 1.031 0.366
0.100 0.584 1.007 0.504
0.053 0.569 1.024 0.510
(0.200) (0.258) (0.191) (0.354)
(0.201) (0.258) (0.190) (0.358)
(0.201) (0.258) (0.194) (0.359)
0.133 (0.018) –
0.131 (0.018) 0.334 (0.138)
0.128 (0.019) 0.255 (0.143)
Organizational ecology Total density (Logged) Total density Squared Density at founding (Logged) First Mover
– – – 0.837 (0.308)
– – – 0.830 (0.307)
0.950 0.036 0.894 0.590
(0.412) (0.020) (0.279) (0.313)
Environment War Year(s) Medical Practice Act Examining board Period 2 (1847–1910) Period 3 (1911–1930)
0.127 0.413 0.696 0.868 1.747
(0.252) (0.256) (0.256) (0.386) (0.435)
0.003 0.819 0.420 0.481 1.392
(0.260) (0.298) (0.274) (0.654) (0.733)
Number of events Condition index 2 log likelihood (df)
(0.251) (0.256) (0.256) (0.386) (0.435)
200 17.165 1658.38 (14)
0.087 0.459 0.709 0.867 1.793
200 18.656 1651.74 (15)
200 90.182 1637.85 (18)
po0.05. po0.01 (one-tailed tests for hypothesized effects, two tailed otherwise). y
Homeopathic and other sectarian schools (e.g. botanical, physio-medical) represent the omitted category. z To reduce skewness, these variables were transformed using a square root function.
With respect to the effects of the social environment, my findings here are also consistent with previous research, particularly among historians. One external catalyst to the shakeout among U.S. medical schools can be attributed to the creation of examining boards in various states in the late 19th and early 20th century (see Starr, 1982), which leads to more than a twofold
62
MARTIN RUEF
increase in school exit rates. The dissemination of the Flexner Report contributes to an equally dramatic rise in school closings and acquisitions after 1910. Finally, there is substantial first-mover effect for those schools that introduce medical education in a given state. The first movers reduce their odds of dissolution or acquisition by nearly 60%, reflecting their ability to secure ongoing support from state legislators and local philanthropists and to develop a regional identity that wards off potential challengers. In Model 2, I add a covariate for entrepreneurial inertia, the number of years that elapsed between the initial organization of a medical school and operational startup. Consistent with Hypothesis 5, those schools exhibiting substantial lags in their founding process are significantly less likely to face subsequent closure or acquisition. For instance, a school with a 4-year lag between initiation and startup is estimated to have an exit rate only half that of a school with no entrepreneurial lag. Medical schools with prolonged periods of founding were able to invest more time in resource mobilization, faculty recruitment, curriculum planning, staff socialization, and the development of clinical and teaching facilities. Inertia itself may also serve as a signal of the reliability and accountability of an educational institution (Hannan & Freeman, 1984). Notably, the effect holds even after controlling for the liability of newness and opposing first-mover benefits. A final model considers whether this finding holds up when covariates are added for a density-dependent specification of organizational exits (Model 3). Consistent with previous research on organizational ecology, the exit rate tends to fall with initial increases in population density and rise once competition intensifies. There is also evidence of a density delay effect. Medical schools that are founded during periods of high organizational density have persistently higher exit rates, supporting the ecological notion that adverse environmental conditions from the time of founding may be ‘‘imprinted’’ on formal organizations (Carroll & Hannan, 1989). The estimate for entrepreneurial inertia continues to be significant in this specification, although its magnitude is slightly reduced. A reason that may be offered for this result is the substantive overlap between the mechanism underlying entrepreneurial inertia and those underlying density delay. As Carroll and Hannan (2000) emphasize, one process contributing to density delay involves a liability-of-scarcity, in which intense competition at the time of founding leads preoperational organizations to scramble toward operational startup. The consequence is that new ventures ‘‘cannot devote much time, attention, and resources to organization building’’ (p. 241). Naturally, the same problem is implied by low levels of entrepreneurial inertia. It should also be noted that tolerance diagnostics suggest that there may be
63
Boom and Bust
collinearity in this model, as evidenced by the fairly high condition index (90.18). As Belsey, Kuh, and Welsch (1980) have discussed, condition indices over 20 are often linked to possible problems of collinearity, including inflated standard errors. In light of these theoretical and methodological considerations, it is unsurprising that entrepreneurial inertia is only significant at the 0.05 level in this model.
DISCUSSION The mechanism of entrepreneurial inertia suggests a basic paradox for the viability of organizational populations. On the one hand, careful planning and resource mobilization can enhance the survival of a nascent organization once it becomes operational. Entrepreneurs hoping to create sustainable ventures are thus advised to avoid rushing into startup activity. At the same time, the delay inherent in this process contributes to perceptual inaccuracies on the part of other organizations and entrepreneurs who want to forecast the competitive landscape in a niche. Depending on the amount of delay relative to intrinsic startup and exit rates in a population, entrepreneurial inertia can lead to a single boom-and-bust event, sustained boomand-bust cycles, or even the extinction of an organizational population. Although a meta-analysis of multiple industries would be required to explore the value of these predictions more definitively, several preliminary insights can be gained from an observation plan that focuses on a single organizational population. For U.S. medical schools, the paradox of entrepreneurial inertia finds empirical support in both population- and organization-level accounts. The population-level analysis provides a basic tool for diagnosing appropriate models of evolution in organizational populations. In this empirical case, the preferred model of population evolution is neither one that relies on classic density dependence (Hannan & Freeman, 1977) nor one that attributes uniform inertia to both startup and exit processes. Instead, a hybrid model seems most appropriate, emphasizing the delays that often accompany startup activities and the relative immediacy of factors contributing to organizational exit. A sensitivity analysis of this model suggests considerable potential for instability in organizational populations, given even small entrepreneurial delays. The level of relative inertia in a population (calculated as the product of average lag times and intrinsic growth rates) seems especially useful as a diagnostic parameter to identify when boom-and-bust cycles are likely to
64
MARTIN RUEF
occur and when organizational populations are likely to converge to their carrying capacity in a stable fashion. In other respects, the population-level model requires relatively few substantive assumptions for the occurrence of cycles. As long as market entry decisions on the part of entrepreneurs are dependent on the density of operational incumbents and those decisions are not revised significantly as a consequence of changes in density during the startup process, the structural conditions are ripe for boom-and-bust outcomes. An organizational model serves to flush out the processes contributing to entrepreneurial inertia. One crucial process parallels discussions of structural inertia more generally. Inertia in organizational structure and routines tends to be favored by society, because it connotes reliability in performance and accountability on the part of management (Hannan & Freeman, 1984). By the same token, entrepreneurs who take their time entering a market are thought to have invested time in ensuring reliable performance and documenting their activities for the sake of accountability. The resulting lag between organizational initiation and operational startup is also influenced by a number of structural factors, including the extent to which entrepreneurs seek to adopt the dominant organizational form in a population and the number of exemplars available for that dominant form. Beyond the specific application of boom-and-bust cycles in organizational populations, the framework presented here may have more general implications for both organizational theory and economic sociology. With respect to organization theory, this empirical investigation helps to connect two previously disparate strands of research – organizational ecology, which characterizes the evolution of industries in the aggregate (e.g. Carroll & Hannan, 2000), and entrepreneurial strategy, which emphasizes the activities of individual entrepreneurs or entrepreneurial teams (e.g. Ruef, Aldrich, & Carter, 2003). In drawing micro–macro connections between these strands of research, scholars are confronted with the obvious question as to how the actions of ordinary entrepreneurs, assessed on a timescale of months or years, could possibly make any difference in the evolution of industries, assessed on a timescale of decades or centuries (Carroll & Khessina, 2005). Surprisingly, analyses of entrepreneurial inertia imply that even relatively minute changes in the timing of founding processes can have profound implications for the vitality and composition of organizational populations. For economic sociology, a broader implication concerns the relevance of rational actors for the existence of stable markets. It has often been assumed that speculative bubbles in markets tend to arise due to overconfident entrants or the irrationality of bandwagon effects (Kindleberger, 1996;
Boom and Bust
65
Galbraith, 1997). On the other hand, if the culprits are methodical strategists, who carefully plan their entry into an industry or market so as to maximize their viability, then individual rationality and market instability may be ironic bedfellows. As Jovanovic (2004) demonstrates, markets tend to be most stable (in terms of price and output) when entrepreneurial lags are either zero or constrained to be equal. Heterogeneous lags, resulting from either structural constraint or entrepreneurial choice, lead to more complex market dynamics. Adam Smith’s self-interested ‘‘homo economicus’’ can therefore become the progenitor of a rather shaky ‘‘invisible hand’’. This raises several general questions for future research in economic sociology. How debilitating is rational planning on the part of market participants for a market as a whole? Is the problem mitigated or exacerbated by stronger forms of rational planning, in which entrepreneurs consider both their opportunity costs – e.g. in terms of wages foregone during an entrepreneurial lag – as well as the benefits received from a long-lived organization? When is nonstrategic social action crucial to the survival of economic markets or industries over the long-run? These issues are complicated by the fact that the number of levels at which inertia may be assessed are not two but three – individual, organizational, and population. Consider the basic paradoxes raised by structural inertia theory (Hannan & Freeman, 1984, 1989) and the parallels that may be identified with respect to entrepreneurial inertia (see Fig. 7). As suggested by both the classic theory of structural inertia and its logical reformulations (e.g. Pe´li et al., 1994, 2000), inertia may be a desirable property for specific organizations due to its association with the reproducibility of administrative structure and routines and, in turn, organizational reliability and accountability. At the same time, however, inertia is seen as problematic for the individual managers running such static organizations. It can serve as a signal of managerial conservatism – or even incompetence – and may lead stakeholders to question the usefulness of management more generally. Another paradox is apparent at the population level. Even though structural inertia can enhance the survival of individual organizations, its proliferation ultimately limits the evolution of entire organizational populations. This renders mature organizational forms vulnerable to displacement by novel arrangements. A similar set of mechanisms applies when we turn to entrepreneurial inertia. For nascent organizations, inertia during an entry period allows for the careful development of capabilities and resources. The benefits of inertia for the entrepreneurs themselves are more questionable. From an economic
66
MARTIN RUEF
Creates Vulnerability to Displacement from other Populations
Populations of Organizations
Contributes to 'Boom-andBust' Cycles in Populations
Associated with Reliability and Accountability of Organization
Individual Organizations
Allows New Organizations to Develop Capabilities and Resources
Signal of Managerial Conservatism / Incompetence
Individual Managers / Entrepreneurs
Reflects 'Opportunity Cost' of Entrepreneurship
Structural Inertia
Fig. 7.
Entrepreneurial Inertia
Hypothetical Impact of Organizational Inertia at Multiple Levels of Analysis.
perspective, they forego alternative opportunities to earn income or invest during this waiting period (Jovanovic, 2004). An organizational population as a whole may also suffer from high levels of entrepreneurial inertia, not owing to interpopulation competition as in the case of structural inertia, but due to instability within the population and a propensity toward cyclical expansion and consolidation. Inertia poses difficulties for organizational participants, social scientists, and policy makers, given inherent differences in its causal effects across levels of analysis. Moreover, there is often a strong inclination among observers to reinterpret the structural effects of inertia in psychological terms. Boom-and-bust cycles are thus characterized as arising due to overvaluation – i.e. investors’ departures from ‘‘market fundamentals’’ – rather than delays in investment or entrepreneurial activity. Similarly, organizational inaction is often seen as a problem of motivating – and adequately compensating – managers, rather than as an evolutionary by-product of organizational survival. For enlightened conceptions of strategic management, a central research task involves understanding the cultural conditions that encourage participants in modern society to analyze the effects of structural inertia along such individualist lines.
67
Boom and Bust
NOTES 1. This micro-analytical research is supported by the work of economic historians, who find that stock market booms such as the 1929 bull market tend to be driven by firms using new technologies (Rappoport & White, 1993, p. 551). 2. Since the functional form of the logistic model is non-linear, it is estimated using a Levenberg–Marquardt algorithm rather than OLS. The intrinsic growth rate (r) and carrying capacity (K) serve as model parameters. 3. The trajectories are generated using model (2), with K ¼ 150, r ¼ 0.25, N(0) ¼ N(1) ¼ N(2) ¼ 1, and t varying as shown in the figure. 4. Using this criterion, pre-operational entry necessarily occurs before operational startup, although an entrepreneurial ‘‘lag’’ of zero may be registered when a school is founded and graduates students in the same calendar year (19 cases). Previous research designs examining the effect of founding processes on population dynamics (e.g. Rao (2001); Carroll & Hannan (2000), Chapter 15) have not been able to rely on this strict temporal sequencing. 5. Aside from the homeopaths, other medical sects – including the botanical and physiomedical forms – represented fairly radical alternatives to allopathic medicine in the early 19th century (Waite, 1946; Starr, 1982). These subsequently fed into the less ‘‘deviant’’ form of eclectic medicine. 6. In separate analyses, I considered whether this effect may be curvilinear – i.e. if there is a point where the widespread prevalence of a dominant design actually inhibits the transition of schools to operational startup. No empirical support for this hypothesis was found.
REFERENCES Aldrich, H. (1999). Organizations evolving. Thousand Oaks, CA: Sage. Aldrich, H., & Fiol, C. M. (1994). Fools rush in? The institutional context of industry creation. Academy of Management Review, 19, 645–670. Amburgey, T., Kelly, D., & Barnett, W. (1993). Resetting the clock: The dynamics of organizational change and failure. Administrative Science Quarterly, 38, 51–73. Anderson, P., & Tushman, M. (1990). Technological discontinuities and dominant designs: A cyclical model of technological change. Administrative Science Quarterly, 35, 604–633. Barnett, W. (1997). The dynamics of competitive intensity. Administrative Science Quarterly, 42, 128–160. Barnett, W., & Carroll, G. (1995). Modeling internal organizational change. Annual Review of Sociology, 21, 217–236. Baron, J. (2004). Employing identities in organizational ecology. Industrial and Corporate Change, 13, 3–32. Baron, J., Hannan, M., & Burton, M. D. (1999). Building the iron cage: Determinants of managerial intensity in the early years of organizations. American Sociological Review, 64, 527–547. Barron, D. (1999). The structuring of organizational populations. American Sociological Review, 64, 421–445.
68
MARTIN RUEF
Belsey, D., Kuh, E., & Welsch, R. (1980). Regression diagnostics: Identifying influential data and sources of collinearity. New York: Wiley. Berk, R. (1983). An introduction to sample selection bias in sociological data. American Sociological Review, 48, 386–398. Bevan, A. (1908). Medical education in the United States: The need of a uniform standard. Journal of the American Medical Association, 51, 594–602. Blossfeld, H-P., & Rohwer, G. (1995). Techniques of event history modeling: New approaches to causal analysis. Mahaweh, NJ: Erlbaum. Cameron, A., & Trivedi, P. (1986). Econometric models based on count data: Comparisons and applications of some estimators and tests. Journal of Applied Econometrics, 1, 29–53. Carroll, G., & Hannan, M. (1989). Density delay and the evolution of organizational populations: A model and five empirical tests. Administrative Science Quarterly, 34, 411–430. Carroll, G., & Hannan, M. (2000). The demography of corporations and industries. Princeton, NJ: Princeton University Press. Carroll, G., & Khessina, O. (2005). The ecology of entrepreneurship. In: R. Agarwal, S. Alvarez, & O. Sorenson (Eds), Handbook of entrepreneurship: Disciplinary perspectives (pp. 167–200). New York: Springer. Carroll, G., & Wade, J. (1991). Density dependence in the organizational evolution of the American brewing industry across different levels of analysis. Social Science Research, 20, 271–302. Davis, N. S. (1877). Contributions to the history of medical education and medical institutions in the United States of America, 1776–1876. Washington: Government Printing Office. Delaney, K. (1992). Strategic bankruptcy: How corporations and creditors use chapter 11 to their advantage. Berkeley: University of California Press. Dew, N., Goldfarb, B., Sarasvathy, S. (2006). Optimal inertia: When organizations should fail. In: J. A. C. Baum, S. D. Dobrev, & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic Management (Vol. 23, pp. 72–102).. Oxford, UK: JAI/Elsevier. DiMaggio, P., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48, 147–160. Dixit, A. (1992). Investment and hysteresis. Journal of Economic Perspectives, 6, 107–132. Flexner, A. (1910). Medical education in the United States and Canada. New York: Carnegie Foundation for the Advancement of Teaching. Flexner, A. (1960). Abraham Flexner: An autobiography. New York: Simon and Schuster (originally published in 1940 as ‘‘I Remember’’). French, J. (1946). A history of the university founded by Johns Hopkins. Baltimore, MD: The Johns Hopkins Press. Galbraith, J. K. (1997). The Great Crash 1929. Boston: Houghton Mifflin (7th ed., first published in 1954). Granovetter, M. (1985). Economic action and social structure: The problem of embeddedness. American Journal of Sociology, 91, 481–510. Hannan, M. (1997). Inertia, density, and the structure of organizational populations: Entries in European automobile industries, 1886–1981. Organization Studies, 18, 193–228. Hannan, M., & Freeman, J. (1977). The population ecology of organizations. American Journal of Sociology, 82, 929–964. Hannan, M., & Freeman, J. (1984). Structural inertia and organizational change. American Sociological Review, 49, 149–164.
Boom and Bust
69
Hannan, M., & Freeman, J. (1989). Organizational ecology. Cambridge, MA: Harvard University Press. Hawkins, H. (1960). Pioneer: A history of the Johns Hopkins University, 1874–1889. Ithaca, New York: Cornell University Press. Heckman, J. (1979). Sample selection bias as a specification error. Econometrica, 147, 153–161. Hiatt, M., & Stockton, C. (2003). The impact of the Flexner report on the fate of medical schools in North America after 1909. Journal of American Physicians and Surgeons, 8, 37–40. Journal of the American Medical Association (JAMA). (1901–1930). Medical Education in the United States 37–66 (various volumes). Journal of the American Medical Association (JAMA). (1901). Laws regulating the practice of medicine in the various states and territories of the United States, 37, 1304–1319. Jovanovic, B. (2004). The pre-producers. Working Paper. Department of Economics, New York University. Kalleberg, A., Marsden, P., Aldrich, H., & Cassell, J. (1990). Comparing organizational sampling frames. Administrative Science Quarterly, 35, 658–688. Kaufman, M. (1976). American medical education: The formative years, 1765–1910. Westport, CN: Greenwood Press. Kimberley, J. (1979). Issues in the creation of organizations: Initiation, innovation and institutionalization. Academy of Management Journal, 22, 437–457. Kindleberger, C. (1996). Manias, panics and crashes: A history of financial crises. New York: Wiley (3rd ed., first published in 1978). Lant, T., & Baum, J. (1995). Cognitive sources of socially constructed competitive groups: Examples from the Manhattan hotel industry. In: W. R. Scott, & S. Christensen (Eds), The institutional construction of organizations (pp. 15–38). Thousand Oaks, CA: Sage. Lieberman, M., & Montgomery, D. (1998). First-mover (dis)advantages: Retrospective and link with the resource-based view. Strategic Management Journal, 19, 1111–1125. Little, R. (1992). Regression with missing X’s: A review. Journal of the American Statistical Association, 87, 1227–1237. Lomi, A., Larsen, E., & Freeman, J. (2005). Things change: Dynamic resource constraints and system-dependent selection in the evolution of organizational populations. Management Science, 51, 882–903. Meyer, M., & Zucker, L. (1989). Permanently failing organizations. Newbury Park, CA: Sage. Moore, D., & Cain, D. (2004). Strategic myopia and entrepreneurial entry: An experimental approach. Working Paper. GSIA, Carnegie-Mellon University. Nielsen, F., & Hannan, M. (1977). The expansion of national education systems: Tests of a population ecology model. American Sociological Review, 42, 479–490. Nisbet, R. M., & Gurney, W. S. C. (1982). Modeling fluctuating populations. New York: Wiley. Norwood, W. F. (1944). Medical education in the United States before the civil war. Philadelphia: University of Pennsylvania Press. Paich, M., & Sterman, J. (1993). Boom, bust, and failures to learn in experimental markets. Management Science, 39, 1439–1458. Pe´li, G., Bruggeman, J., Masuch, M., & Nualla´in, B. O´. (1994). A logical approach to formalizing organizational ecology. American Sociological Review, 59, 571–593. Pe´li, G., Po´los, L., & Hannan, M. (2000). Back to inertia: Theoretical implications of alternative styles of logical formalization. Sociological Theory, 18, 195–215.
70
MARTIN RUEF
Po´los, L., Hannan, M., & Carroll, G. (2002). Foundations of a theory of social forms. Industrial and Corporate Change, 11, 85–115. Rao, H. (2001). The power of public competition: Promoting cognitive legitimacy through certification contests. In: C. B. Schoonhoven, & E. Romanelli (Eds), The entrepreneurship dynamic: Origins of entrepreneurship and the evolution of industries (pp. 262–285). Stanford, CA: Stanford University Press. Rappoport, P., & White, E. (1993). Was there a bubble in the 1929 stock market?. Journal of Economic History, 53, 549–574. Rauch, J. (1891). Medical education, medical colleges and the regulation of the practice of medicine in the United States and Canada, 1765–1891. Springfield, IL: Illinois State Board of Health. Ruef, M. (1997). Assessing organizational fitness on a dynamic landscape: An empirical test of the relative inertia thesis. Strategic Management Journal, 18, 837–853. Ruef, M. (2002). Unpacking the liability of aging: Toward a socially embedded account of organizational disbanding. Research in the Sociology of Organizations, 19, 195–229. Ruef, M. (2004). For whom the bell tolls: Ecological perspectives on industrial decline and resurgence. Industrial and Corporate Change, 13, 61–89. Ruef, M. (2005). Origins of organizations: The entrepreneurial process. Research in the Sociology of Work, 15, 63–100. Ruef, M., Aldrich, H., & Carter, N. (2003). The structure of founding teams: Homophily, strong ties, and isolation among U.S. entrepreneurs. American Sociological Review, 68, 195–222. Schoonhoven, C., Eisenhardt, K., & Lyman, K. (1990). Speeding products to market: Waiting time to first product introduction in new firms. Administrative Science Quarterly, 35, 177–207. Sorenson, O., & Audia, P. (2000). The social structure of entrepreneurial activity: Geographic concentration of footwear production in the United States, 1940–1989. American Journal of Sociology, 106, 424–462. Starr, P. (1982). The social transformation of American medicine. New York: Basic Books. Sterman, J. (1989). Modeling managerial behavior: Misperceptions of feedback in a dynamic decision making experiment. Management Science, 35, 321–339. Tuma, N., & Hannan, M. (1984). Social dynamics: Models and methods. New York: Academic Press. Turchin, P. (2003). Complex population dynamics: A theoretical/empirical synthesis. Princeton: Princeton University Press. Tushman, M., & Murmann, J. P. (1998). Dominant designs, technology cycles, and organizational outcomes. In: B. Staw, & L. Cummings (Eds), Research in organizational behavior, 20 (pp. 231–266). reenwich, CN: JAI Press. Ullman, D. (1988). Homeopathy: Medicine for the 21st century. Berkeley, CA: North Atlantic. Waite, F. (1946). American sectarian medical colleges before the civil war. Bulletin of the History of Medicine, 19, 148–166. Wilder, A. (1901). History of medicine: A brief outline of medical history and sects of physicians. New Sharon, ME: New England Eclectic. Zuckerman, E. (1999). The categorical imperative: Securities analysts and the illegitimacy discount. American Journal of Sociology, 104, 1398–1438.
71
Boom and Bust
APPENDIX. STABILITY ANALYSIS Following Nisbet and Gurney (1982, pp. 287–289), I analyze the stability of the delayed logistic model through two simultaneous equations that prescribe the permissible behaviors of the model, given the controlling parameter for relative inertia (r t): m0 ¼ rt expðm0 Þ cos o0
(A1)
o0 ¼ rt expðm0 Þ sin o0
(A2)
where m0 is the so-called damping constant, indicating whether the population trajectory is stable (m0 X0) or unstable (m0 o0), and o0 is referred to as the natural frequency, indicating whether the population is subject to predictable oscillations (o0 40) or not (o0 ¼ 0). Cross-tabulating the two parameters leads to four qualitatively distinct types of population trajectories (Table A1). The first three (A–C) match the population trajectories displayed in Fig. 1, while the last (D) corresponds to an outcome in which the average number of participants in a population or market cannot be predicted over the long-run, even in the absence of exogenous changes or population heterogeneity. Table A1.
Analyzing the Stability of Organizational Populations. Natural Frequency (o0 ) ¼0
Damping constant (m0 )
X0 o0
(A) Stable (overdamped) (D) Unstable (no steady state)
40 (B) Stable (underdamped) (C) Unstable (limit cycles)
The simultaneous equations can be solved numerically for rt, thus indicating the levels of relative inertia leading to each of the population trajectories. Using the FINDROOT function in Mathematica, numerical solutions for m0 and o0 were obtained using values of relative inertia ranging from 0 (i.e. no entrepreneurial lag) to 2.5 (reflecting an extremely long lag and/or high intrinsic rate of population growth). Fig. A1 illustrates the three
72
MARTIN RUEF
2.0 Stable Stable (Oscillatory) 1.5 Unstable
1.0
0.5
0.0 Damping Constant
U.S. Medical Schools -0.5 0.0
Fig. A1.
Natural Frequency 0.5
1.0 1.5 Relative Inertia (r x Tau)
2.0
2.5
Steady States of Organizational Populations in Delayed Logistic Model of Growth.
distinctive types of population steady states corresponding to different values of relative inertia over the range.
OPTIMAL INERTIA: WHEN ORGANIZATIONS SHOULD FAIL$ Nick Dew, Brent Goldfarb and Saras Sarasvathy ABSTRACT We challenge the premise that the CEO’s job is to keep the corporation alive and thriving at all costs and under all circumstances. We briefly review the differing normative views of strategic management theorists and organizational theorists about organizational inertia. We then develop an economic model of incumbent behavior in the face of challenger competition that accommodates complementary assets. The model predicts and describes conditions under which organizational inertia, as subsequent organizational failure, is optimal. We then extend the logic and propose that the failure of entrepreneurial firms does not necessarily imply the failure of entrepreneurs. We conclude with a call to study ‘‘exit’’ as a viable strategic option.
$
Authors are listed alphabetically; all contributed equally.
Ecology and Strategy Advances in Strategic Management, Volume 23, 73–99 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23003-1
73
74
NICK DEW ET AL.
INTRODUCTION In its March 20, 2002 issue, The Onion carried the following news item: Corporation Reaches Goal, Shuts Down March 20, 2002|Issue 3810 AUSTIN, TX – After 18 years of striving, Dell Computer finally reached its long-stated goal to be the worldwide leader in computing systems Monday and promptly ceased operations.
Of course, The Onion is not a real newspaper. And no organization shuts down with the statement ‘‘Mission Accomplished’’ emblazoned on the wall behind a picture of its CEO. Yet, the spoof in The Onion poses an interesting puzzle, given that corporations were invented as useful instruments in the achievement of human/societal purposes, should not organizations sometimes choose to cease operations? If so, when and why? These puzzles are the topics we explore in this essay. There are many ways in which these questions can be approached and we cannot cover them all in one short chapter; here, we limit ourselves and approach these questions from the perspective of the innovation literature. We take the prototypical competitive innovation scenario featuring incumbents and challengers and use a model to show just a few of the circumstances under which organizational inertia – and subsequent exit – is optimal. Both the results of our model and much of the exit literature (Ghemawat & Nalebuff, 1985; Jovanovic, 1982; Knott & Posen, 2005; van Witteloostuijn, 1998) point to the need to further explore why much of strategy research so rarely differentiates between failure and exit. Within the literature there are several different views of failure: that firms fail because of incompetent managers who do not recognize changing market conditions and react to them appropriately; that they fail because they never develop appropriate capabilities for establishing a competitive advantage in dynamic environments; or that they fail simply as a by-product of their past successes. Yet much of this literature does not differentiate very well between failure and exit: exit can be forced (failure), or a deliberate and voluntary strategic gambit (hence a success story). In this paper we take-up this point, showing that inertia can be optimal for successful exit, just as it can be a terminal by-product of prior success or an organizational pathology. The paper is organized as follows. The second section outlines what we call the optimist’s view of the possibilities of keeping organizations alive even in the face of some serious problems posed by innovation dynamics. The third section discusses the rationale for pessimism about organizational
Optimal Inertia: When Organizations Should Fail
75
adaptation as argued by certain strategic management thinkers. The fourth section discusses the point of view of the extreme pessimists, a point of view espoused by organizational ecologists, who take inertia as central to the study of organizations. The fifth section presents the same problem from an economic perspective, coming to rather different conclusions about inertia. Instead of viewing inertia as an unwanted by-product of organizational design that cannot be overcome, it is considered an optimal choice for profit maximizing firms. Here we introduce a model (formally developed in the appendix) through which we articulate these arguments. Finally, we argue that from the perspective of entrepreneurship, organizational suicide often makes a lot of sense. Organizations should at times exit, cease operation, pack-up and go home. Many in fact do just that, though strategic management researchers rarely investigate this phenomenon as a choice. The wheel thus comes full circle, from the perspective of the promoters of organizational life, to a perspective that embraces organizational death as an optimal strategy.
THE OPTIMIST’S VIEW: STRATEGIC MANAGEMENT THEORISTS Most research in strategic management is motivated by the assumption that the CEO’s job is to keep the corporation alive and thriving at all costs and under all circumstances. The innovation literature is a good place to start looking for the key issues surrounding survival because scholars have long argued that technological change promoted by challenger firms can cause market disruptions that incumbent firms must master in order to survive (King & Tucci, 2002, p. 171). One of the latest wrinkles in this tapestry is Christensen’s ‘‘Innovator’s Dilemma’’ (Christensen, 2000), which points to cases where entrepreneurial firms initially touting inferior technologies eventually displace established firms. The ‘‘dilemma’’ consists in the fact that by doing the ‘‘right’’ thing, i.e., by listening to their customers, well-established and well-run companies such as IBM can end up losing those customers to relative newcomers – upstart firms that bring to market new technologies for which ‘‘no customers as yet exist’’ (Christensen & Bower, 1996). Christensen’s framework – while having its share of admirers and detractors – has certainly heightened practitioner awareness of the basic phenomenon of creative destruction described by Schumpeter, i.e., that innovation is a fundamental feature of competition. The literature spurred by his framework is another measure of its impact: we found 145 peer-reviewed articles that specifically refer to ‘‘Christensen’’, ‘‘the innovator’s dilemma’’, or both.1
76
NICK DEW ET AL.
A central feature of the innovator’s dilemma consists in the underlying mechanisms that make incumbents succumb to disruptive technologies introduced by challengers. Christensen and Bower (1996) excavated a process model for resource allocation within established firms faced with such disruptive challengers. This process model brought to the surface the fact that many of the new and disruptive technologies were either developed within established firms or were easily accessible to them, but were not pursued owing to estimates of high-market risk and/or low revenue and profit projections. While individuals within the established firms saw great potential in the new technologies, the firm’s processes were set up to allocate resources based on ‘‘rational’’ assessments of data about returns and risks: Projects targeted at the known needs of big customers in established markets consistently won the rational debates over resource allocation. Sophisticated systems for planning and compensation ensured that this would be the case. The contrast between the innovative behavior of some individuals in the firm, vs. the manner in which the firm’s processes allocated resources across competing projects, is an important feature of this model (Christensen & Bower, 1996).
Christensen has taken his insights gleaned from this empirical study of the disk drive industry and directly translated them into far-reaching prescriptions for practice (Christensen, 2000). Like many other strategic management theorists, he deplores corporations that are not ‘‘nimble’’ enough to cope with market changes, not only describing how this happens, but also prescribing how and why corporations can change their ways (Christensen & Raynor, 2003). An underlying but central proposition of this literature is that organizational inertia is a management problem that is fixable, at least in principle. While inertia may be difficult to solve, Christensen is optimistic that it is a force that better managers are able to overcome, and one to which incompetent managers succumb.
THE PESSIMIST’S VIEW: DIFFERENT STRATEGIC MANAGEMENT THEORISTS There are a number of reasons for thinking that the task of keeping the corporation alive and thriving at all costs and under all circumstances is actually even more difficult than the optimists suppose. To do so would require the firm to have a source of sustainable competitive advantage, a concept that has been the holy grail of strategic management. For instance, take the opening sentence of a seminal article by Teece et al.: ‘‘The
Optimal Inertia: When Organizations Should Fail
77
fundamental question in the field of strategic management is how firms achieve and sustain competitive advantage’’ (Teece, Pisano, & Shuen, 1997, p. 509). Yet it is a holy grail that has mostly proved elusive so far. In pursuit of it, the field has experienced several paradigm shifts, the latest perspective being that of ‘‘dynamic capabilities.’’ In dynamic market environments where new markets continually emerge, the ability to respond competitively is part of a class of organizational abilities called ‘‘dynamic capabilities’’ (King & Tucci, 2002, p. 171). What is needed is a strategic advantage in coping with change, a sort of meta-skill. The generic idea is that the ability to renew organizations occurs in the form of these types of dynamic capabilities: ‘‘New product development, as practiced in many firms, is a prototypical example of a y ‘dynamic capability’’’ (Winter, 2003, p. 992). However, the very existence of this ultimate source of sustainable advantage has been argued as an impossibility by some strategic management theorists (Collis, 1994; Winter, 2003). Collis (1994), for example, practically sounds the death knell of the concept as follows: Organizational capabilities, appropriately defined, can meet the conditions, articulated by the resource-based view of the firm, for being a source of sustainable competitive advantage. However, this paper observes that there are limits to the extent of the importance of such capabilities. They are vulnerable to threats of erosion, substitution, and above all to being superseded by a higher-order capability of the ‘learning to learn’ variety. This suggests that there can be an infinite regress in the explanation for, and prediction of, sustainable competitive advantage. The problem is resolved by arguing that the value of organizational capabilities is context dependent, and by recognizing that the strategy field will never find the ultimate source of sustainable competitive advantage (Collis, 1994, p. 143).
Winter (2003), building on Collis’ formulation of higher order capabilities, introduced the notion of ‘‘ad-hoc problem solving’’ as a counter-punch against the idea of dynamic capabilities: From a logical point of view, the ‘existence’ of higher order rates of change is in question only in the mathematical sense that some derivatives might not exist; and from a computational point of view, a time sequence of N+1 values of a variable suffices to compute one value of the Nth order rate of change. But if dynamic capabilities are similar to capabilities in that they involve patterned activity oriented to relatively specific objectives, then there is no guarantee that the organizational processes governing high-order change are highly patterned, and substantial reason to think otherwise. In this important substantive sense, high-order dynamic capabilities do not necessarily exist (Winter, 2003, p. 992).
In the face of these arguments the top management team’s task of keeping an organization alive becomes even more gargantuan: the task is being
78
NICK DEW ET AL.
nimble in the absence of any holy grail at all. Some are even more pessimistic. According to sociologists, there are even deeper difficulties the manager faces, owing to the inherent difficulties in changing organizations. It is to these arguments we now turn.
THE ‘‘REALLY’’ PESSIMISTIC VIEW: THE SOCIOLOGY OF ORGANIZATIONS For many researchers the central reality of strategic management is the concept of ‘‘organizational inertia.’’ As Ross and Staw (1993, p. 701) have put it: ‘‘Much of organizational theory can be reduced to two fundamental questions – how do we get organizations moving, and how do we get them stopped once they are moving in a particular direction?’’ Organization theorists working in sociological traditions are even more pessimistic than their management counterparts about the ability of organizations to overcome their established market positions, technological commitments and organizational structures, i.e., to stop persistently moving in the same direction and move, instead, in new directions. But instead of considering this some kind of pathology (Peli, Polos, & Hannan, 2000, p. 195), organizational ecologists think of inertia as a by-product of success. According to Hannan and Freeman’s (1984) original statement of structural inertia, the source of inertia lies in structural forces, not poor management. Peli et al. (2000, p. 196) state that: Hannan and Freeman (1984) argued that structural inertia arises as an inadvertent byproduct of a particular social selection process. They hold that inertia derives from the very characteristics that make organizations the favored kinds of corporate actors in contemporary societies: accountability and reliability. Both characteristics depend on a structure’s being reproduced with high fidelity over time. Yet, high reproducibility means that structures resist transformation. Therefore, core aspects of organization can be transformed only slowly and at considerable cost. Highly reproducible structures have a deadweight quality y [I]nertia can block transformation completely.’’
The underlying argument here will be familiar to economists: there is no free lunch. Organizational ecologists view inertia as a cost incurred in order to attain more valuable benefits. The process of environmental selection eliminates out firms that are not inert (i.e., are not sufficiently reliable and accountable) so inertia confers a survival advantage in a relatively stable environment. But when environmental disruptions are substantial, inert firms fail – not because of bad management – but because their managers had been in the past successful in using the firm’s design (architecture,
Optimal Inertia: When Organizations Should Fail
79
routines and culture) to underpin a competitive advantage. Therefore, removing inertia involves trade-offs in other dimensions (accountability and reliability) that are more valuable to organizations. Better management cannot ‘‘fix the problem.’’ Organizational ecologists have built a successful academic mini-industry by examining organizational dynamics based on assumptions about the role played by these structural factors (Dobrev, Kim, & Carroll, 2003; Dobrev, Kim, & Solari, 2004; Po´los, Carroll, & Hannan, 2002; Po´los & van Witteloostuijn, 2005; van Witteloostuijn, 1998, 2003). Ruef’s chapter in this volume is also informed by these ideas (Ruef, 2006). The underlying tenor of these arguments suggests that the strategic manager’s job is in fact futile in the face of environmental disruptions: there is no chance of corporations staying alive and thriving under all circumstances; nimbleness is simply not what organizations are about. Instead, inertia is structurally built into organizations, and failure in one period is often a concomitant of success in a previous one.
A CONTINGENT VIEW: THE ECONOMICS OF INERTIA Economists take a different approach to the incumbent-challenger interaction asking instead: under what circumstances is organizational inertia the optimal strategy for the firm? The economic incentives for incumbent firms to engage in uncertain innovative activities have been examined in detail in the literature on the economics of technological innovation and firm strategy (e.g., Reinganum, 1983). The central proposition of this stream of research is that incumbent firms will experience disincentives to create new technologies that disrupt existing technologies because the new technology cannibalizes the rent stream from the old. Non-incumbent challengers do not face this disincentive. So they rationally invest more and as a result will contribute a disproportionally large share of major innovations: ‘‘Since a successful incumbent merely ‘replaces himself’ y the incumbent firm has a lower marginal incentive to invest in R&D than does the challenger’’ (Reinganum, 1983, p. 741). The main addendum to Reinganum’s insight is that complementary assets play an important role by attenuating the disincentives incumbents face (Teece, 1986; Tripsas, 1997; Gans & Stern, 2002). For instance, large pharmaceutical firms control extensive sales networks and brands, which attenuate the threat of disruption by new products from biotechnology
80
NICK DEW ET AL.
start-ups and partly explain partnering arrangements between incumbents and challengers in the pharmaceutical industry. However, the role of complementary assets is complex and in need of further research (Helfat & Lieberman, 2002). Absent complementary assets, the incentive structure shows why the innovator faces a dilemma: incumbents face the unsavory prospect of having to decide when to start cannibalizing themselves in the full knowledge that considerable uncertainty pervades this choice. If they stay with their existing technology too long, they face being disrupted by an entrepreneurial attacker; if they migrate to a new technology too early, then they lose the rent stream from the old one. The uncertainty surrounding this choice is indeed deep, as Rosenberg has cautioned, ‘‘The simultaneous advance in new technology, along with the substantial upgrading of old technology, underlines the pervasive uncertainty confronting industrial decision makers in a world of rapid technological change’’ (Rosenberg, 1996, p. 107). In order to understand these trade-offs better, we have explored several extensions of the basic Reinganum model (see the appendix).2 Following Reinganum’s model, there is an incumbent and a challenger.3 Reinganum’s fundamental result is that the incumbent faces lower incentives vis-a`-vis the challenger to innovate if the output of the innovation will cannibalize its current revenue streams (a result we replicate in Proposition 2). Our extension demonstrates that Reinganum’s result is weakened if there is uncertainty as to whether the innovative activity will cannibalize the incumbent’s market and be reversed if the incumbent has significant complementary assets. However, it also shows that when an incumbent’s assets are detrimental to the pursuit of the new business, then Reinganum’s results are strengthened, the incumbent should invest less in researching potentially cannibalizing technologies. In the model, firms choose to invest in R&D that may lead to new technologies. Success is stochastic. Therefore, the model makes probabilistic statements about firm investment success, failure and consequent market positions. The model states conditions under which challenging firms will rationally invest more (less) than incumbent firms and hence be more likely to succeed (fail) in their innovation efforts. Markets are modeled as winnertake-all markets, and successful innovation by one player implies market exit of the other. Perhaps more importantly in our context, one can interpret low levels of research and development in the model as a strategy of purposeful inertia – as this is a decision to decrease the likelihood of change. Our model is consistent with observations that incumbent firms often do invest in the new technologies, as was the case with Polaroid and the digital
Optimal Inertia: When Organizations Should Fail
81
camera. While Polaroid proved unable to compete in the digital camera market and later filed for bankruptcy, the model suggests that perhaps it should not have invested more in digital technology, rather less. Upon realizing it had lost the digital camera race, it might have focused on extracting any remaining value from its existing businesses, and liquidated other assets, which is arguably its strategy coming out of bankruptcy. Of course, the model is also consistent with existing firms investing and at times succeeding and thereby warding off threats from incumbents. We begin with a summary of the basic findings of the model. First, Lemma 1 points out the expected result that the existence of the competitor encourages the incumbent to invest in research and development in an effort to defend its turf. Lemma 2 reports that the incumbent will be slower to react to the challenger’s investments in R&D than will the challenger to the incumbent so long as incumbent enjoys a rent stream from its current business. Note, this does not imply that in equilibrium, the incumbent will invest less than the challenger. Market Uncertainty Lemma 3 explores how Reinganum’s results change if the firm engages in R&D and it is uncertain whether, upon success, the firm’s current market will be cannibalized. As the model is stochastic, there is risk in this strategy – it could discover a technology that does indeed cannibalize its market.4 The key insight of Lemma 3 is that as the probability of cannibalization increases, the incumbent may still increase investment in R&D if the potential rewards of cannibalization are sufficiently high. The flip side implies that if the market post-cannibalization is weak, the incumbent will rationally under-invest. This result is akin to the implications of King and Tucci (2002) who found that firms chose not to enter next generation disk drive markets when they learned that these markets would be highly competitive. That is, even absent a story about complementary assets, an incumbent should be wary about entering the next generation market if it promises to be very competitive. Proposition 2 shows that in equilibrium, the incumbent indeed invests less than the challenger, and this is true even if the probability of cannibalization is low. Complementary Assets We take a very simplified view of complementary assets, in that we allow the benefits of successful R&D efforts to vary between the incumbent and the
82
NICK DEW ET AL.
challenger. When the value of the innovation is higher to the incumbent than to the challenger, this is a case when complementary assets are important. When the value of the innovation is higher to the challenger than to the incumbent, this is a case when the incumbent’s assets are detrimental. Lemma 4 states the obvious result that firms will react more aggressively when the potential benefits are higher. Lemma 5 reports the result that the incumbent may react more aggressively than the challenger when it has complementary assets that increase its valuation of the cannibalizing technology. Proposition 3 states the important alternative result that when the incumbent has significant complementary assets, in equilibrium the incumbent will invest more than the challenger even if the new technology will cannibalize its current revenue streams. In other words, if the incumbent possesses complementary assets, it will invest more aggressively than the challenger. Observing an incumbent’s successful defense of its market position may be indicative of strong complementary assets. Such a firm may be successful because there was not so much inertia to overcome. This result formalizes the arguments put forward originally by Teece and later by Gans and Stern, which explores the conditions under which a challenger will enter a market. Importantly, Proposition 4 also implies that as the challenger’s relative advantage grows, the incumbent will invest less and creative destruction at the hands of the challenging firm becomes more likely.
Multiple Challengers As the number of challengers increases, the incumbent’s chance of becoming unseated increases, therefore the incumbent will invest more so as to reduce the probability of this outcome. Furthermore, and while we do not prove this result, we speculate that in more competitive markets the likelihood that the incumbent will lose the race increases – and if it is a winner-take-all market, this implies that the likelihood of exit increases as well. This effect is similar to Aron and Lazear’s (1990) results where incumbents rationally follow challengers, and when the number of challengers increases, it becomes more likely that an incumbent will follow challengers.5 To be clear, the model does not necessarily predict firm failure, as it does not distinguish between organizational exit and an organization’s decision to exit a particular market. In cases where an organization’s activities are concentrated in a single market, however, this distinction is not important.
Optimal Inertia: When Organizations Should Fail
83
Bottom Line of the Model Under a fairly wide range of assumptions, in the absence of complementary assets, it is optimal for incumbents to exit, i.e., optimizing managers should utilize a strategy of inertia in order to maximize profits. It follows that: Inertia is not, from this perspective, an unwanted result of unavoidable and inherent ‘‘social processes’’; it is, at times, an optimizing decision on the part of strategic managers. Neither is inertia necessarily a function of the failure, or inability of managers to harvest the fruits of dynamic capabilities or to create those capabilities, regardless of the difficulties therein. Finally, inertia is not necessarily the result of management failure or mismanagement (Christensen & Bower, 1996). Efforts to improve or ‘‘fix’’ the management should not inevitably seek to overcome inertia, but merely be considered as one of many rational strategic choices on the part of the firm (King & Tucci, 2002).
AN INSTRUMENTAL VIEW: THE ENTREPRENEURSHIP OF INERTIA The model in this paper leads us to draw different conclusions and look in different theoretical directions than much of the strategic management literature that deals either directly or indirectly with the issue of organizational inertia. Under the circumstances outlined in the previous section, we see inertia and subsequent exit as fundamentally optimal for an organization. Organizations should fail themselves. This is equivalent to saying that firms should, under a fairly broad set of circumstances, adopt exit strategies. This leads us to our final discussion point: what would a strategic management that took an instrumental view of exit look like? First, let us consider some anecdotal evidence that illustrates an instrumental view of exit is more than just a toy problem, but has a grounding in organizational realities. Toys-‘R’-Us, the chain toy retailer – unable to compete with superstores Wal-Mart and Target in the retail toy business, has recently embarked on a long-term exit strategy. Wal-Mart, with its radical superstore technology, has been able to use toys as a loss-leader to draw customers into their stores, especially around the holiday season. At first Toys–‘R’-Us publicly flirted with the idea of outright exit, though in the
84
NICK DEW ET AL.
end they sold their assets to a group of private investors who closed all but the most profitable stores, killed their Kids-‘R’-Us and imaginarium brands and 182 related stores in 2003 and refocused attention on the more profitable Babies-‘R’-Us market segment (O’ Leary, 2003). We note that if the market trends continue as is, there is little reason to believe Toys-R-Us will not continue to exit individual markets, their recent sale to a group of private investors notwithstanding. Similarly, board member and business tycoon Carl Icahn recently lobbied to shut down Blockbusters video rental business in the face of looming competition from mail-order video rentals such as Netflix and increasing video-on-demand capabilities. Or, as Randal W. Forsyth wrote in the November 22nd, 2004 issue of Barron’s, Creative destruction is alive and well in retailing. The creative ones, such as Wal-Mart, Target, Home Depot, Lowe’s, Best Buy and Costco, not to mention Amazon.com, may well wind up destroying the likes of Sears and Kmart. If so, count on [controlling shareholder Edward Lampert] to redeploy [Sears and Kmart’s] assets in the quest for higher returns for shareholders, even if it means getting out of retailing. That is how capitalism is supposed to work.
There are two basic perspectives on instrumental exit that we wish to introduce here. Our view is that exit dynamics might be best understood by taking an integrated view of entry and exit together, i.e., considering strategic management as a long series of entries and exits that are interrelated in a non-trivial and potentially strategic fashion. Consider two cycles: a cycle of organizational entry and exit, and a cycle of exit and subsequent (re)entry. Taking first the entry and exit cycle, our suggestion is that the division of labor among strategy researchers has obscured the fact that many firms pursue integrated entry and exit strategies, sometimes described as ‘‘Hit and Run’’ strategies and predicted by contestable market theory (Baumol, 1982). The critical difference in looking at entry and exit together is that entry and exit costs can be offset by profits made during the firm’s tenure. Recent data is suggestive in this regard. For instance, in their study of exit in the banking industry, Knott and Posen (2005) found that ‘‘failed entrants’’ in their sample (those who entered and exited in the period they reviewed) in fact incurred no private losses. Instead, 790 firms had a mean net profit of $15.14 million over their tenure. The obvious question arising from this and other studies is, in what sense did these firms fail? A recent analysis of business closure by Headd found that, in fact, ‘‘about a third of closed businesses were successful at closure,’’ at least in the opinion of their owners (Headd, 2003, p. 51). We in strategic management have a habit of equating exit with failure, yet there is enough
Optimal Inertia: When Organizations Should Fail
85
accumulated evidence to suggest that a reformulation of some of our habits of analysis is in order. Instead of studying entry, or exit, we need to study the entry–exit cycle as an integrated whole, being mindful that it is quite possible that the entrepreneurs who found, manage and close or sell firms view this cycle as a single-strategic gambit – as one among a sequence of ‘‘hit and run’’ opportunities. Very deliberate ‘‘optimal inertia’’ may make considerable sense when viewed as a component of this strategy.6 In this connection, it is interesting to note Ruef’s paper in this volume, which relates the period during which entrepreneurs ‘‘prepare’’ to enter the market (a ‘‘pre-production’’ period) with population dynamics. Ruef’s story is that long-waiting times generate inertia that results in destabilizing the market at the macrolevel, even if the delays may be ‘‘rational’’ for the individual entrepreneur who is trying to carefully monitor his/her opportunity costs in deciding to found a new enterprise. Deliberate optimization of inertia leading to intentional exit may in turn mitigate or even prevent some of the boom and bust effects that result from organizational managers trying to react frantically to a dying niche. By ‘‘rationally’’ exiting with notice to key stakeholders to relocate and without cash reserves thrown away on futile attempts to keep the organization alive at all costs, markets could very well come to more peaceful ends even as their components get recycled and transformed into new industries. The second cycle to consider is the exit and (re)entry cycle. Our suggestion here is that for some purposes, such as exit and (re)entry, it might make sense to consider an organization as a diverse collection of stakeholders instead of as a holistic entity (Freeman, 1984). Through this lens we might view the entrepreneurial process as periodically tearing up old stakeholder configurations and reconstituting stakeholders around new organizational arrangements (Venkataraman, 2002). Given this, it might be useful to segment a long string of organizational entries and exits differently: instead of studying entry and subsequent exit, we can study exit as a prelude to subsequent (re)entry. This forward-looking reorientation of the ‘‘problem’’ of exit might encourage us to think that, from an entrepreneurial perspective, exit can make a lot of sense because it portends new entry. As ‘‘old’’ market rents erode because of changing environments, ‘‘new’’ opportunities may simply be better. Why, necessarily, should new opportunities be pursued by the same configuration of stakeholders that pursued the old opportunities? Perhaps it is useful to consider whether the costs of exit might be offset by subsequent benefits that a variety of stakeholders harvest upon (re)entry. For instance, Sarasvathy and Menon (2004) suggest that entrepreneurs
86
NICK DEW ET AL.
might amplify their expected success rates by exploiting learning that occurs through failed start-ups: ‘‘[F]or the multiple entrepreneur, each firm, whether successful or failed, is an instrument of learning that enables him or her to achieve better performance over time.’’ Similarly, King and Tucci (2002) interpret entry as a mechanism by which firms leverage prior management experience gained in an industry. This experience has a greater value if the firm enters a subsequent market; therefore King and Tucci observe that firms enter these markets in order to gain that increase in value (King & Tucci, 2002, p. 171). Taking these logics one step further, optimizing stakeholders would exit whenever they expect that subsequent (re)entry opportunities are more valuable than continuing with the current venture. This results in a perfectly rational and simple logic for what might be described as organizational euthanasia, self-murder or suicide. To understand this phenomenon properly we need much more data on voluntary versus involuntary exit, for both individual entrepreneurs as well as other investors, such as corporate venturers, venture capitalists and angels, all of which might be usefully viewed as managing temporal portfolios (instead of the conventional view that portfolios are contemporaneous diversification attempts). In all of these settings, the notion that exit costs are offset by the learning opportunities they engender for subsequent (re)entry has at least the air of plausibility about it. This suggests that there is some merit in analysis of strategic behavior based on an exit–(re)entry cycle, i.e., in taking an instrumental view of organizations. In closing, we stress that, in truth, this idea is neither new nor revolutionary. The joint stock company was invented so creative individuals could assume risk without putting their lives and livelihoods on the line for economic growth. Limited liability and the resultant immortality of the corporation were originally designed as instruments of individuals. In keeping with this design philosophy, organizational inertia can be optimal; we should study it as so. Firms should fail; it is sometimes optimal for them to do so. As researchers we should study exit more often, and when we do study exit, we need to study it with a view to the interaction of entry with exit and exit with (re)entry. The popularity of ecological metaphors notwithstanding, we need to consider organizational suicide, euthanasia and self-murder as rational actions in human-designed systems. As well as studying Darwin, researchers might also pick up Shakespeare and consider the human element in patterns of purposeful life – and death – therein. Et tu, Brute?7
Optimal Inertia: When Organizations Should Fail
87
NOTES 1. For this paper we searched EBSCO Business Resource Premier for all peerreviewed articles that contained the search terms ‘‘Christensen’’ and/or ‘‘innovator’s dilemma.’’ This resulted in 145 articles. From this sample, we short-listed the most relevant articles reviewed in this paper. 2. Baye and Hoppe’s (2003) results find a strategic equivalence between winner-take-all games and post-innovation rent-sharing games so long as there is a positive association between investment and profits. Thus, our results relating R&D expenditures to complementary assets are quite general. Under certain assumptions, however, this result goes away. For example, when there are information spillovers (i.e., a follower can learn from the leader about profitability), there can be an advantage to waiting, i.e., a second mover advantage (Hoppe, 2000). 3. For theoretical exposition, in our model we assume firms use discrete strategies and stick to modeling the ‘‘pure’’ game played by an incumbent and a challenger. In reality firms might use mixed strategies, i.e., intermediate between inertia (old technology) and innovation (new technology). Such mixed strategies are often referred to as striking a balance between exploration and exploitation (March, 1991). 4. In the model, the firm will, at this point, proceed with cannibalization. However, one might imagine it attempting to suppress the results and, perhaps, rogue employees starting spinouts. 5. Aron and Lazear do not assume winner-take-all markets. Hence, in their model one can observe leaders and followers in the new market, whereas in our model the loser exits. 6. The reader might consider two game theoretic literatures that might be viewed as complementary to the model we use in this paper. van Witteloostuijn (1998) explores the issue of optimal inertia by seeking to calculate how inertia maximizes a focal firm’s profits in the face of flexible rivals. There, flexibility leads to exit. In the industrial organization literature (Ghemawat & Nalebuff, 1985) sunk costs are associated with commitments and inertia. Here flexibility is also associated with exit. These models are independent from model we explore in this paper, which does not depend on sunk costs or flexibility to generate its results. These different models might therefore be seen as complementary to one another. 7. ‘‘Et tu, Brute?’’ – from Julius Caesar (III, i, 77). According to Moore (2002): ‘‘Perhaps the most famous three words uttered in literature, ‘Et tu, Brute?’ (Even you, Brutus?) this expression has come down in history to mean the ultimate betrayal by one’s closest friend. This scene, in which the conspirators in the Senate assassinate Caesar, is one of the most dramatic moments on the Shakespearean stage. The audience has just witnessed the arrogance and hubris of a ruler who has sought, within a republic, to become a monarch, comparing himself to the gods. Brutus, a friend of Caesar and yet a man who loves Rome (and freedom) more, has joined the conspirators in the assassination, a betrayal which is captured by the three words above.’’
88
NICK DEW ET AL.
REFERENCES Aron, D. J., & Lazear, E. P. (1990). The introduction of new products. American Economic Review, 80(2), 421–426. Baumol, W. (1982). Contestable markets: An uprising in the theory of industry structure. American Economic Review, 72, 1–15. Baye, M. R., & Hoppe, H. C. (2003). The strategic equivalence of rent-seeking, innovation, and patent-race games. Games and Economic Behavior, 44, 217–226. Christensen, C. M. (2000). The innovator’s dilemma. Boston, MA: Harvard Business School Press. Christensen, C. M., & Bower, J. L. (1996). Customer power, strategic investment and the failure of leading firms. Strategic Management Journal, 17, 197–218. Christensen, C. M., & Raynor, M. E. (2003). The innovator’s solution: Creating and sustaining successful growth. Boston, MA: Harvard Business School Press. Collis, D. J. (1994). How valuable are organizational capabilities? Strategic Management Journal, 15, 143–148. Dasgupta, P., & Stiglitz, J. (1980). Uncertainty, industrial structure, and the speed of R&D. Bell Journal of Economics, The RAND Corporation, 11(1), 1–28. Dobrev, S. D., Kim, T. Y., & Carroll, G. R. (2003). Shifting gears, shifting niches: Organizational inertia and change in the evolution of the U.S. automobile industry, 1885–1981. Organization Science, 14(3), 264–282. Dobrev, S. D., Kim, T. Y., & Solari, L. (2004). The two sides of the coin: Core competence as capability and obsolescence. In: J. A. C. Baum, & A. M. McGahan (Eds), Business strategy over the industry lifecycle: Advances in strategic management (Vol. 21, pp. 255– 287). JAI Press. Freeman, R. E. (1984). Strategic management: A stakeholder approach. New York: Pitman Publishing. Gans, J. S., & Stern, S. (2002). The product market and the market for ‘Ideas’: Commercialization strategies for technology entrepreneurs. Research Policy, 1441(32), 333–350. Ghemawat, P., & Nalebuff, B. (1985). Exit. Rand Journal of Economics, 16, 184–194. Hannan, M. T., & Freeman, J. (1984). Structural inertia and organizational change. American Sociological Review, 49, 149–164. Headd, B. (2003). Redefining business success: Distinguishing between closure and failure. Small Business Economics, 21, 51–61. Helfat, C. E., & Lieberman, M. B. (2002). The birth of capabilities: Market entry and the importance of pre-history. Industrial and Corporate Change, 11(4), 725–760. Henderson, R. (1993). Underinvestment and incompetence as responses to radical innovation: Evidence from the photolithographic alignment equipment industry. Rand Journal of Economics, 24(2), 248–270. Hoppe, H. C. (2000). Second-mover advantages in the strategic adoption of new technology under uncertainty. International Journal of Industrial Organization, 18, 315–338. Jovanovic, B. (1982). Selection and the evolution of industry. Econometrica, 50(3), 649–670. King, A. A., & Tucci, C. L. (2002). Incumbent entry into new market niches: The role of experience and managerial choice in the creation of dynamic capabilities. Management Science, 48(2), 171–186. Knott, A. M., & Posen, H. E. (2005). Is failure good? Strategic Management Journal, 26, 617–641.
Optimal Inertia: When Organizations Should Fail
89
March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2(1), 71–87. Moore, R. (2002). Julius Caesar: Et tu, Brute? In: P. Satoris (Ed.), eNotes: Julius Caesar. Seattle: Enotes.com LLC, October. Accessed 19 December 2005. http://www.enotes.com/jc/257. O’Leary, C. (2003). Toys ‘R’ us in trouble again: Muchtouted 2002 restructuring falls prey to Wal-Mart pricing steamroller. Investment Dealers Digest, Nov. 24. Peli, G. L., Polos, L., & Hannan, M. T. (2000). Back to inertia : Theoretical implications of alternative styles of logical formalization. Sociological Theory, 18(2), 195–215. Po´los, L., Hannan, M. T., & Carroll, G. R. (2002). Foundations of a theory of social forms. Industrial and Corporate Change, 11, 85–115. Po´los, L., & van Witteloostuijn, A. (2005). Who should dare to change? The theory of organizational inertia. Working Paper. University of Durham. Reinganum, J. F. (1983). Uncertain innovation and the persistence of monopoly. The American Economic Review, 73(4), 741–748. Rosenberg, N. (1996). Uncertainty and technological change. In: J. C. Fuhrer, & J. Sneddon Little (Eds), Technology and growth: Conference series no. 40, Boston, MA: Federal Reserve Bank of Boston. Ross, J., & Staw, B. M. (1993). Organizational escalation and exit: Lessons from the Shoreham Nuclear Power Plant. Academy of Management Journal, 36(4), 701–732. Ruef, M. (2006). Boom and bust: The effect of entrepreneurial inertia on organizational populations. In: J. A. C. Baum, S. D. Dobrev & A. van Witteloostuijn (Eds) Strategy and ecology: Advances in strategic management (Vol. 23, pp. 29–72). Oxford, UK: JAI/ Elsevier. Sarasvathy, S. D., & Menon, A. R. (2004). Failing firms and successful entrepreneurs: Serial entrepreneurship as a temporal portfolio. Working Paper. Teece, D. J. (1986). Profiting from technological innovation: Implications for integration, collaboration, licensing and public policy. Research Policy, 15, 285–305. Teece, D. J., Pisano, G., & Shuen, A. (1997). Dynamic capabilities and strategic management. Strategic Management Journal, 18(7), 509–540. Tripsas, M. (1997). Unraveling the process of creative destruction: Complementary assets and incumbent survival in the typesetter industry. Strategic Management Journal, 18, 119–142. Van Witteloostuijn, A. (1998). Bridging behavioral and economic theories of decline: Organizational inertia, strategic competition and chronic failure. Management Science, 44(4), 501–519. Venkataraman, S. (2002). Stakeholder value equilibration and the entrepreneurial process. Business Ethics Quarterly, Ruffin Series #3. Winter, S. G. (2003). Understanding dynamic capabilities. Strategic Management Journal, 24(10), 991–995.
90
NICK DEW ET AL.
APPENDIX: MODEL A Model under Market Uncertainty Our model is an extension of Reinganum (1983). However, in this model we only consider technologies that are radical, in the sense that they may disrupt current markets or create new markets. Indeed, it is this second possibility, that R&D expenses may produce a technology that creates a new market and does not cannibalize an incumbent’s existing market that distinguishes our model from that of Reinganum. We begin our discussion with a description of the model under conditions of technological certainty, but allow market uncertainty. We consider a case in which there are two firms, an incumbent and an entrant. The incumbent enjoys a revenue stream R. The firms are engaged in an innovation race. If the incumbent wins the race and patents the new technology, it earns a present value of monopoly profits from the invention, P2 : However, there is a probability that the new technology is not a substitute for the incumbent’s product, and instead can be sold in a complementary market. This probability is a. The present value profit in the new market is P1 : Thus if the incumbent wins the patent race, it earns aðP1 þ RÞ þ ð1 aÞP2 : If the entrant wins the patent race, then it earns aP1 þ ð1 aÞP2 : Note that we have assumed winner-take-all markets that are common in high-technology industries. The difference between the two expressions is the continued flows, R, from the current market that the incumbent enjoys when the new technology does not make its old technology obsolete. Following Gilbert and Newbery (AER 82), if the invention process is deterministic, then whoever is willing to bid the most for the new technology receives the first patent with probability 1. The incumbent will be willing to bid up to aðP1 þ RÞ þ ð1 aÞP2 while the challenger will be willing to bid up to aP1 þ ð1 aÞP2 . Since the value is always higher to the incumbent, the incumbent will preemptively patent the new technology. Note that this differs from the result of Gilbert and Newbury in that when there is market uncertainty, but no technological uncertainty, the incumbent always preemptively patents – even in the case of a radical innovation. In the above discussion we have implicitly assumed that the needed investment, x , is less than the challenger’s expected profits aP1 þ ð1 aÞP2 , a condition necessary to ensure the challenger enters. If there is no credible challenger, that is x 4aP1 þ ð1 aÞP2 ; the incumbent will still invest if aðP1 þ RÞ þ ð1 aÞP2 x 40: Rearranging this condition we find aR4 x ðaP1 þ ð1 aÞP2 Þ: if the probability of the innovation leading to
Optimal Inertia: When Organizations Should Fail
91
a new market is high enough, the firm will invest in it even though there is no threat of entry and the innovation may cannibalize the firm’s own market. This is due to the incumbent betting that the new technology will lead to a new market opportunity. This outcome becomes more likely when the value of the current technology, R, is low, or the market potential for the new technology aP1 þ ð1 aÞP2 is high. Assumption 1. Monopoly profits after the innovation P2 always exceed the present value of the monopoly profits before the innovation R=r; where r is the discount rate. Assumption 1 would be implied if we explicitly assumed that the innovation was cost-reducing.
A Model under Technological Uncertainty An incumbent and a challenger are engaged in a search for a particular radical technology. With probability a the technology will not cannibalize the incumbent’s existing market and with probability 1 a the technology will cannibalize the existing market. This characterizes the market uncertainty. We borrow the characterization of technological from Reinganum (1983). In particular, let xI represent the rate of investment of the incumbent and tI ðxI Þ the random success date of the incumbent. Following an exponential distribution, PrftI ðxI Þ tg ¼ 1
e
hðxI Þt
for t 2 ½0; 1Þ: Likewise, xC represents the rate of investment of the challenger and tC ðxC Þ the random success date of the challenger, so PrftC ðxC Þ tg ¼ 1 e hðxC Þt : 1=hðxi Þ is the expected success date for firm i, where i 2 fI; Cg: We note that h(xi) is the hazard function. Following Reinganum (Assumption 2), we make the common assumption that the R&D process exhibits decreasing returns to scale: Assumption 2. The hazard function h( ) is twice continuously differentiable, with h0 ðxÞ40 and h00 ðxÞo0 for all x 2 ½0; 1Þ: Furthermore, hð0Þ ¼ 0 ¼ limx!1 h0 ðxÞ: We assume perfect intellectual property right protection so that success is preemptive. The incumbent receives P1 if success leads to a new market when this occurs; it retains its revenue stream R from the old market. This outcome occurs with probability density ahðxI Þe ðhðxI ÞþhðxC ÞÞt : With
92
NICK DEW ET AL.
probability density ð1 aÞhðxI Þe ðhðxI ÞþhðxC ÞÞt technological success will lead to market cannibalization, and a (NPV) profit P2 : An important distinction between this model and that of Reinganum is that the incumbent can retain its market position even when the challenger patents preemptively if the new technology is not a cannibalizing one. The probability density for this event is ahðxC Þe ðhðxI ÞþhðxC ÞÞt : With probability density e ðhðxI ÞþhðxC ÞÞ neither firm realizes technological success and the incumbent continues to earn R and incurs the flow costs xI. Hence, the expected profit for the incumbent conditional on the pair of investment rates ðxI ; xC Þ is Z 1 e rt e ðhðxI ÞþhðxC ÞÞt V I ðxI ; xC Þ ¼ 0 ahðxI ÞðP1 þ R=rÞ þ ð1 aÞhðxI ÞP2 þ ahðxC ÞR=r þ R xI dt ahðxI ÞðP1 þ R=rÞ þ ð1
¼
aÞhðxI ÞP2 þ ahðxC ÞR=r þ R
r þ hðxI Þ þ hðxC Þ
xI
ð1Þ
The challenger can benefit only from success and receives a profit stream P1 if the product does not cannibalize the incumbent’s existing market, an event that occurs with probability a. With probability 1 a the product does cannibalize the incumbent’s existing market and the challenger receives P2 : Note that Reinganum’s model is a special case of this model when a ¼ 0. When a ¼ 1, the existing profit stream does not influence the maximization problem, and the game is a standard patent race (cf. Dasgupta & Stiglitz, 1980). The challenger’s payoff is Z 1 C e rt e ðhðxI ÞþhðxC ÞÞt ½ahðxC ÞP1 þ ð1 aÞhðxC ÞP2 xC dt V ðxI ; xC Þ ¼ 0
ahðxC ÞP1 þ ð1
¼
aÞhðxC ÞP2
r þ hðxI Þ þ hðxC Þ
xC
ð2Þ
Note that the incumbent’s payoff in Eq. (1) is equal to the challenger’s payoff in Eq. (2) when R ¼ 0. The incumbent’s strategy is a choice of xI, while the challenger’s strategy is a choice of xC. We solve the model as follows. We define the best response function of the two players: fI : ½0; 1Þ ! ½0; 1Þ such that V I ðfðxC Þ; xC Þ V I ðxI ; xC Þ 8xI 2 ½0; 1Þ for the incumbent and fC : ½0; 1Þ ! ½0; 1Þ such that V C ðxI ; fðxI ÞÞ V I ðxI ; xC Þ 8xC 2 ½0; 1Þ: The best response functions will also depend upon parameters (a, R). We now search for a Nash equilibrium.
Optimal Inertia: When Organizations Should Fail
93
Proposition 1. The best response functions are well behaved and there exist Nash equilibrium strategies. Proof. See Reinganum (1982), Proposition 1. & The first-order conditions implicitly define the best response functions @V I ðfI ; xC Þ 0 / ah ðfI ÞðP1 þ R=rÞ þ ð1 aÞh0 ðfI ÞP2 1 r þ hðfI Þ þ hðxC Þ @xI h0 ðfI Þ ahðfI ÞðP1 þ R=rÞ þ ð1 aÞhðfI ÞP2 þahðxC ÞR=r þ R fI ¼ 0 @V C ðfI ; xC Þ 0 / ah ðfC ÞðP1 þ ð1 aÞÞh0 ðfC ÞP2 1 r þ hðxI Þ þ hðfC Þ @xI h0 ðfC Þ ahðfC ÞP1 þ ð1 aÞhðfC ÞP2 fC ¼ 0
Noting the definitions of V I ðfI ; xC Þ and V C ðxI ; fC Þ we can rearrange and find 0 ¼ ah0 ðxI ÞðP1 þ R=rÞ þ ð1 aÞh0 ðxI ÞP2 1 ½r þ hðxI Þ þ hðxC Þ h0 ðxC ÞV I ðfI ; xC Þ½r þ hðxI Þ þ hðxC Þ
Rearranging V I ðfI ; xC Þ ¼
h0 ðfI ÞðaðP1 þ R=rÞ þ ð1 h0 ðfI Þ
aÞP2 Þ
1
Similarly V C ðxI ; fC Þ ¼
h0 ðfC ÞðaP1 þ ð1 aÞP2 Þ h0 ðfC Þ
1
Remark 1. Payoffs are non-negative and therefore h0 ðfI ÞðaðP1 þ R=rÞ þ ð1 aÞP2 Þ 140 and h0 ðfC ÞðaP1 þ ð1 aÞP2 Þ 1: Lemma 1. @fI ðxC ; a; RÞ=@xC 40 and @fC ðxI ; a; RÞ=@xI 0: The existence of the challenger provokes the incumbent to invest more than it otherwise would on the innovation. Proof. By the implicit function theorem, @fI ¼ @xC
@2 V I ðfI ; xC Þ=@xC @xI . @2 V I ðfI ; xC Þ=@x2I
94
NICK DEW ET AL.
The denominator is negative by the second-order condition. The numerator is 0 ah ðfI ÞðP1 þ R=rÞ þ ð1 aÞh0 ðfI ÞP2 1 h0 ðxC Þ h0 ðfI ÞaðR=rÞh0 ðxC Þ ¼ h0 ðxC Þ ah0 ðfI ÞP1 þ ð1 aÞh0 ðfI ÞP2 1 ¼ h0 ðxC Þh0 ðfI ÞðV I ðfI ; xC Þ aR=rÞ which is positive if V I ðfI ; xC Þ aR=r40: Since V I ðfI ; xC Þ V I ð0; xC Þ because it is a best response value, substituting in V I ð0; xC Þ
aR=r ¼
ahðxC ÞR=r þ R r þ hðxC Þ
aR=r ¼
Rð1 aÞ 40 r þ hðxC Þ
Hence the numerator is positive. By the implicit function theorem @fC ¼ @xI
@2 V C ðxI ; fC Þ=@xI @xC @2 V C ðxI ; fC Þ=@x2C
The denominator is negative by the second-order condition. The numerator is h0 ðxI Þ½ah0 ðfC ÞP1 þ ð1
aÞh0 ðfC ÞP2
1
which is non-positive by Remark 1. & Lemma 2. If R40 and ao1, then fI ðx; R; aÞofC ðx; aÞ8x: Proof. The only difference between Eqs. (1) and (2) is the probabilities of retaining the current revenue stream R. In particular, V I ðxÞ V C ðxÞ ¼ ahðxÞR=r þ ahðxÞR=r þ R: If R ¼ 0 and then fI ðx; 0; aÞ ¼ fC ðx; aÞ8a; x; again using the implicit function theorem, @fI ¼ @R
@V I ðfI ; xÞ=@R@xI @2 V I ðfI ; xÞ=@x2I
Since the denominator is negative and the numerator is 0 ah ðfI Þ a 0 ½r þ hðfI Þ þ hðxC Þ h ðfI Þ½ ðhðfI Þ þ hðxC ÞÞ þ 1 r r ¼ h0 ðfÞð1
aÞ40
we have that fI ðx; R; aÞofI ðx; 0; aÞ ¼ fC ðx; aÞ for all R40, ao1 and all x. & Lemma 3. The incumbent’s best response function is 1. Increasing in a when P1 4P2 or both P1 þ R=r4P2 4P1 and rðP1 P2 Þ þ R=ðP2 P1 Þ4hðxC Þ:
Optimal Inertia: When Organizations Should Fail
95
2. Decreasing in a when P1 þ R=roP2 : The challenger’s best response function is 3. Increasing in a when P2 4P1 : 4. Decreasing in a when P2 oP1 ; and neutral when P2 ¼ P1 :
Proof. Using the implicit function theorem, @fI ¼ @a
@V I ðfI ; xÞ=@a@xI @2 V I ðfI ; xÞ=@x2I
Since the denominator is negative and the numerator is h0 ðfI ÞðP1 þ R=r P2 Þ½r þ hðfI Þ þ hðxC Þ h0 ðfI Þ½hðfI ÞðP1 þ R=r þhðxC ÞR=r ¼ ðh0 ðfI ÞððP1 þ R=r P2 Þðr þ hðxC ÞÞ hðxC ÞR=rÞÞ ¼
ðh0 ðfI ÞððP1 þ R=r
P2 Þr þ ðP1
P2 Þ
P2 ÞhðxC ÞÞÞ
There are three cases: 1. If P1 4P2 ; then the numerator is negative and @fI =@a40: 2. If P1 þ R=roP2 ; the numerator is positive and @fI =@ao0: I 3. If P1 þ R=r4P2 4P1 ; then the numerator is negative ð@f @a 40Þ if rðP1 þ R=r P2 Þ=ðP2 P1 Þ4hðxC Þ: For the challenger’s best response function note that @fC ¼ @a
@V C ðxI ; fC Þ=@a@xC . @2 V C ðxI ; fC Þ=@x2C
The denominator is negative by the second-order condition and the numerator can easily be shown to equal h0 ðfC ÞðP1 P2 Þðr þ hðxI ÞÞo0 and hence @fC =@a40 when the complementary market is bigger than the current market and negative otherwise. & Following Lemma 3, the incumbent’s investment in R&D can increase even if the likelihood of cannibalization (a) is increasing. This will occur when the profits from the new alternative market are larger than the profits from the new technology in the old market (cannibalization). However, following Lemma 2, the incumbent’s reaction is still more tempered than the challenger’s. The incumbent’s investment is decreasing with the probability of cannibalization when the alternative market is inconsequential in relation to the old market after cannibalization ðP1 þ R=roP2 Þ:
96
NICK DEW ET AL.
Proposition 2. If R40, then the incumbent invests less than the challenger. That is, xnI ða; RÞ xnC ða; RÞ Proof. The proof follows the logic of Reinganum (1983), Proposition 2. Suppose, contrary to the proposition, that xnI ða; RÞ xnC ða; RÞ: Then Lemmas 1 and 2 and the definition of a Nash equilibrium imply that xnC ða; RÞ ¼ fC ðxnI ða; RÞ; aÞ fC ðxnC ða; RÞ; aÞ fnI ðxnC ða; RÞ; aÞ ¼ xnI ða; RÞ. But this is a contradiction. Thus xnI ða; RÞoxnC ðaÞ: & The proposition shows that the introduction of market uncertainty does not change Reinganum’s central result that the incumbent rationally invests less than the challenger in the new technology when there is market uncertainty. This result suggests that incumbents rationally invest less than challengers as they are trading off the potential loss of the old market with the potential gains associated with the new market. This is true even if the chance of cannibalizing the new market is small. We speculate (i.e., state without proof) that as the chance of cannibalization becomes small, the incumbent’s investment will approach that of the challenger.
Complementary Assets A key point in the literature is that the incumbent and the challenger might have different cost structures when bringing the product to market. For example, Henderson (1993) finds that challengers’ R&D in the photolithographic industry was more productive than incumbents and hence they were much more successful in exploiting radical technologies than were incumbents. In this section we allow the rewards to R&D to vary between the challenger and the incumbent, thereby allowing the incumbent’s existing assets to be either complementary or detrimental toward the successful commercialization of the radical technology. Our formalization of detrimental assets can be thought of as a formalization of organizational inertia – i.e., an increased cost of shifting to the new business. In the extreme case this cost rises to infinity and it is not possible for the organization to move to the new business model. To capture this idea let the incumbent earn present value profits P1I when it wins the patent race and its profit stream R is not cannibalized and P2I when R is cannibalized. Similarly, let P1C be the present value profits of the challenger when the incumbents profit stream R is not
Optimal Inertia: When Organizations Should Fail
97
cannibalized and P2C be the present value profits of the challenger when the incumbent’s profit stream is cannibalized. Extreme organizational inertia occurs as P2I approaches negative infinity and becomes interesting when this is a case that the challenger would pursue, that is a case where P2C is positive. We state the following lemma without proof: Lemma 4. fI is increasing in P1I and P2I ; fC is increasing in P1C and P2C : In the following proposition we consider the case where there is only technological uncertainty. In particular, we consider the case where a ¼ 0. ~ 2I 0 Lemma 5. There exists P ~ 2C Þ even when R40. fC ðx; a; 0; P
such
that
~ 2I Þ fI ðx; 0; a; 0; P
Proof. If R40, P1I ¼ P1C and a ¼ 0, then fI ðx; R; aÞofC ðx; aÞ8x: The only differences between Eqs. (1) and (2) are the probabilities of retaining the current revenue stream R. In particular, V I ðxÞ V C ðxÞ ¼ hðxÞ=r þ 2hðxÞ½P2I P2C þ R: If R ¼ 0 and P2I ¼ P2C ; then fI ðx; 0; aÞ ¼ fC ðx; aÞ8a; x: From Lemma 4 @fI =@P2 ja¼0 ¼ h0 ðfI Þ½r þ hðxC Þ40: It immediately follows that fI ðx; 0; a; P2I Þ4fC ðx; a; P2C Þ8a; x if P2I 4P2C : Since the best response functions are continuous in R from Assumption 2, there exists a neighborhood N ¼ ð0; Z such that R 2 N; where fI ðx; R; P2I Þ4fC ðx; P2C Þ: & Proposition 3. If R 2 N and P2I 4P2C ; then the incumbent invests more than the challenger ðxnI ðR; P2I ; P2C Þ4xnC ðP2I ; P2C ÞÞ: Proof. The proof follows the logic of the proof of Proposition 2: suppose, contrary to the proposition, that xnI ðR; P2I ; P2C ÞoxnC ðP2I ; P2C Þ: Then Lemmas 4 and 5 and the definition of a Nash equilibrium imply that xnC ðP2I ; P2C Þ ¼ fC ðxnI ðR; P2I ; P2C Þ; P2I Þ fC ðxnC ðP2I ; P2C Þ; P2I Þ fnI ðxnC ðP2I ; P2C Þ; P2I Þ ¼ xnI ðR; P2I ; P2C Þ. But this is a contradiction. Thus xnI ðR; P2I ; P2C Þ4xnC ðP2I ; P2C Þ: & The proposition states that when the returns to radical innovation for the incumbent are sufficiently higher than the returns to the challenger and the present revenue streams are sufficiently modest, then the incumbent will invest more than the challenger despite its current positive revenue stream. This result formalizes the arguments put forward originally by Teece and later by Gans and Stern, which explores the conditions under which a challenger will
98
NICK DEW ET AL.
enter a market. In particular, if the incumbent has complementary assets that are critical to the new radical technology, modeled here as a higher present value of the project, then the incumbent will invest more and creative destruction is less likely (at least by a challenging firm). Note that the incumbent’s investment levels are decreasing as complementary assets become less advantageous. We state this formally in the following proposition: Proposition 4. @fI ðxC ðP2I ; P2C ÞÞ=@P2C o0: Proof. The proof follows immediately from Lemmas 1 and 4. & Proposition 4 implies that as the challenger’s relative advantage grows, the incumbent will invest less and creative destruction at the hands of the challenging firm becomes more likely. Multiple Challengers We now explore a situation where there are multiple challengers. Of particular interest is the case where the probability of a non-cannibalizing market is high as this approximates a case of high market uncertainty. This is the situation that Intel experienced in the early 1980s when it listed 50 possible applications of the microprocessor none of which were the personal computer. To model this, let there be n firms. Without loss of generality, let firm 1 be the incumbent and firms 2; :::; n be the challengers. Note that the model depicted P in the second section does not change except that the term hðxC Þ becomes n1 hi ðxi Þ: All the results still obtain. Therefore, the incumbent will invest more as the number of challengers increases. This proof is the same as the other lemmas above when we notice that the first-order condition for the new problem becomes @V I ðfI ; xC Þ 0 / ah ðfI ÞðP1 þ R=rÞ @xI " # n X 0 hi ðxi Þ þð1 aÞh ðfI ÞP2 1 r þ hðfI Þ þ i¼2
"
h0 ðfI Þ ahðfI ÞðP1 þ R=rÞ þ ð1 þa
n X i¼2
hi ðxi ÞR=r þ R
fI ¼ 0
#
aÞhðfI ÞP2
Optimal Inertia: When Organizations Should Fail
99
By symmetry this can be rewritten as @V I ðfI ; xC Þ 0 / ah ðfI ÞðP1 þ R=rÞ @xI þð1 aÞh0 ðfI ÞP2 1 r þ hðfI Þ þ ðn 1ÞhðxC Þ h0 ðfI Þ ahðfI ÞðP1 þ R=rÞ þ ð1 aÞhðfI ÞP2 þaðn 1ÞhðxC ÞR=r þ R fI ¼ 0
We can then show that =@xI @n @fI =@n ¼ @V ¼ @V =@2 x I
½ah0 ðfI ÞðP1 þR=rÞþð1
aÞh0 ðfI ÞP2 1hðxC Þ h0 ðfI ÞahðxC ÞR=r
40.
Simply stated, the more competition, the more the incumbent invests to deter the competition.
This page intentionally left blank
100
PART II: TOP MANAGEMENT TEAMS
101
102
TOP MANAGEMENT TEAM COMPOSITION AND ORGANIZATIONAL ECOLOGY: A NESTED HIERARCHICAL SELECTION THEORY OF TEAM REPRODUCTION AND ORGANIZATIONAL DIVERSITY Christophe Boone, Filippo C. Wezel and Arjen van Witteloostuijn ABSTRACT The ‘‘upper echelon’’ literature has mainly produced static empirical studies on the impact of top management team composition on organizational outcomes, ignoring the dynamics of industrial demography. Organizational ecology explicitly studied the dynamics of organizational diversity at the population level, however largely ignoring how the entry and exit of executives shapes organizational diversity over time. In this paper, we try to integrate both streams of demography research and develop a multi-level behavioral theory of organizational diversity, linking Ecology and Strategy Advances in Strategic Management, Volume 23, 103–135 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23004-3
103
104
CHRISTOPHE BOONE ET AL.
selection processes at both levels of analysis. The behavioral mechanism connecting the two levels of analysis is the stylized empirical fact that small groups, including top management teams, routinely reproduce their demographic characteristics over time. We argue that, under certain conditions, the potent forces of team homogenization coevolve with those of population-level selection to sustain between-firm diversity.
INTRODUCTION One of the most important research questions in the field of organization sciences is why some organizations are successful, while other organizations linger or fail. Different strands of research tend to predominantly attribute the causes of the fate of organizations either to the external environment in which organizations operate or to the features of their internal functioning. Most organization theories can be put on a continuum ranging from macro to micro with respect to their underlying focus. At the macro side, organizational ecology (OE) has flourished ever since the publication of Hannan and Freeman’s seminal 1977 paper. Given that organizations require reliable routines to survive, OE-researchers have systematically analyzed the vital (i.e., founding, change, and mortality) rates of organizations as the main drivers of change and of diversity within organizational populations. This has cumulated into an impressive body of knowledge and well-established theory fragments, which has recently been summarized by Carroll and Hannan (2000). Heavily relying on the concept of organizational inertia, ecologists have focused on cycles of variation, selection, and retention at the population level, often making abstraction from what happens inside the organization. At the other end of the continuum, more recent streams (re)stress that people should more prominently picture in theories of organizations as they are the ‘‘guts’’ of formal organizations (Stinchcombe, 1997, pp. 17–18; see also, e.g., Pennings, Lee, & van Witteloostuijn, 1998; Haveman, 2000). Especially, Pfeffer (1983) and Hambrick and Mason (1984), departing from a behavioral standpoint, emphasized that organizational actions are historydependent and based on routines. Because individuals enact routines, it becomes central to study managers’ demographic profiles since their characteristics are presumed to be associated with specific psychological dispositions and subsequent strategic choices (Finkelstein & Hambrick, 1996). The argument is that organizations are, to a certain extent, a reflection of
Top Management Team Composition and Organizational Ecology
105
the characteristics of their (upper echelon) members and/or the distribution of their members’ traits. These researchers inspired many scholars to empirically investigate the impact of the demographic composition of (groups in) organizations (especially top management teams) with respect to dimensions such as tenure, gender, ethnic background, and age on a myriad of organizational outcomes. The focus generally is on the impact of the mean and the spread (i.e., diversity) of these characteristics on criterion variables such as turnover, innovation, diversification, and organizational performance (Boone, van Olffen, & van Witteloostuijn, 2005). The potential importance of these contributions should not be underestimated as ‘‘they put the individual back into the predominantly macro-level organization theory’’ (Sørensen, 1999a). However, although many findings of this young research field are very encouraging, a review shows that the results are not very conclusive and in some instances even contradictory (Williams & O’Reilly, 1998; Boone, van Olffen, & van Witteloostuijn, 2005). One reason for this might be that organizations in these studies are treated as isolated entities that do not compete for human capital on input markets. Similarly, the consequences of within-firm organizational demography for between-firm competition and interaction are not explored (but see Sørensen, 1999a; Wezel, Cattani, & Pennings, 2005). To advance organizational demography research we need to develop a dynamic theory in which micro (team) selection processes are integrated into macro (population) ones and vice versa (see also Baum & Singh, 1994a; Haveman & Cohen, 1994; March, 1994; Haveman, 1995). As Haveman (2000) stressed, different phenomena at different levels of analysis can be ‘‘understood as parts of a single, broad evolutionary system. Only when we analyze the whole system will we come to grasp the interdependencies among these phenomena’’ (Haveman, 2000, p. 482). We do so by applying the principles of variation, selection, and retention (Campbell, 1969) at the top management team level of analysis through the lens of OE, with the goal of understanding the evolutionary dynamics of populations of organizations. In building our argument, we will try to bring together several disparate pieces of literature, ranging from organizational behavior (attraction, selection, and attrition processes: Schneider, 1987), via evolutionary theory and sociology (homophily: McPherson, Smith-Lovin, & Cook, 2001; Ruef, Aldrich, & Carter, 2003), to OE (Darwinian population-level selection: Carroll & Hannan, 2000) and market-partitioning theory (multi-form co-evolution: Carroll, 1985), as well as organizational demography and upper-echelon theory (top management team composition: Pfeffer, 1983;
106
CHRISTOPHE BOONE ET AL.
Hambrick, & Mason, 1984). We will also borrow from and build on McPherson’s ecological theory of affiliation in voluntary organizations (McPherson, 1983) and Sørensen’s (1999a) first attempt to develop a dynamic inter-firm approach to top management team composition. So, on the one hand, we contribute to the predominantly ‘‘static’’ top management team literature by elaborating a dynamic theory of team composition and its implications, in interaction with higher-level processes. Moreover, on the other hand, we contribute to organizational ecology by explicitly putting the individual back into macro-level population dynamics (cf. Pennings, Lee, & Witteloostuijn, 1998; Wezel, Cattani, & Pennings, 2005). In particular, we aim at shedding light on the role of managerial turnover for the maintenance of those routines on which selection operates (Baum & Singh, 1994a; Miner, 1994). Ultimately, therefore, this paper deals with an old classical tension in the social sciences between individual agency and the dominating pressure of social aggregates (Haveman, 2000).
MICRO-LEVEL HOMOSOCIAL REPRODUCTION Drive Toward Homogeneity A pervasive fact in social life is that social groups (including top management teams and organizations) are not random samples of people (McPherson, Popielarz, & Drobnic, 1992). Instead, people are systematically sorted into groups whose members have similar sociodemographic characteristics. Blau (1977) convincingly argued that demographic characteristics influence social interaction: social interaction is more likely to occur between people who are similar with respect to demographic features (McPherson et al., 1992, 2001). Indeed, evidence shows that ‘‘distances along sociodemographic dimensions translate into probability of contact between individuals for almost all kinds of messages passing through the system, whether the messages are money, sociation, attitudes, group formation, or the like’’ (McPherson et al., 1992, p. 155). This self-reinforcing relationship between similarity and connection is also known as the ‘homophily principle’. With respect to group formation, research shows that (groups in) organizations differentiate by carving out niches in social space. The result is that members of groups are, on average, more similar to each other than to members of other groups. McPherson and colleagues (McPherson, 1983; McPherson & Smith-Lovin, 1987; McPherson et al., 1992; McPherson & Rotolo, 1996), building on the seminal work of Blau (1977), systematically
Top Management Team Composition and Organizational Ecology
107
studied a large variety of voluntary organizations, ranging from sport clubs to churches in the U.S. They found that in the course of competition for members, these organizations specialize in specific local regions of sociodemographic space – the so-called Blau space (McPherson et al., 1992). When mapping voluntary organizations onto the demographic dimensions of the member’s occupation and education level, they observe that the means in both dimensions differ between organizations and that the withinfirm standard deviations are much smaller than a random sample of individuals would produce. Similar compositional differences between organizations in the for-profit sector were reported in the organizational behavior literature. Interestingly, here researchers did not focus on demographics, but rather on aspects of member personality. For instance, Schneider, Smith, Taylor, and Fleenor (1998) used the Myers–Briggs type indicator to assess the personality of almost 13,000 managers from 142 organizations, representing a broad crosssample of U.S. industries. Their key finding is that a substantial part of personality variation can be explained by organizational membership. In addition, nested models of organizations within industries revealed remarkable regularities, demonstrating that sorting does not only occur across populations (as in the studies of McPherson and colleagues), but also within the same population. This underscores that ‘‘homophily is a powerful force of group homogeneity’’ (McPherson et al., 1992, p. 156). The same homophily principle also applies to small groups and management teams. Ruef, Aldrich, and Carter (2003) analyzed a unique representative cross-sectional sample of 816 organizational founding teams from the U.S. population, and found that homophily and network constraints based on strong ties were the most important forces driving founding team composition. This appeared to be the case not only for ascriptive characteristics such as gender and ethnicity, but also for achieved characteristics. Ruef et al. (2003, p. 217) conclude that ‘‘even in a situation where we might reasonably expect stringent economic rationality to prevail – and thus lead to choice based on the functional diversification of achieved characteristics – we find that team composition is driven by similarity not differences.’’ The above examples make clear that important cross-sectional differences in the demographic composition of groups can be observed, and that homophily and group homogeneity are two sides of the same coin. However, they do not fully explain how these regularities unfold. In order to be able to fully appreciate the potential consequences of homophily and group composition in an organizational context, one therefore also needs to understand where demographic homogeneity (or, for that matter, diversity)
108
CHRISTOPHE BOONE ET AL.
comes from in the first place (Lawrence, 1997; Pfeffer, 1997; Hinds, Carley, Krackhardt, & Wholey, 2000), as well as the determinants of the evolution of demographic composition (Ruef et al., 2003; Boone, van Olffen, van Witteloostuijn, & De Brabander, 2004). Dynamic Models of Group Composition Two different theoretical accounts have been offered in the past to explain the dynamics of group composition: McPherson’s ecology of affiliation (McPherson, 1983; McPherson et al., 1992) and Schneider’s attraction–selection–attrition (ASA) theory (Schneider, 1987). Both theories propose a Darwinian mechanism of systematic variation, selection, and retention of members in groups. Moreover, both theories agree that groups have the tendency to reproduce themselves by the selective recruitment of similar people in the group and by facilitating the turnover of dissimilar people, a directional selection process labeled ‘‘homosocial reproduction’’ by Kanter (1977). These two theories offer seemingly different but complementary explanations about the unfolding of group homophily. McPherson’s (macro-)sociological theory, building on Blau’s Inequality and Heterogeneity (1977), stresses the importance of network ties associated with the position of individuals in sociodemographic space. In this view, social structure, via its impact on network ties, drives the selective entry and exit of members into and out of groups. As people tend to develop network ties with other people sharing similar sociodemographic characteristics, people joining to form groups are relatively similar. This similarity is perpetuated due to conservative, selective recruitment of new members (McPherson et al., 1992; Popielarz & McPherson, 1995). In a similar vein, Feld (1982) argued that the social homogeneity of organizational foci strongly enhances the formation of homophilous network ties, and found that these foci produced ties that were twice as homogeneous as would be expected by chance. So, selective recruitment implies selection via social contact.1 Selective recruitment is reinforced by the principle that members who are atypical of the group will leave the group first (the niche-edge hypothesis), especially when groups are subject to competition from other groups (Popielarz & McPherson, 1995). The putative reason is that demographic dissimilarity from group members acts as a centrifugal force because homophily implies that atypical members have more external ties to nonmembers and are less closely tied to fellow group members (Popielarz & McPherson, 1995). Empirical research on voluntary organizations indeed suggests (1) that entry and exit of group members depend upon the number
Top Management Team Composition and Organizational Ecology
109
and strength of social network ties that connect group members to each other and to non-members (McPherson et al., 1992), and (2) that atypical members will leave the group first (Popielarz & McPherson, 1995). McPherson et al. (1992) claim that homophily mainly follows from social structure, and not from human agency (cf. Popielarz & McPherson, 1995). Group formation is determined by general constraints in the social network with respect to logically possible choices, rather than by ‘‘individual utilities or imputed production functions’’ that guide our choices to join and stay in groups (McPherson & Ranger-Moore, 1991, p. 38). That is, an individual’s position in the social structure determines her or his opportunity set. The ASA theory of Schneider (1987) offers a complementary view. ASA theory summarizes many different strands of research in the field of organizational behavior (see also Schneider, Goldstein, & Smith, 1995), stressing the importance of human agency and choice in producing homophily.2 Although the original model focused on ‘‘soft’’ people characteristics such as attitudes and personality, it also applies to demographics (Westphal & Zajac, 1995; Zajac & Westphal, 1996; Boone, van Olffen, van Witteloostuijn, & De Brabander, 2004).3 The ASA model outlines the forces operating to restrict organizational diversity in members’ characteristics, producing so-called trait homogenization (Schaubroeck, Ganster, & Jones, 1998). That is, organizations move toward member homogeneity because individuals are attracted to, selected by, and stay with organizations that suit their personality characteristics (Schaubroeck et al., 1998). ASA theory provides a behavioral rather than a structural (i.e., network ties) account of the dynamics of homosocial team reproduction. Why would people prefer to associate or to work with similar ‘alter egos’? Several related reasons can be offered (see also Westhpal & Zajac, 1995; Zajac & Westphal, 1996; Boone, van Olffen, van Witteloostuijn, & De Brabander, 2004). First, there is ample evidence in social psychology that similarity on a salient dimension enhances (dyadic) interpersonal attraction (for a review, see Huston & Levinger, 1978). Although many underlying mechanisms have been proposed, theory and evidence suggest that this similarity – attraction response is likely to be deeply rooted in human beings as it is directly reinforced by positive affectivity (Byrne, 1971; Clore & Byrne, 1974). Second, self-categorization theory posits that people derive selfesteem and self-identity from perceived group membership. As demographic similarity provides a salient basis for group membership, people may seek to construct or maintain homogeneous groups in order to sustain or enhance their self-esteem and identity (Westphal & Zajac, 1995). Third, Hogg, and Mullin (1999) argue that reducing uncertainty is a primary individual mo-
110
CHRISTOPHE BOONE ET AL.
tivator or ‘‘fundamental need’’ guiding behavior. Given that similarity enhances the (perceived) predictability of the behavior of others, the psychological need for stability strongly motivates people to reduce uncertainty by forming homogeneous groups (see also Hinds et al., 2000). A final set of reasons follow from the fact that groups in most cases (e.g., in organizations) compete with other groups for resources or are nested in higher-order groupings (Baum & Singh, 1994a; Campbell, 1994). It is well established that inter-group competition alters behavior of members in important ways (an insight dating back to, at least, the Robber’s cave experiments of Sherif, Harvey, White, Hood, & Sherif, 1961; see also LeVine & Campbell, 1972). For instance, it is a stylized fact that people are more likely to cooperate in a social dilemma when it is embedded in the context of intergroup conflict (Bornstein & Ben-Yossef, 1994; Bornstein, 2003). The reason for this is that from the standpoint of the individual such behavior increases the survival chances of the group. However, within-group cooperation is only sustainable when people trust each other and/or deviance from reciprocity can easily be monitored and sanctioned (Campbell, 1994; van Witteloostuijn, 2003). It is likely that group homogeneity facilitates trust and reciprocal altruism in face-to-face groups (Ruef et al., 2003; Boone, van Olffen, van Witteloostuijn, & De Brabander, 2004) – clique selfishness in Campbell’s (1994) words. As he (1994, p. 28) puts it: ‘‘All group uniformities on trait – specifically neutral features – would be useful signs in such a reciprocal altruistic pact.’’ Closely related to this, group homogeneity might also enhance group survival because of sociopolitical reasons. That is, homogeneity is likely to increase a group’s power to control competition between groups in nested settings. For instance, top managers might prefer to hire and promote people in the top management team who are similar – e.g., having the same functional background and sharing the same strategic preferences – to perpetuate and institutionalize their managerial power. At the same time, similarity facilitates communication within a team, and diminishes the likelihood of conflict and within-team power struggles (Pfeffer, 1983; Boone, van Olffen, & van Witteloostuijn, 1998). At the organization level, Beckman, Haunschild, and Phillips (2004) argue that when uncertainty is high, organizations will strive for homogeneity, reducing uncertainty through interactions with similar others. Specifically, organizations, as a threat-rigidity response (Staw, Sandelands, & Dutton, 1981), will seek to establish stability and trust by increasing their commitment toward existing partners instead of seeking new (uncertain) relationships. Their analyses reveal that ‘‘market uncertainty leads firms to reinforce their existing net-
Top Management Team Composition and Organizational Ecology
111
works, and firm-specific uncertainty y leads firms to reduce their broadening networks’’ (Beckman et al., 2004, p. 273). This type of findings show that uncertainty or, in more general terms, pressure on the group increases the tendency of the social entity to close its ranks as a threat-rigidity and uncertainty-reducing response. The fact that exogenous forces trigger this process suggests to us that the role of choice, albeit unconscious, should not be underestimated.4 We expect that especially in top management teams choice-driven homophily might be important, as in most cases the market for managerial talent is carefully screened before candidates are selected into the team (Schneider et al., 1998; Sørensen, 1999b). Note that whatever the source of homophily, structural or choice-based, the end result is the same: groups tend to become more homogeneous over time by selectively replacing ‘‘distant’’ members with people who are similar to themselves (see also Lawrence, 1997). Empirical Evidence and Theorem on Firm-level Homogenization A very limited number of top management team studies focus on team composition from a dynamic standpoint, and all reveal, this general tendency toward homophily: top executive management teams tend to selectively hire and fire to the effect that their own demographic characteristics are strengthened, so promoting their homosocial reproduction. Westphal and Zajac (1995), who apply these ideas to the board of director selection process, hypothesized that the extent to which incumbent CEOs can realize their preference for demographically similar new directors depends on the relative power of the CEO vis-a`-vis the board of directors. They test their argument on a sample of 413 Fortune/Forbes 500 companies from 1986 to 1991. Consistent with their arguments, they found that ‘‘(1) when incumbent CEOs are more powerful than their boards of directors, new directors are likely to be demographically similar to the firm’s CEO; (2) when boards are more powerful than their CEOs, new directors resemble the existing board.’’ Indeed, this pair of findings provide evidence for homosocial reproduction, where power determines which group’s demographics is being reproduced – the executive or the non-executive team. These ideas of Westphal and Zajac can also be applied in the context of the selection of new executive managers into top management teams. Indeed, Jackson et al. (1991), studying 93 top management teams in bank holding companies over a four-year period, found that reliance on internal recruitment as a mean for filling team vacancies resulted in greater subsequent team homogeneity. With respect to the turnover process, the au-
112
CHRISTOPHE BOONE ET AL.
thors demonstrated that (1) teams, on average, are relatively homogeneous, (2) heterogeneity is a relatively strong predictor of team turnover rates, and (3) team members whose personal attributes are dissimilar to their team mates are more likely to leave the team than team members with similar attributes. Finally, Boone, van Olffen, van Witteloostuijn, and De Brabander (2004) systematically analyzed the processes of selective entry into and exit from top management teams of the five largest newspaper-publishing companies in the Netherlands in the period from 1970 to 1994. They argued that top management teams, when having the power to do so, tend to hire likes (in terms of demographics), while at the same time fire or ‘‘relieve’’ unlikes, making the team gravitate toward homogeneity. The authors also suggested that rational-economic forces might operate as a countervailing force at the organizational level to undermine these ASA cycles. Specifically, they theorized that the cycle of homosocial reproduction cannot be sustained when top management teams face a compelling need for team composition diversity caused by conditions such as poor organizational performance, high corporate diversification, and tough market competition. Interestingly, many of their expectations were not supported. Relating to entry, they report that poor performance and high diversification causes teams to select likes, which is exactly opposite to what they expected. In addition, although more powerful teams do tend to select likes, their tendency to do so is even stronger when competitive intensity increases. Concerning exit, they found that poor organizational performance increases the overall likelihood of executive exit, and that dissimilar managers tend to leave first. In addition, the likelihood of dissimilar managers’ leaving appeared to be greater when firm diversification is high, which was again not as expected. Taking together these findings, they conclude that homosocial reproduction does occur, and particularly so when the organizations face conditions that at first glance require more team diversity. Apparently, top management teams tend to close ranks when environmental pressure or uncertainty increases. At the inter-firm level, these findings are remarkably similar to the conclusions of the network tie research of Beckman et al. (2004); (see also Podolny, 1994). At the individual level, these results are in line with behavioral research into the uncertainty-reducing effect of routinized behaviors (Heiner, 1983; Egidi, 1996). Uncertainty, broadly defined, apparently induces a threat-rigidity response with respect to demographic team composition, too (Boone, van Olffen, van Witteloostuijn, & De Brabander, 2004). The consistency of this observation in very different settings suggests that this reaction is strongly embedded in
Top Management Team Composition and Organizational Ecology
113
human and, by extension, organizational behavior. So, in an important respect – the composition of their upper echelon – firms reveal a clear tendency toward homogenization. By way of summary, we therefore propose Theorem 1. Particularly – but not exclusively – in the face of uncertainty, top management teams tend to selectively hire and fire to the effect that their own demographic characteristics are strengthened (homosocial reproduction), leading to team homogeneity.
HOMOSOCIAL REPRODUCTION AND DIFFERENTIATION Team Homogenization and Competitive Differentiation Do homosocial reproduction and ASA processes serve an adaptive purpose or do they undermine organizational performance? Schneider et al. (1995) speculate that homogeneity might be beneficial for organizational performance in the short run, for young organizations in particular, because it facilitates coordination, communication, and cooperation. However, survival in the long run and for old organizations may be undermined by the atrophying consequences of homogeneity. That is, heterogeneity ‘‘should provide the kinds of alternative perspectives and conflicts that stimulate accurate sensing of the environment and the concomitant changes required to adapt and cope with changes that might not otherwise be perceived’’ (Schneider et al., 1995, p. 766). However, the upper echelon research reviewed above remains inconclusive on the impact of top management team heterogeneity (or homogeneity) on performance (Williams & O’Reilly, 1998). An important reason of these inconclusive findings is that top management team research focuses on short-run outcome variables (e.g., Return on Equity (ROE)), failing to include long-run organizational performance measures such as growth, survival, and innovation. Incorporating such measures is, however, necessary to unravel the complicated relationship between team diversity and organizational performance (Boone, van Olffen, van Witteloostuijn, & De Brabander, 2004). Moreover, a full appreciation of the survival consequences of homosocial reproduction – and, for that matter, of top management diversity – can be reached only by moving beyond the atomistic perspective of the theory and conceiving the organization as part of an environment within which it competes for managerial talents and resources in the output market (see also Sørensen, 1999a; Pennings & Wezel, 2005).
114
CHRISTOPHE BOONE ET AL.
In more general terms, any evolutionary theory of organizational adaptation requires the study of the interplay between selection processes at different, nested levels of analysis (Baum & Singh, 1994a). To assess the adaptiveness of top management team selection processes (Wiersema & Bantel, 1993) one needs to deal with the fact that units of adaptation are located within ecologies of other units: ‘‘Units of adaptations are nested, so that some adapting units (e.g., individuals) are integral parts of other adapting units (e.g., organizations). The structure of relations among them arises from an interaction among the various nested units responding to a shifting environment and their own internal dynamics. These features of organization considerably complicate any multilevel evolutionary story’’ (March, 1994, p. 43). Such nested selection processes imply that what might be local adaptation at the team level, can be dysfunctional at the organizational level (Campbell, 1994; Baum & McKelvey, 1999; Miller, 1999). Meyer (1994) is very explicit on this, and argues that uncertainty and bounded rationality make organizational adaptation to the external environment almost impossible. Most people do not adapt to environments, but to organizational cues such as goals and sub-goals. This local adaptation is especially triggered when organizations are faced with bad news (e.g., low performance) or external threats. According to Meyer (1994, p. 110), these strategies can be very adaptive for individual people in order to preserve a modicum of certainty and power over their own lives: ‘‘internal adaptation (the kind used by people) may be independent of external adaptation (the kind made by organizations), and sometimes inimical to it.’’ In a similar vein, homosocial team reproduction processes can be considered as local team-level adaptations to external uncertainty and pressure. Whether such top management team-level adaptations are functional or dysfunctional will depend on the consequences of these processes for the higher-level ecology in the population of competing organizations to which these teams belong.5 Here, our starting point is Campbell’s (1994) claim that internal group homogeneity and inter-group variability set the stage for higher-level selection. Specifically, migration into and out of top management teams is such that selection and retention processes reduce within team-variation, implying that teams will, on average, become more homogeneous over time. As organizational features, such as routines and strategies, can partly be considered to be a reflection of top management team composition (Hambrick & Mason, 1984), these homosocial reproduction processes have an impact on the higher-level selection entities – i.e., organizations that compete for growth and survival in the marketplace. The upper echelon research tradition has indeed demonstrated that organizational routines and strategies
Top Management Team Composition and Organizational Ecology
115
do not exist independently of the characteristics of individual human beings (Miner, 1994). The unfolding of homosocial reproduction along specific demographic characteristics and experiences of executives implies that capabilities, decisions, and policies get reinforced over time. That is, by hiring likes and relieving unlikes routines and strategies are perpetuated, which in turn reinforces the tendency toward homosocial team reproduction. As the degree of heterogeneity in demographic characteristics amounts to a ‘‘proxy for cognitive heterogeneity, representing innovativeness, problem-solving abilities, creativity, diversity of information sources and perspectives, openness to change and willingness to challenge and be challenged’’ (Finkelstein & Hambrick, 1996, p. 125), the unfolding of homosocial processes will inevitably reduce the spectrum of opportunities available to the organization. The result is that organizations gradually carve out a specialized niche in resource space. As these processes apply to all organizations within a given population, homosocial team reproduction goes hand in hand with betweenfirm differentiation in top management team composition, reducing niche overlap among organizations. Paradoxically, team homogenization processes may well trigger and sustain population-level organizational diversity (Boyd & Richerson, 1985; Campbell, 1994).6,7 Need for Differentiation and Organizational Diversity Before connecting micro to macro logic more precisely, we must first briefly review the latter. A central question in OE relates to the explanation of organizational diversity: where does organizational diversity come from, and what explains its evolution? In the past three decades, a series of theory fragments in OE has sought – and still does so – to answer this key question (Hannan, 2005). In the context of the current paper, we briefly discuss three of such theory fragments: the localized competition argument (Hannan & Freeman, 1977), the resource-partitioning model (Carroll, 1985), and nicheoverlap logic (Baum & Singh, 1994b, 1994c). What this set of theory fragments have in common, is that they start from the ecological concept of the niche. Organizations address with their offer certain client or customer tastes, indicated by points in the n-dimensional resource space. First, localized competition was already introduced in Hannan and Freeman’s (1977) classic contribution. The key argument is that an organization is particularly subject to competition from rivals that are located on close distance in resource space. That is, the closer organization i and j are in resource space, the more intense their competition will be. Subsequent work
116
CHRISTOPHE BOONE ET AL.
further specified the underlying distance notion. For example, size-localized competition (Hannan, Ranger Moore, & Banaszak-Holl, 1990; Amburgey, Dacin, & Kelly, 1994) argues that competition is partitioned along the size gradient: i.e., similarly sized organizations compete most intensely. The reason for this is that the kind of resources firms’ use or the type of strategy they pursue is closely associated with their size. More generally, localized competition relates the intensity of rivalry to distance in a multidimensional space, introducing such aspects as capacity, geography, price, product, and technology next to mere size (Baum & Mezias, 1992; Baum & Haveman, 1997; Dobrev, Kim, & Carroll, 2002). Second, in 1985, Glenn Carroll introduced a resource-partitioning argument to explain a paradoxical phenomenon observed in many industrial populations: increasing firm concentration can open the way for entry of new organizations at the market peripheries. Resource-partitioning theory distinguishes two organizational types according to their niche spans. On the one hand, generalist organizations are characterized by a broad fundamental niche: their products attract people from very different taste groups. On the other hand, specialist organizations have a narrow niche: their offer focuses on specific tastes. The generalists’ advantage derives from their broad and rich potential customer base. But because of their broad appeal, their offer cannot be as precisely tuned at the customers’ wishes as that of specialists. This is in contrast with the sharp stance of specialist organizations, which can exploit their niche with high effectiveness, attracting a high percentage of the clients from their narrower customer base. The resourcepartitioning model describes how selection processes structure organizational populations according to their niche spans when resources are unevenly distributed in the environment, forming a market center abundant in demand vis-a`-vis a resource-scarce market periphery (Carroll, 1985; Carroll & Hannan, 2000). Evidence abounds that indeed generalist concentration is positively associated with specialist viability (for an overview, see Carroll & Swaminathan, 2000). Third, niche overlap is argued to have two potentially opposing effects on organizational performance (Baum & Haveman, 1997; Boone, Wezel, & van Witteloostuijn, 2005). On the negative side, more niche overlap implies more intense crowding competition for similar resources; on the positive side, more niche overlap may produce mutualistic agglomeration-type of benefits. On the one hand, in a series of studies, Baum and colleagues explore the crowding effect of niche overlap (and its complement: the mutualistic impact of non-overlap). In their study of Torontonian day care centers, Baum and Singh (1994b) find support for their hypothesis that niche-overlap density
Top Management Team Composition and Organizational Ecology
117
decreases entry in the focal niche, whereas non-overlap density increases founding rates. The former effect is argued to relate to crowding and competition, while the latter is explained by the potential benefits of mutualism when entrepreneurs enter niches that border other viable niches with low overlap. Similarly, Baum and Singh (1994c) reveal that niche-overlap density and non-overlap density are positively and negatively associated with mortality rates, respectively. In line with this finding, Baum and Singh (1996) report that differentiation – i.e., moving to less competitive niches – increases the day care centers’ survival chances. Finally, Baum and Oliver (1996) report evidence in the context of founding rates in Torontonian day care centers that mutualistic non-overlap is enhanced in the presence of crowding. On the other hand, Baum and Haveman (1997) and Boone, Wezel, & van Witteloostuijn (2005) add to this logic that niche overlap might have a mutualistic impact as well, pointing to a set of spillover effects that may benefit organizations that cluster together. Whatever their differences, the alternative theories of market partitioning share a common key assumption: at the demand side of the market, the variety of tastes is such that a need for differentiation arises. That is, customers in niche i prefer a product offer that is quite different from their counterparts in niche j. The further apart niches i and j are in resource space, the larger is this taste difference. Take, again, Carroll’s (1985) resourcepartitioning model, to start with. Resource-partitioning processes emerge if a number of necessary conditions are met (Boone, Carroll, & van Witteloostuijn, 2002, 2004; Boone & van Witteloostuijn, 2004). A critical one is that the resource space must reveal sufficient heterogeneity, with a resourceabundant center and a resource-poor periphery. In a similar vein, other theories of market partitioning need such a resource space heterogeneity assumption. For instance, niche-packing theory (Pe´li & Nooteboom, 1999) assumes a flat resource space, implying that resources are distributed over a (wide) variety of taste niches. Without taste variety, market partitioning will not emerge. For instance, if the resource space features resource-rich homogeneity only, then specialist organizations cannot operate viably. After all, in such a market, there is no peripheral demand for specific offerings, as all customers prefer the same middle-of-the-road variety. For instance, demand for special types of petrol or salt is very limited: for by far the majority of customers, only price matters. Then, the demand side’s low variety of tastes is reflected in supply side homogeneity. Market partitioning will not emerge, because the demand side’s preferences reflect a need for homogeneity, rather than a desire for heterogeneity.
118
CHRISTOPHE BOONE ET AL.
In all, the above logic gives our second theorem. Theorem 2. Under the condition of sufficient demand-side heterogeneity (i.e., taste variety), firms tend to differentiate themselves away from their competitors to the extent that they spread across the resource space’s niches, leading to industry heterogeneity.
A MULTI-LEVEL THEORY OF REPRODUCTION AND DIVERSITY Conditions and Propositions We are now ready to illustrate how, under certain conditions, macro-level partitioning may unfold from micro-level processes of homosocial reproduction, linking Theorem 1 to Theorem 2. Before turning to more specific propositions, it is useful to make explicit the major assumptions underlying the theory. That is, how this micro–macro interaction evolves over time, very much depends upon the underlying conditions at the supply side (potential managers) and at the demand side (potential customers). A minimum level of resource diversity is needed at both the input and output side of organizations to produce the between-firm differentiation effects described above. It is clear that if the distribution of demographic characteristics of top managers who may enter into a population has low variance, then the compositional differences between teams in the industry will be low as well, irrespective of the potency of homosocial reproduction. As a result, team reproduction can then not be a source of market-level partitioning.8 The same reasoning applies to output markets. A minimum level of resource heterogeneity with respect to customer tastes is required for organizations to be able to differentiate and occupy isolated niches (for an overview of the impact of different resource environments, see van Witteloostuijn & Boone, 2006).9 So, the main necessary conditions are: Condition 1. Supply side heterogeneity with respect to demographic characteristics of top managers is sufficiently high. Condition 2. Demand-side heterogeneity with respect to customer tastes is sufficiently high. By way of illustration, several more specific propositions can be derived from our theory. Without losing generality, we focus on two firms only (firm i and j), for the sake of parsimony. The model distinguishes three levels of
Top Management Team Composition and Organizational Ecology
119
analysis – team, organization, and market (i.e., population), – linking micro-level processes of team adaptation with ecological selection at the market level. At the most general level, we predict that markets over time will show features of market partitioning (e.g., localized competition, monopolistic competition, niche packing, or resource partitioning) the more top managers are sorted into teams based on the principle of homophily. That is, every organization eventually specializes to serve niches with minimal overlap with the niches of other organizations. In economics terms, in the extreme, every firm may then act as a local monopolist in the niche of the market where its product offer comes closest to the taste of local customers – this is the classic case of monopolistic competition (Hotelling, 1929). The reasoning is as follows. If the ASA processes within the firms’ upper echelons lead to different team characteristics, organizations will carve out different niches, resulting in strategic differentiation. If sets of organizations, however, happen to reproduce around similar top manager characteristics, they will carve out similar niches, resulting in niche overlap. Niche overlap will put pressure on organizations to differentiate, especially when the carrying capacity of the niche is low. Fig. 1 illustrates how the process unfolds. To summarize, our theory suggests that market partitioning may result from top management team homogenization (given that teams gravitate toward different team demographic compositions). Because differentiation can boost organizational performance (Hotelling, 1929; Baum & Mezias, 1992; Boone, Carroll, & van Witteloostuijn, 2002, 2004), it reinforces the process of homosocial reproduction at the team level (a feedback loop not shown in Fig. 1, for the sake of parsimony). That is, in order to avoid the downsides of crowding, spreading across the resource space’s niche structure stimulates organizational performance by reducing niche overlap, on average. Similarly, competition among organizations – i.e., niche overlap – directly spurs team homogenization. Indeed, McPherson and colleagues showed that competitive pressures from other organizations for members sharpen and focus the compositional features of the group (the so-called social niche), resulting in organizational homogeneity (McPherson, 1983; McPherson & Smith-Lovin, 1987) – a result that is in line with the findings from research on inter-firm, team and individual behavior (e.g., Heiner, 1983; Beckman et al., 2004; Boone, van Olffen, van Witteloostuijn, & De Brabander, 2004). The end result of such positive feedback cycles is that there will be a fairly tight match between top management team composition and organizational niche position in the long run. In all, this suggests
120
Time Team i
Team homogenization
Niche carving
Firm i
Competition as a function of degree of niche overlap
Market
Fig. 1.
Niche carving
Strategic differentiation
Organizational performance
Market partitioning
Organizational performance
Team homogenization
A Process Model of Homosocial Reproduction, Competition and Long-Run Organizational Performance.
CHRISTOPHE BOONE ET AL.
Firm j
Team j
Strategic differentiation
Top Management Team Composition and Organizational Ecology
121
Proposition 1. Provided that Conditions 1 and 2 hold, the higher the potency of homosocial reproduction at the top management team level, (a) the higher the extent of market partitioning and (b) the lower the level of niche overlap. Given our prediction that uncertainty increases the tendency toward homosocial reproduction, we expect that the process of market partitioning will especially materialize in periods of high uncertainty. Podolny (1994) also suggested that uncertainty reinforces market partitioning. In a sense, one could argue that the team-level process of homosocial reproduction is adaptive at the population level. This is because the threat-rigidity response at the team level eventually decreases niche overlap and competition at the market level, reducing the uncertainty that triggered the process in the first place. Note that a tightly packed resource space reduces uncertainty in another way, too, as it hampers entry (Pe´li & Nooteboom, 1999) and stabilizes organizational density. Moreover, the model predicts that homosocial reproduction sets in motion a process of differentiation between pairs of firms, both with respect to the compositional characteristics of their top management teams, as well as the strategic niche positions these organizations occupy. Different organizations will gravitate toward different specific top management team demographic characteristics. Given that average compositional differences imply variance in capabilities and preferences, these firms will also develop different routines and strategies (Finkelstein & Hambrick, 1996; Sørensen, 1999a; Pennings & Wezel, 2005). Therefore, we expect for each dyad of organizations, and given that the carving of specialized niches will be more forcefully present when teams are relatively homogeneous: Proposition 2. (a) A focal firm’s distance from a competitor with respect to its top management team’s average demographic composition is positively related to its strategic distance vis-a`-vis this competitor, (b) especially under conditions of high uncertainty. From the literature on crowding and niche overlap cited above follows that differentiation decreases competition, and therefore increases organizational performance. As Hotelling (1929) explained, for example, positive price premia emerge if products are different, since then each firm can operate as a local monopolist in the niche of the market where its product offer comes closest to the taste of local customers. Similar predictions have been made in the strategic management literature. Gimeno and Woo (1996), for instance, find in their sample of more than 3000 city-pair markets in the U.S. airline
122
CHRISTOPHE BOONE ET AL.
industry that strategic similarity among airlines increases the intensity of rivalry. Specifically, the yield to an airline, defined as the average price charged by a firm to passengers in a city-pair market divided by the distance of the market, decreases when the average strategic similarity to competitors increases.10 Moreover, Sørenson (1999a) found that commercial television stations grow faster the more the average tenure of the top management team differs from that of its competitors. Taken together, this suggests that the positive impact of differentiation with respect to team composition on organizational performance is mediated by strategic dissimilarity resulting from demographic differences. This argument leads to Proposition 3. There is a positive impact of a focal firm’s (a) average distance from competitors with respect to its top management team’s mean demographic composition on (long-run) organizational performance,11 where (b) this relationship is mediated by the average strategic distance of the focal firm from its competitors. Finally, the model also helps to unravel the performance implications of top management team demographic diversity. Most authors have suggested and presented evidence that top management team homogeneity facilitates (or, at least, does not hamper) performance in the short run (Schneider, 1987; Boone, Olffen, & van Witteloostuijn, 1998; Boone, van Olffen, van Witteloostuijn, & De Brabander, 2004; McPherson & Smith-Lovin, 1987). As mentioned earlier, research is not conclusive with respect to the long-run implications of top management team diversity. Our ecologically informed model suggests that the (long-run) performance consequences of top team homogeneity depend on market-partitioning processes and the position of other firms in product space. On a general level, we expect that homogeneity will only be positively associated with (long-run) organizational performance, particularly the likelihood of survival, in partitioned markets with low niche overlap among competitors. That is, an individual organization’s likelihood of survival depends critically upon its position in resource space. More specifically, a focal organization with a homogeneous top team will outperform its competitors when it occupies isolated positions in product market space and, mutatis mutandis, in the distribution of managerial demographic characteristics. Conversely, if the organization happens to be located in an overly crowded niche, long-run organizational performance will be harmed. Again, homosocial reproduction will fuel this process. If an organization is located in an unfavorable niche, short-run performance will suffer. As a threat-rigidity and uncertainty-reducing response, ranks will be closed by hiring ‘clones’ and ‘relieving’ atypical executives (Boone, van
Top Management Team Composition and Organizational Ecology
123
Olffen, van Witteloostuijn, & De Brabander, 2004). This implies that the organization further digs its own grave by investing in routines and strategies that fit with the ‘wrong’ niche. Overall, this logic results in Proposition 4. Top management team homogeneity increases (long-run) organizational performance, when (a) the focal firm’s average distance from competitors with respect to its top management team’s mean demographic composition is large,12 where this relationship is mediated by the average strategic distance of the focal firm from its competitors.
APPRAISAL AND CONCLUSION Two important streams of demographic research have developed independently over the last 25 years or so: within-firm organizational demography (Pfeffer, 1983), with an emphasis on (top management) team composition (Hambrick & Mason, 1984), and population-level demography or organizational ecology (Hannan & Freeman, 1977), studying the dynamics of diversity among organizations. Both research traditions strictly focused on the selection processes operating within the boundaries of the chosen level of analysis: micro (individual and team) versus macro (organization and population). We argued, however, that in order to advance the field of top management team research (and organizational ecology, for that matter) we need to build more comprehensive evolutionary models of organizational adaptation explicating the interplay between selection processes at different, nested levels of analysis (Baum & Singh, 1994a). In this paper, we made a first attempt by theorizing about how the within-team variety-reducing behavioral mechanism of homophily shapes between-firm competitive outcomes and organizational diversity, and vice versa. Six final comments and limitations are worth mentioning. First, discovering the general mechanisms behind the origin of organizational diversity has attracted the attention of many scholars for more than a century. Durkheim (1893/1933, p. 266) already speculated on how organizational diversity is driven by endogenous forces as follows: ‘‘if work becomes divided more as societies become more voluminous and denser, it is not because external circumstances are more varied, but because the struggle for existence is more acute.’’ In the present paper, we build on this insight by explicating one such endogenous mechanism, arguing that team homogenization processes, paradoxically, drive population-level organizational diversity. In other words, homosocial reproduction magnifies social
124
CHRISTOPHE BOONE ET AL.
differences, rather than mitigating them (see also Popielarz & McPherson, 1995, p. 699). We acknowledge that the behavioral theory of organizational diversity we presented here is extremely general and, as a result, that the predictive power in any specific setting will be relatively low, on average. To be able to develop more specific hypotheses, additional conditions must be introduced, reflecting the specificities of the case at hand. Above, we illustrated this for the case of market partitioning, imposing two additional conditions upon the argument as to sufficient heterogeneity at the demand or output side (i.e., potential customers) and the supply or input side (i.e., potential managers) of the market. Specifically, we think that the field of organization theory and strategy badly needs general theories that unify the increasingly disparate pieces of literature and provide a parsimonious baseline logic to understand complicated phenomena based on firmly grounded behavioral principles. In this respect, we follow McPherson and Ranger-Moore (1991, p. 37), who defend their approach in modeling the impressive diversity of forms found in the voluntary sector as follows: ‘‘our model is a unified view of one process that underlies all this otherwise confusing diversity. By grounding the model in universal aspects of organizations (i.e., all organizations contain people), sociodemographic variables (i.e., all individuals have a value of age), and social networks (i.e., all networks are homophilous), and ignoring the aspects for which we cannot yet account y we avoid the pitfall of focusing on uniqueness at the expense of generality and redirect attention to those phenomena that we may be able to explain’’ (see also Mark, 1998). Similarly, we hope to have illustrated that simple general mechanisms such as homosocial reproduction, niche carving, and strategic distance generate novel propositions with respect to team composition outcomes. This really implies integrating ecology and strategy arguments, which we think has great potential. In future work, we hope to explore the ecology – strategy interface further, particularly by developing similar arguments for other cases than market partitioning. Second, because top management research in the past has mainly focused on demographic characteristics of executives, we also did so in developing our theory, ignoring the social networks of executives. One should be aware that there exists an ongoing debate in the literature about the relative explanatory power of the attributes of individuals per se versus the relations between them (Popielarz & McPherson, 1995; Reagans, Zuckerman, & McEvily, 2004). In future work, it might be worthwhile (though probably even more demanding) to also collect longitudinal data on the network ties of executives (McPherson et al., 2001). Luckily, however, the work of
Top Management Team Composition and Organizational Ecology
125
McPherson and colleagues shows that the relative positions of individuals in social space can act as a proxy for the network connections people have, making the debate less salient. Specifically, ‘‘through the organizing structure of social space, the attributes of individuals summarize their homophilous relations with others who are near and distant in social space’’ (Popielarz & McPherson, 1995, p. 716). Third, we were silent about executive migration. It may be, though, that the market-partitioning outcome of the process is only sustainable when top management team mobility between organizations is relatively low. This is because frequent within-industry migration fosters transmission of routines and strategies across the population’s organizations, increasing their similarity (Wezel, Cattani, & Pennings, 2005). Indeed, Sørenson’s (1999b) study on executive migration among commercial television stations suggests that recruitment of executives from competitors increases niche overlap and competition. Two additional comments on within-industry executive migration are worth making, though. For one, migration probably also follows predictable patterns (see the introduction in Baum & Singh, 1994a). Specifically, it is not unlikely that migration of managerial characteristics and related capabilities (Sørenson, 1999b) mainly occurs among organizations that are similar. If this is the case, then the countervailing impact of managerial mobility on the market-partitioning process will be tempered. Moreover, recent empirical research shows that within-industry executive mobility is especially high in homogeneous compared to heterogeneous industries. Parrino (1997, p. 195), studying 977 CEO succession events in largepublic firms between 1969 and 1989, finds that ‘‘the likelihood of turnover, forced turnover and outside succession increase with the similarity of the firms in an industry. Furthermore, the likelihood that a fired CEO is replaced by an executive from another firm in the same industry also increases with industry homogeneity.’’ In homogeneous industries, organizations pursue similar strategies, and as a result need similar managerial talent. This facilitates the exchange of executives among organizations within the industry. Interestingly, this suggests that the market-partitioning process we described above is difficult to revoke, as organizational diversity hampers the transmission of routines between organizations via executive migration. Fourth, note that the ideas developed in this paper are similar to the arguments presented in the seminal work of McPherson and colleagues on voluntary organizations, and Sørensen’s work on the ecology of managerial tenure distributions. However, there are important differences. McPherson’s ecological analyses are restricted to competition among voluntary organizations for members, invoking the principle of homophily to predict the location of organiza-
126
CHRISTOPHE BOONE ET AL.
tions in Blau (demographic) space. In other words, their focus is on organizational input markets, ignoring the specific output of these voluntary organizations, and the competition resulting thereof in output or product markets. Although it might be interesting to do so, we did not focus on competition among organizations for managerial talent (for an exception, see Sørensen, 1999b). Instead, we theorized on how homosocial reproduction impinges on competition in output (product) markets, which is important when one wants to extend the study of the implications of homophily to for-profit organizations. Sørensen (1999a) does focus on competition between organizations in products markets in his study of organizational growth of commercial television stations. His main finding is that the distance of a focal firm’s top management team mean tenure to the mean tenure of competitors increases a focal firm’s growth rate. The putative reason is that overlap in tenure goes hand in hand with overlap in managerial capabilities, which leads to greater competition for resources since managers shape a firm’s pattern of resource utilization. Although Sørensen also urges researchers to analyze the ecological interplay of top management team composition in relation to other organizations, we extend this logic in three ways. First, Sørensen restricts his analysis to the firm’s mean managerial tenure, whereas our focus is on demographic homogeneity. Second, Sørensen does not theorize on where specific demographic team distributions come from in the first place, which is the central starting point of our theory. Third, Sørensen empirically models firm growth, whereas we try to develop a more comprehensive model linking micro-level team composition dynamics, strategic behavior and market-level outcomes. Fifth, we would like to reflect on the adaptation-selection debate with reference to the accompanying paper by Wiersema and Moliterno in this volume. We believe that our and their contributions are complementary in at least three ways. First, our paper stresses, as do Campbell (1994) and Meyer (1994), the importance of internal adaptation, whereas Wiersema and Moliterno (2006) focus on external adaptation. That is, in our model, managers adapt to internal goals, preferences, and opportunities and threats, resulting in team reproduction and homogeneity. This is a baseline process that does not exclude the fact that managers need to adapt – and do try so – to external (exogenous) shifts in the environment as well (cf. Boone, van Olffen, van Witteloostuijn, & De Brabander, 2004). Second, the internal adaptation processes we described above are evolutionary in nature, and are well placed to explain incrementally emerging and evolving processes. Wiersema and Moliterno (2006) define external adaptation as reactions to punctuated or discontinuous change, which disturb periods of incremental (internal) adaptation. So, one could argue that punctuated shocks from time to
Top Management Team Composition and Organizational Ecology
127
time disrupt the target of team reproduction. Third, all hypotheses of Wiersema and Moliterno (2006) are consistent with our argument. Homosocial reproduction is most forcefully operating when TMTs have the power to reproduce their characteristics. Hence, in settings where the financial stakeholders are relatively powerful (e.g., in the case of many institutional investors and large blockholders), the process of TMT reproduction may well be undermined. A more fundamental issue is that Wiersema and Moliterno do not make a distinction between the event of CEO dismissal and the adaptive value of such an event. The adaptive power of CEO dismissal is, however, highly questionable, as Wiersema and Moliterno acknowledge in their discussion (see also Wiersema, 2002). For one, evidence to date has not revealed that CEO dismissal leads to better performance. Moreover, our multi-level theory suggests that the massive hypish wave of CEO dismissals, resulting from pressures of the financial community, might even be non-adaptive (cf. Sorge & van Witteloostuijn, 2004). That is, if firms hire similar CEOs (e.g., with a financial background), then homogeneity and competition are likely to increase, which may ultimately undermine firm performance in the long run. In this respect, Wiersema and Moliterno claim that diversity in the US automobile industry was lower in the old era compared to the new era. These opposite claims (is CEO diversity decreasing or increasing over time?) suggest an interesting avenue for future research. Finally, we realize that the theory we presented here is very demanding with respect to data collection. One needs longitudinal demographic data on executives, executive entry and exit events have to be carefully recorded, and team-level data have to be linked to organization-level information on strategies, niches and performance. So, the research design involves a combination of the already demanding requirements of ecology and demography studies. Nevertheless, recent work, focusing on parts of the model, shows that it is doable and, more importantly, that the results are very promising (Sørensen, 1999a; Boone, van Olffen, van Witteloostuijn, & De Brabander, 2004; Pennings & Wezel, 2005). Indeed, we strongly believe that this is the type of ecology – strategy dialogue that is very likely to produce new insights in the future, linking different levels of analyses in an overarching adaptation – selection logic.
NOTES 1. Note that this role of network ties is assumed rather than estimated in the McPhersonian line of work. Indeed, Dobrev (2005) job-flocking’s argument suggests that ‘‘ecological ties of observability’’ may be enough. That is, even without direct
128
CHRISTOPHE BOONE ET AL.
social ties, career outcomes are affected by homophily in larger professional communities (such as top managers). 2. Admittedly, McPherson and colleagues do not argue that preference and choice are not important. In fact, they (ex post) defend their choice to focus on voluntary associations because ‘‘they represent a unique arena for watching the strong interplay of structurally induced and choice-produced homophily’’ (McPherson, SmithLovin, & Cook, 2001, p. 432). Notwithstanding this disclaimer, these authors overwhelmingly focused on the structural sources of homophily. 3. Interestingly, in the majority of recent sociological work on homophily researchers limit attention to ascriptive demographic characteristics such as gender and ethnicity. There is, however, no reason to exclude other individual characteristics from the analysis. Note, in fact, that most classical sociological treatments of homophily such as Lazarsfeld and Merton (1954) did not restrict it to ascriptive characteristics (McPherson et al., 2001; Ruef et al., 2003). The bottom line for a purely homophilous mechanism to apply to both achieved and ascribed characteristics is that the functional contributions of those characteristics must be ruled out (Ruef et al., 2003, p. 196). That is, functional theories would suffice if group members are mainly selected based on the valuable and complementary competences they possess to ensure the success of the collectivity. 4. Note that the set of reasons we offer for people’s preference for homogeneity do not necessarily imply conscious, deliberate choices that people make with respect to, for instance, the recruitment of similar members in groups – on the contrary. As already suggested, these preferences are firmly rooted in human beings as they probably have evolutionary origins. Specifically, evolutionary theories of human cooperation, such as kin selection (Hamilton, 1964) and reciprocity (Axelrod & Hamilton, 1981), trace the capacity of humans to behave cooperatively back to the evolutionary advantage resulting from fostering kin and from tit-for-tat behavior. What both theories have in common is that cooperation is expected to be more common among similar individuals, even if the underlying mechanisms are different. 5. Yet another insight of Campbell (1994), related to Meyer’s (1994) argument, deserves further attention: executives as parasites. His argument is that the executives’ efforts to maximize their own fitness are likely to be at odds with the fitness of their firm. If so, we need to shift the level of analysis from firms to executives. In this paper, we developed a similar logic, arguing that particularly homogeneous top management teams may be more interested in their own team fitness than in the fitness of their firms. Combining this with Campbell’s argument would imply that the likelihood of survival of a firm decreases with the homogeneity of its top management team. In their attempt to survive as a group, particularly if under threat, the executives as parasites will be involved in homosocial reproduction strategies that will negatively impact upon the survival chances of their host – i.e., the firm they are heading. In future work, we hope to test this hypothesis. 6. This suggests a further refinement of the argument. Top management team homogeneity is positively associated with inertia, because ‘cloned’ executives share a preference for similar routines. Such inertia can be an advantage or a disadvantage, from the organization’s perspective. On the one hand, organizational ecology argues that such inertia is positively related with survival by providing a buffer against changes that would harm the organization’s accountability, reliability, and repro-
Top Management Team Composition and Organizational Ecology
129
ducibility, as well as its identity. On the other hand, this inertia may turn into a disadvantage in turbulent times, since the stifled routines may no longer fit with the new environmental conditions. So, it may be that top management team homogeneity is positively associated with the likelihood of organizational survival in stable environments, but negatively so in dynamic ones. Note that this hypothesis has already been explored in group research, providing evidence for the above logic (Boone, van Olffen, van Witteloostuijn, & De Brabander, 2004). 7. A similar logic might be applied at the population level (cf. Miner & Haunschild, 1995; Miner & Anderson, 1999). Top management team reproduction may be either beneficial or harmful for the survival changes of the population as a whole by promoting organizational diversity, depending upon the nature of environmental change the population is facing. 8. Interestingly, this implies that organizational diversity will be relatively low in populations where entry to a specific profession is highly regulated and institutionalized by, for instance, professional associations – a proposition consistent with institutional theory (Meyer & Rowan, 1977; DiMaggio & Powell, 1983). In such cases, higher-order imitation and selection processes homogenize the pool of managerial talent that enters into an industry (for an example, see the Dutch audit industry: Maijoor & van Witteloostuijn, 1996; cf. Campbell, 1994). More broadly, homophily processes may occur at the level of the population as a whole, rather of the organization. In the current paper, we ignore this case. 9. Both conditions are here taken to be exogenous. Of course, we could complicate the argument further by taking aboard endogeneity of this pair of conditions. For instance, clever top managers may be able to carve out new niches by creating new demand for a novel product, or smart non-executives may decide to attract new types of executives from outside the current ‘‘population’’ of (potential) candidates. We leave these endogeneity issues for future work. 10. Note that this finding only appears after controlling for multi-market contact among rivals. As predicted, multi-market contact decreases rivalry because the threat of competitive retaliation increases when competitors meet in several markets (van Witteloostuijn & van Wegberg, 1992; van Wegberg & van Witteloostuijn, 2001). This underscores Gimeno and Woo’s plea to disentangle crowding from multi-market contact effects. 11. Unlike Proposition 2, this proposition is formulated at the level of the organization, and not at the dyad level. This is because the dependent variable is organizational performance. It is not very meaningful to use an absolute distance measure of performance between two firms as a criterion variable. 12. Note that as interaction effects are symmetric, we also expect that specialist organizations (i.e., with a large strategic distance from competitors) will perform well especially when their top management teams are homogeneously composed.
ACKNOWLEDGMENT We would like to thank Joel Baum and Stanislav Dobrev for their very insightful comments. Of course, all remaining errors are ours.
130
CHRISTOPHE BOONE ET AL.
REFERENCES Amburgey, T. L., Dacin, T., & Kelly, D. (1994). Disruptive selection and population segmentation: Interpopulation competition as a segregating process In: J. A. C. Baum, & J. V. Singh, (Eds), Evolutionary dynamics of organizations, (pp. 240–254). New York: Oxford University Press. Axelrod, R., Hamilton, W. D. (1981). The Evolution of Cooperation. Science, 211, 1136–1139. Baum, J. A. C., & Haveman, H. (1997). Love thy neighbor?: Differentiation and agglomeration in the Manhattan hotel industry, 1898–1990. Administrative Science Quarterly, 42, 304– 338. Baum, J. A. C., & McKelvey, B. (1999). Variations in organization science: In honor of Donald T. Campbell. Oxford, UK: Sage. Baum, J. A. C., & Mezias, S. (1992). Localized competition and organizational failure in the Manhattan hotel industry 1898–1990. Administrative Science Quarterly, 37, 580–604. Baum, J. A. C., & Oliver, C. (1996). Toward an institutional ecology of organizational founding. Academy of Management Journal, 39, 1378–1427. Baum, J. A. C., & Singh, J. V. (1994a). Organizational hierarchies and evolutionary processes: Some reflections on a theory of organizational evolution. In: J. A. C. Baum, & J. V. Singh (Eds), Evolutionary dynamics of organizations (pp. 3–22). New York: Oxford University Press. Baum, J. A. C., & Singh, J. V. (1994b). Organizational niche overlap and the dynamics of organizational founding. Organization Science, 5, 483–502. Baum, J. A. C., & Singh, J. V. (1994c). Organizational niche overlap and the dynamics of organizational mortality. American Journal of Sociology, 100, 346–380. Baum, J. A. C., & Singh, J. V. (1996). Dynamics of organizational responses to competition. Social Forces, 74, 1261–1297. Beckman, C. M., Haunschild, P. R., & Phillips, D. J. (2004). Friends or strangers? Firm-specific uncertainty, market uncertainty, and network partner selection. Organization Science, 15, 259–275. Blau, P. M. (1977). Inequality and heterogeneity. New York: The Free Press. Boone, C., Carroll, G. R., & van Witteloostuijn, A. (2002). Resource distributions and market partitioning: Dutch daily newspapers, 1968–1994. American Sociological Review, 67, 408–431. Boone, C., Carroll, G. R., & van Witteloostuijn, A. (2004). Size, differentiation and the growth and profitability of Dutch daily newspapers. Industrial and Corporate Change, 13, 117– 148. Boone, C., van Olffen, W., & van Witteloostuijn, A. (1998). Psychological top management team make up and economic firm performance: An experimental study. Journal of Economic Psychology, 19, 43–73. Boone, C., van Olffen, W., & van Witteloostuijn, A. (2005). Team locus-of-control composition, leadership structure, information acquisition, and financial performance: A business simulation study. Academy of Management Journal, 48, 889–909. Boone, C., van Olffen, W., van Witteloostuijn, A., & De Brabander, B. (2004). The genesis of top management team diversity: Selective turnover among top management teams in Dutch newspaper publishing, 1970–1994. Academy of Management Journal, 47, 633–656. Boone, C., Wezel, F. C., & van Witteloostuijn, A. (2005). An ecological theory of populationlevel diversity dependence. Working Paper, Antwerpen: University of Antwerpen.
Top Management Team Composition and Organizational Ecology
131
Boone, C., & van Witteloostuijn, A. (2004). A unified theory of market partitioning: An integration of resource-partitioning and sunk cost theories.. Industrial and corporate change, 13, 701–725. Bornstein, G., & Ben-Yossef, M. (1994). Cooperation in intergroup and single-group social dilemmas. Journal of Experimental Social Psychology, 30, 52–67. Bornstein, G. (2003). Intergroup conflict: Individual, group, and collective interests. Personality and Social Psychology Review, 7, 129–145. Boyd, R., & Richerson, P. J. (1985). Culture and the evolutionary process. Chicago: University of Chicago Press. Byrne, D. (1971). The attraction paradigm. New York: Academic Press. Campbell, D. T. (1969). Variation and selective retention in socio-cultural evolution. General Systems, 16, 69–85. Campbell, D. T. (1994). How individual and face-to-face group selection undermine firm selection in organizational evolution. In: J. A. C. Baum, & J. V. Singh (Eds), Evolutionary dynamics of organizations (pp. 23–38). New York: Oxford University Press. Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in populations of organizations. American Journal of Sociology, 90, 1262–1283. Carroll, G. R., & Hannan, M. T. (2000). The demography of corporations and industries. Princeton, NJ: Princeton University Press. Carroll, G. R., & Swaminathan, A. (2000). Why the microbrewery movement? Organizational dynamics of resource partitioning in the American brewing industry after prohibition. American Journal of Sociology, 106, 715–762. Clore, G. L., & Byrne, D. (1974). A reinforcement affect model of attraction. In: T. L. Huston (Ed.), Foundations of interpersonal attraction (pp. 143–170). New York: Academic Press. DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48, 147– 160. Dobrev, S. D. (2005). Career mobility and job flocking. Social Sciences Research, 34, 800–820. Dobrev, S. D., Kim, T.-Y., & Carroll, G. R. (2002). The evolution of organizational niches: U.S. automobile manufacturers 1885–1981. Administrative Science Quarterly, 47, 233– 264. Durkheim, E. [1893] (1933). The division of labor in society. Translated by George Simpson. New York: Free Press. Egidi, M. (1996). Routines, hierarchies of problems, procedural behaviour: Some evidence from experiments. In: K. Arrow, E. Colombatto, M. Perlman, & C. Schmidt (Eds), The rational foundations of economic behaviour, (pp. 303–333). London: Macmillan. Feld, S. (1982). Structural determinants of similarity among associates. American Sociological Review, 47, 797–801. Finkelstein, S., & Hambrick, D. C. (1996). Strategic leadership: Top executives and their effects on organizations. St. Paul, MN: West Publishing Company. Gimeno, J., & Woo, C. Y. (1996). Hypercompetition in a multimarket environment: The role of strategic similarity and multimarket contact in competitive de-escalation. Organization Science, 7, 322–341. Hambrick, D. C., & Mason, P. A. (1984). Upper echelons: Organizations as a reflection of its top managers. Academy of Management Review, 9, 193–206. Hamilton, D. (1964). The genetic evolution of social behavior. Journal of Theoretical Biology, 7, 1–16.
132
CHRISTOPHE BOONE ET AL.
Hannan, M. T. (2005). Ecologies of organizations: Diversity and identity. Journal of Economic Perspectives, 19, 51–70. Hannan, M. T., & Freeman, J. (1977). The population ecology of organizations. American Journal of Sociology, 83, 929–984. Hannan, M. T., Ranger-Moore, J., & Banaszak-Holl, J. (1990). Competition and the evolution of organizational size distributions. In: J. V. Singh (Ed.), Organizational evolution: New directions (pp. 246–268). Beverly Hills, CA: Sage. Haveman, H. A. (1995). The demographic metabolism of organizations: Industry dynamics, turnover, and tenure distributions. Administrative Science Quarterly, 40, 586–618. Haveman, H. A. (2000). The future of organizational sociology: Forging ties among paradigms. Contemporary Sociology, 29, 476–486. Haveman, H. A., & Cohen, L. E. (1994). The ecological dynamics of careers: The impact of organizational founding, dissolution, and merger on job mobility. American Journal of Sociology, 100, 153–195. Heiner, R. (1983). The origin of predictable behaviour. American Economic Review, 73, 560– 595. Hinds, P. J., Carley, K. M., Krackhardt, D., & Wholey, D. (2000). Choosing work group members: Balancing similarity, competence, and familiarity. Organizational Behavior and Human Decision Processes, 81, 226–251. Hogg, M. A., Mullin, B. A. (1999). Joining groups to reduce uncertainty: Subjective uncertainty reduction and group identification. In: D. Abrams, M. A. Hogg (Eds), Social identity and social contagion (pp. 249–279). Malden, MA: Blackwell. Hotelling, H. (1929). Stability in competition. Economic Journal, 39, 41–57. Huston, T. L., & Levinger, G. (1978). Interpersonal attraction and relationships. Annual Review Psychology, 29, 115–156. Jackson, S. E., Brett, J. F., Sessa, V. I., Cooper, D. M., Julin, J. A., & Peyronnin, K. (1991). Some differences make a difference: Individual dissimilarity and group heterogeneity as correlates of recruitment, promotions, and turnover. Journal of Applied Psychology, 76, 675–689. Kanter, R. M. (1977). Men and women of the corporation. New York: Basic Books. Lawrence, B. S. (1997). Opening the black box of organizational demography. Organization Science, 8, 1–22. Lazarsfeld, P. F., & Merton, R. K. (1954). Friendship as a social process: A substantive and methodological analysis. In: M. Berger, T. Abel, & C. H. Page (Eds), Freedom and control in modern society (pp. 18–66). New York: Van Nostrand. LeVine, R. A., & Campbell, D. T. (1972). Ethnocentrism: Theories of conflict, ethnic attitude and group behavior. New York: Wiley. Maijoor, S. J., & van Witteloostuijn, A. (1996). An empirical test of the resource based theory: Strategic regulation in the Dutch audit industry. Strategic Management Journal, 17, 549– 569. March, J. M. (1994). The evolution of evolution. In: J. A. C. Baum, & J. V. Singh (Eds), Evolutionary dynamics of organizations (pp. 239–252). Oxford, UK: Oxford University Press. Mark, N. (1998). Birds of a feather sing together. Social Forces, 77, 453–485. McPherson, J. M. (1983). An ecology of affiliation. American Sociological Review, 48, 519–532. McPherson, J. M., Popielarz, P. A., & Drobnic, S. (1992). Social networks and organizational dynamics. American Sociological Review, 57, 153–170.
Top Management Team Composition and Organizational Ecology
133
McPherson, J. M., & Ranger-Moore, J. R. (1991). Evolution on a dancing landscape: Organizations and networks in dynamic Blau space. Social Forces, 70, 19–42. McPherson, J. M., & Rotolo, T. (1996). Testing a dynamic model of social composition: Diversity and change in voluntary groups. American Sociological Review, 61, 179–202. McPherson, J. M., & Smith-Lovin, L. (1987). Homophily in voluntary organizations: Status distance and the composition of face-to-face groups. American Sociological Review, 52, 370–379. McPherson, J. M., Smith-Lovin, L., & Cook, J. (2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415–444. Meyer, M. W. (1994). Turning evolution inside the organization. In: J. A. C. Baum, & J. V. Singh (Eds), Evolutionary dynamics of organizations (pp. 109–116). Oxford, UK: Oxford University Press. Meyer, J. W., & Rowan, B. (1977). Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83, 340–363. Miner, A. S. (1994). Seeking adaptive advantage: Evolutionary theory and managerial action. In: J. A. C. Baum, & J. V. Singh (Eds), Evolutionary dynamics of organizations (pp. 76– 89). Oxford, UK: Oxford University Press. Miner, A. S., & Anderson, P. (1999). Industry and population-level learning: Organizational, inter-organizational and collective learning processes. In: A. S. Miner, & P. Anderson (Eds), Advances in strategic management: Population-level learning and industry change (pp. 1–31). Stamford CT: JAI Press. Miner, A. S., & Haunschild, P. R. (1995). Population-level learning. In: L. L. Cummings, & B. M. Staw (Eds), Research in organizational behavior (pp. 115–166). Greenwich, CN: JAI Press. Miller, D. (1999). Selection processes inside organizations: The self-reinforcing consequences of success In: J. A. C. Baum, & B. McKelvey (Eds), Variations in organization science (pp. 93–112). Thousand Oaks: Sage. Parrino, R. (1997). CEO turnover and outside succession: A cross-sectional analysis. Journal of Financial Economics, 46, 165–197. Pe´li, G., & Nooteboom, B. (1999). Market partitioning and the geometry of the resource space. American Journal of Sociology, 104, 1132–1153. Pennings, J. M., Lee, K., & van Witteloostuijn, A. (1998). Human capital, social capital, and firm dissolution. Academy of Management Journal, 41, 425–440. Pennings, J. M., & Wezel, F. C. (2005). Heterogeneity shifts due to member mobility: New insights on diversity and performance. Working Paper, Tilburg: Tilburg University. Pfeffer, J. (1983). Organizational demography. Research in Organizational Behavior, 5, 299–357. Pfeffer, J. (1997). New directions for organization theory: Problems and prospects. New York: Oxford University Press. Podolny, J. M. (1994). Market uncertainty and the social character of economic exchange. Administrative Science Quarterly, 39, 458–483. Popielarz, P. A., & McPherson, J. M. (1995). On the edge or in between: Niche position, niche overlap, and the duration of voluntary association memberships. American Journal of Sociology, 101, 698–720. Reagans, R., Zuckerman, E., & McEvily, B. (2004). How to make the team: Social networks vs demography as criteria for designing effective teams. Administrative Science Quarterly, 49, 101–133.
134
CHRISTOPHE BOONE ET AL.
Ruef, M., Aldrich, H. E., & Carter, N. M. (2003). The structure of founding teams: Homophily, strong ties, and isolation among U.S. entrepreneurs. American Sociological Review, 68, 195–222. Schaubroeck, J., Ganster, D. C., & Jones, J. R. (1998). Organization and occupational influences in the ASA process. Journal of Applied Psychology, 83, 869–891. Schneider, B. (1987). The people make the place. Personnel Psychology, 40, 437–453. Schneider, B., Goldstein, H. W., & Smith, D. B. (1995). The ASA framework: An update. Personnel Psychology, 48, 747–773. Schneider, B., Smith, D. B., Taylor, S., & Fleenor, J. (1998). Personality and organizations: A test of the homogeneity of personality hypothesis. Journal of Applied Psychology, 83, 462–470. Sherif, M., Harvey, O., White, B., Hood, W., & Sherif, C. (1961). Intergroup conflict and cooperation: The robber’s cave experiment. Norman, OK: Institute of Group Relationships of the Univiversity of Oklahoma. Sørensen, J. B. (1999a). The ecology of organizational demography: Managerial tenure distributions and organizational competition. Industrial and Corporate Change, 8, 713–745. Sørensen, J. B. (1999b). Executive migration and interorganizational competition. Social Science Research, 28, 289–315. Sorge, A. M., & van Witteloostuijn, A. (2004). The (non)sense of organizational change: An essay about universal management hypes, sick consultancy metaphors and healthy organization theories. Organization Studies, 25, 1205–1231. Staw, B. M., Sandelands, L. E., & Dutton, J. E. (1981). Threat-rigidity effects in organizational behavior: A multi-level analysis. Administrative Science Quarterly, 26, 501–524. Stinchcombe, A. (1997). On the virtues of the old institutionalism. Annual Review of Sociology, 23, 1–18. van Wegberg, M. J. A. M., & van Witteloostuijn, A. (2001). Strategic management in the new economy: Modern information technologies and multichannel contact strategies. In: J. A. C. Baum, & H. R. Greve (Eds), Advances in strategic management: Multiunit organization and multimarket strategy (pp. 265–306). Greenwich, CT: JAI Press. Westphal, J. D., & Zajac, E. J. (1995). Who shall govern? CEO/board power, demographic similarity, and new director selection. Administrative Science Quarterly, 40, 60–83. Wezel, F. C., Cattani, G., & Pennings, J. M. (2005). Competitive implications of inter-firm mobility. Working Paper. Tilburg: Tilburg University. Wiersema, M. F. (2002). Holes at the top: Why CEO firings backfire. Harvard Business Review, 80, 70–77. Wiersema, M. F., & Bantel, K. A. (1993). Top management team turnover as an adaptation mechanism: The role of the environment. Strategic Management Journal, 14, 485–505. Wiersema, M. F., & Moliterno, T. P. (2006). CEO turnover in the new era: A dialogue with the financial community. In: J. A. C. Baum, S. D. Dobrev, A. van Witteloostuijn (Eds), Strategy and Ecology: Advances in Strategic Management (Vol. 23, pp. 137–173). Oxford UK: JAI/Elsevier. Williams, K. Y., & O’Reilly, C. A. (1998). Demography and diversity in organizations: A review of 40 years of research. In: B. Staw, & R. Sutton (Eds), Research in organizational behavior (Vol. 20, pp. 77–140). Greenwich, CT: JAI Press. van Witteloostuijn, A. (2003). A game-theoretic framework of trust. International Studies of Management & Organization, 33(2), 53–71 Part II.
Top Management Team Composition and Organizational Ecology
135
van Witteloostuijn, A., & Boone, C. (2006). A resource-based theory of market structure and organizational form. Academy of Management Review (Vol. 31, pp. 409–426). van Witteloostuijn, A., & van Wegberg, M. J. A. M. (1992). Multimarket competition: Theory and evidence. Journal of Economic Behavior and Organization, 18, 273–282. Zajac, E. J., & Westphal, J. D. (1996). Who shall succeed? How CEO/board preferences and power affect the choice of new CEOs. Academy of Management Journal, 39, 64–90.
This page intentionally left blank
136
CEO TURNOVER IN THE NEW ERA: A DIALOGUE WITH THE FINANCIAL COMMUNITY Margarethe F. Wiersema and Thomas P. Moliterno ABSTRACT Scholars working in the strategy area have long held that one of the primary ways in which organizations adapt to external changes is through strategic choice. Inasmuch as a new CEO can result in a new strategic direction for the firm, the CEO turnover event itself is an important way by which organizations can signal an alteration in the direction of the firm. In this chapter, we explore how and why CEO turnover has become one of the most powerful indicators of adaptation the firm can make and propose a research agenda to guide future work on CEO turnover.
INTRODUCTION One of the fundamental research issues in management is how organizations develop and adapt over time in response to environmental change (Lawrence & Lorsch, 1967; Levinthal & March, 1981; Pfeffer & Salancik, 1978; Thompson, 1967; Tushman & Romanelli, 1985). Organizational adaptation represents a ‘‘change in a significant organizational attribute, such Ecology and Strategy Advances in Strategic Management, Volume 23, 137–173 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23005-5
137
138
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
as basic business strategy or organizational structure, in response to environmental change’’ (Levinthal, 1994, p. 171), and so adaptation is seen as the process of organizational change in response to the demands imposed by external forces. Researchers therefore have been primarily interested in understanding the environmental forces – whether economic, sociopolitical, technological, regulatory, or competitive – that may stimulate and influence organizational adaptation (Keats & Hitt, 1988; Kelly & Amburgey, 1991; Romanelli & Tushman, 1988). An emerging body of work in this area argues that organizational change occurs sporadically, and in response to infrequent, but dramatic, changes in the environmental context (Meyer, Brooks, & Goes, 1990). The central premise of this stream of research is that industrial contexts are characterized by relatively long periods of stability, which are interrupted – or ‘‘punctuated’’ – by periods of discontinuous change, and it is these environmental punctuations which ultimately precipitate organizational change (Haveman, 1992, 2001; Meyer, 1982). Accordingly, organizational adaptation is viewed not as slowly emergent process, but rather a fairly dramatic response to distinct periods of equally dramatic contextual change. Scholars have also suggested that these environmental punctuations can have various drivers: political upheavals, changes in governmental regulations, and/or technological innovations (e.g., Astley, 1985; Carroll, 1987; Romanelli, 1989). Thus, organizations experience periods of relative inertia (Hannan & Freeman, 1989), which are interrupted by these dramatic changes in their environmental context, and organizations make changes – that is, they adapt – as a way of responding to these contextual changes. Scholars working in the strategy area have long held that one of the primary ways in which organizations adapt to external changes is through strategic choice (Andrews, 1971; Child, 1972; Lawrence & Lorsch, 1967; Schwenk, 1988). By determining policies and programs, allocating resources, and setting strategic direction, the executives of the firm can enable the firm to be better positioned with respect to external changes and developments (Bower, 1970; Quinn, 1980; Schwenk, 1988; Selznick, 1957). As a result, the firm’s leadership represents an important bridge between the organization and its environment. Given a firm’s senior executives key role at the interface between the firm and the environment, it is reasonable to expect that replacing these individuals charged with determining strategic choices and setting organizational context can serve as an important mechanism for organizational adaptation (Finkelstein & Hambrick, 1990; Helmich & Brown, 1972; Tushman, Virany, & Romanelli, 1985). Coupled, then, with the idea of environmental punctuations, we might well expect that bringing
CEO Turnover in the New Era
139
in a new CEO would be an appropriate organizational response to such dramatic changes in the firm environmental context, and, indeed, Haveman, Russo, and Meyer (2001) have empirically demonstrated just such a relationship. Specifically, they find that following environmental punctuations due to regulatory changes, rates of CEO succession gradually increased in both the California-based hospital and thrift industries. So, while there are strong theoretical and empirical justifications for understanding CEO turnover as a firm-level adaptation mechanism, what is less clear is why firms favor this particular mechanism. Since the CEO sets the strategic direction for the firm, the decision to replace the CEO (i.e., the announcement of a CEO departure) carries with it the signal of the organization’s intended adaptation. Why, then, is the CEO turnover event such a powerful signal adaptation? Is it this question that we endeavor to address in this chapter? To address this question, we begin by observing that since the 1970s, fundamental macroeconomic changes in global competition, technology, deregulation, and demography significantly altered the complexity and uncertainty of the competitive environment for U.S. corporations.1 The documented industry level effects of globalization, for example, clearly indicate that foreign-based competition adversely impacts domestic industry profitability and creates imperatives for domestic firms to respond to a new and often unfamiliar set of competitors.2 Yet despite the adverse impact of these environmental changes on competitive position and profitability, U.S. firms prior to the late-1980s were by and large unresponsive and resistant to taking action to counter these forces in their external environment. We refer to this period as the ‘‘Old Era’’, characterized by considerable organizational inertia, with contextual conditions that shielded the firm’s board of directors and management from competitive conditions that necessitated organizational change. When firms experienced poor performance, CEOs were rarely dismissed due to the difficulty of ascertaining blame, the lack of external demands for change, and the board’s reluctance to hold the CEO accountable. Furthermore, the incumbent CEO was solidly entrenched as head of the organization and largely controlled the board, almost guaranteeing no significant departures from the past. Not surprisingly, CEO dismissals during this period were rare, usually the result of an extreme crisis situation, and generally driven by internal, rather than external, concerns over performance. During the mid- to late-1980s, a period of discontinuous change occurred in the U.S. corporate context that created imperatives for organizations to respond and adapt to macroeconomic trends of global competition,
140
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
technological change, and deregulation. For the first time, the financial community3 emerged as a force exerting considerable pressures for corporate accountability and shareholder wealth maximization, and meeting these expectations become central to how these constituents evaluated the firm and its management. As a result, an increased focus on creating investor value has become the critical issue facing the management of public firms. This contextual shift occurred within a very short period of time (mid- to late-1980s) and was all encompassing: no industry or firm was exempt from these financial market pressures. We view the emergence of the financial community and its attendant ability to place pressures on organizations for change as an environmental punctuation that signified the beginning of a ‘‘New Era’’. In this era of greater corporate accountability failure to meet expected financial performance targets (quarterly earnings per share) inevitably resulted in pressures on the firm’s leadership and its board of directors.4 Faced with dire consequences if they did not respond to an increasingly vocal community of investors, analysts, and stakeholders, public companies and their boards became eager to please. Not surprisingly, then, concomitant with the increasing clout of the financial community, we see a dramatic increase in the rate of CEO turnover – particularly involuntary turnover – in U.S. corporations. Research that examines the period before the mid- to late-1980s indicates that the CEO turnover not accounted for by age-related retirements (e.g., non-routine turnover) ranged from 13% to 45% (Cannella & Lubatkin, 1993; Denis & Denis, 1995; Denis & Kruse, 2000; Denis & Serrano, 1996; Parrino, 1997), whereas by the late 1990s, nonroutine CEO turnover events accounted for as much as 71% of CEO turnover in the 500 largest U.S. corporations (Wiersema, 2002). We argue, then, while macroeconomic drivers of organizational change and adaptation were present in the U.S. corporate context since the 1970s, it was not until the rise in power and visibility of the financial community in the mid- to late-1980s that firms took seriously the need to engage in organizational adaptation, and when they began to do so, CEO turnover surfaced as a particularly power expression of the firm’s intention of adaptation. Changes in the firm’s leadership – and in particular, CEO turnover – can be seen, then, as a signal that the firm is responding to external calls for organizational accountability and adaptation. To understand why firms increasingly turn to CEO turnover as a signal of adaptation after the rise in power of the financial community, we employ the metaphor of a dialogue to characterize the realities of corporate leadership in this New Era. This dialogue can be conceptualized as a two-way process of signaling (Spence, 1974). On one ‘‘side’’ of this dialogue is the financial
CEO Turnover in the New Era
141
community, which is constantly signaling – that is, communicating – its opinion on the firm’s efforts to maximize shareholder wealth: takeover attempts, analysts’ stock ratings, and institutional trading activity can all be understood as modes of communication in this vein. That is, as the participant on one side of this dialogue, the financial community articulates its satisfaction (or lack thereof) with the current direction of the firm. As the financial community became more ‘‘vocal’’ in this dialogue, it significantly altered how the firm’s board and top management viewed the relationship with these external constituents. A sense of accountability to these external constituents represents a major shift in how management perceived the responsibilities of a public corporation. For the first time, the firm’s board of directors and senior management are cognizant and sensitive toward the concerns of the financial community over the firm’s performance and competitive position, and – most importantly – realize the need to communicate ‘‘back’’ to the external community. So for its part, the firm participates in this dialogue with discrete actions that indicate it has ‘‘heard,’’ and is responsive to, the financial community’s opinion. For example, restructuring and downsizing announcements might be seen as articulations from the firm’s side of the dialogue. We propose, then, that scholars might gain traction by perceiving the board’s decision to replace the firm’s CEO in terms of this dialogue. Given the CEO’s role in defining strategic choice, and position in the liminal space between the firm and the environment, the decision to replace the CEO is one of the most powerful signals the firm can send about its intended future direction and willingness to adapt and respond to external demands for change. Thinking in terms of this dialogue metaphor, this chapter examines how and why CEO turnover has become one of the most powerful indicators of adaptation the firm can make. First, we elaborate on the interfacing role that the firm’s leadership plays in aligning the firm with its environment. Literature on top executives and their effects on organizations document the significant ways in which managers can influence organizational context and outcomes (Finkelstein & Hambrick, 1996), and so lead us to conclude that CEO turnover can be a very real form of organizational adaptation, and as such, the turnover event itself has powerful connotations of an ongoing process of adaptation. We then discuss how historically, despite its importance, a change in leadership was rarely used as a mechanism for organizational adaptation because organizations were largely insulated from external pressures – there was no dialogue that necessitated the powerful signal that is sent by CEO turnover. Accordingly, firms during this period were largely subject to inertial forces. This Old Era is then contrasted with
142
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
the New Era where firms are increasingly subject to heightened performance expectations from which few organizations are immune. It is in this context that we see the emergence of the dialogue we have described, and examine why CEO turnover is such a powerful statement in that dialogue. Here, we draw on insights from three theoretical perspectives to advance a research agenda for examining why CEO turnover is increasingly used as part of the dialogue we describe. Specifically, we draw on theories of escalation of commitment (Staw, 1981), TMT socialization (Schein, 1968; Van Maanen, 1978; Van Maanen & Schein, 1979), and institutionalism (DiMaggio & Powell, 1983) to examine how and why CEO turnover is used as a signal of adaptation. The research agenda we outline here proposes that the answer lies in the dialogue that exists between the firm and its external constituents. This interpretation provides an explanation for why CEO turnover, and in particular involuntary CEO turnover, has exploded since the late 1980s, and in addition, lays a foundation for further examination of why CEO turnover can be a potentially powerful expression of the firm’s efforts at adaptation.
THE ROLE OF CORPORATE LEADERSHIP Many organization and strategic management theorists have highlighted the important role that the leadership of the firm plays in aligning the firm with its environment (Barnard, 1938; Hambrick & Mason, 1984; Lawrence & Lorsch, 1967; Thompson, 1967; Tushman & Romanelli, 1985). This contingent perspective on a firm’s leadership assumes that managers differ in their abilities, experiences, knowledge, and cognitive perspectives and that these differences have implications for sensing and interpreting the organization’s environment, as well as determining the plan of action and its execution (Hambrick & Mason, 1984; Schwenk, 1988). Because organizations face different environments and can pursue a variety of strategies, aligning the manager’s skills, knowledge, and cognitive background with the requirements of the job should lead to a better fit and the basis for superior firm performance (Gupta, 1984). Evidence from the literature indicates that managerial traits, demography, and cognitive perspectives all play a role in determining how the firm’s leadership will interpret the external and internal environment that the firm faces, as well as where and how to respond to changes in that environment (Barr, Stimpert, & Huff, 1992). There has been a vast literature that has theoretically proposed, and empirically investigated, the importance of matching manager type to the
CEO Turnover in the New Era
143
firm’s strategy, structure, and environment (Chaganti & Sambharya, 1987; Govindrarajan, 1989; Gupta & Govindarajan, 1984; Snow & Hrebiniak, 1980; Thomas, Litschert, & Ramaswamy, 1991). This literature is premised on the notion that firms face different environments, and as a result require different strategies (Lawrence & Lorsch, 1967; Miles & Snow, 1978; Porter, 1980; Thompson, 1967). These studies also assume that managers are not infinite in their capabilities and skills to manage, and as a result they lack flexibility to function effectively in all strategic contexts (Gupta & Govindarajan, 1984): the backgrounds, experiences, and education of certain managers may render them more adept at managing certain kinds of organizations and environments. That is, due to the unique demands imposed by certain environments and organizational strategies and structures, certain skill sets may be required for optimal performance. The empirical evidence from these studies does indeed show that certain types of strategies are associated with certain managerial background characteristics such as length of organizational tenure and functional background, and that there are significant performance implications of matching strategy to manager. For example, firms with low-cost strategies were found to benefit from top executives with production backgrounds, whereas firms pursuing a differentiation strategy benefited from top executives with R&D backgrounds (Govindrarajan, 1989). In a similar vein, demography theory, while predominantly focused on group composition characteristics, has also highlighted the importance of an individual’s demographic traits as the basis for understanding subsequent organizational outcomes (Hambrick & Mason, 1984). In particular, an individual’s tenure within the organization has been proposed as a key factor in determining his or her sensitivity to new sources of information and the likelihood of taking action that departs from the status quo (Finkelstein & Hambrick, 1990). Greater organizational tenure also has been shown to lead to greater familiarity, producing and adhering to a common perspective and interpretation of events (Allen & Cohen, 1969; March & Simon, 1958). In addition, with increasing tenure, individuals rely more on past experience instead of seeking out new stimuli (Katz, 1982), resulting in increased isolation from outside sources of relevant information (Dubin, 1972; Pelz & Andrews, 1966) and less openness to new information (Finkelstein & Hambrick, 1990). Researchers have also found that increased organizational tenure in positions of leadership leads to an ‘‘internal focus’’ and significantly less change and greater commitment to the status quo (Boeker, 1997; Finkelstein & Hambrick, 1990; Hambrick, Geletkanycz, & Fredrickson, 1993; Miller, 1991; Wiersema, 1992).
144
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
Lastly, in determining strategic choices, managers must absorb, process, and disseminate information about both internal and external issues, problems, and opportunities (McCall & Kaplan, 1985). Organizations have external and internal environments that are complex and ambiguous, forcing managers to employ their knowledge structure to facilitate information processing and decision making (Schwenk, 1988; Starbuck & Milliken, 1988). Knowledge structures are formed by experiences and represent a mental template that individuals impose on environmental information to give it form and meaning (Fisk & Taylor, 1984). The literature on managerial cognition points to the importance of a manager’s perception of reality and problem-solving frameworks in the interpretation of information and decision-making outcomes (Dutton, Fahey, & Narayanan, 1983; Lyles, 1981; March & Simon, 1958; Reger & Huff, 1993; Priem, 1994; Schwenk, 1988; Sims & Gioia, 1986). Similarly, identification of strategic issues as the basis for future action has been shown to be largely influenced by a CEO’s cognitive map (Dutton et al., 1983). Thus, the organization’s ability to sense and interpret the environment, as well as define and determine the strategic importance of issues it encounters, is in large part determined by its leadership. While much of this literature on top executives and their effects on organizations suggest that the firm’s leadership should be a driver of adaptation to changes in its external environment, there is little empirical evidence to indicate that the leadership of the firm actually serves as a vehicle of organizational adaptation. Instead, it has been found that the executive leadership of the firm tends to act as a stabilizing governance structure, and that as executives become increasingly insular to external stimuli they can actually restrict the firm’s ability to change (Boeker, 1997; Brady & Helmich, 1984; Hambrick et al., 1993; Miller, 1991; Tushman & Romanelli, 1985). Problems of organizational inertia as well as tenure effects and the resultant commitment to the status quo and risk adversity prevent incumbent executives from being able to sense external and internal forces pressuring for change, objectively evaluate and incorporate new information, identify the importance of strategic issues, and determine viable alternative strategic options available to the firm. Thus, despite its position between the firm and environment, the firm’s existing leadership often fails to realign the firm when external forces such as technological change, increased global competition, and deregulation create pressures for change. Instead, it has been found that an externally precipitated crisis often initiates the need for organizational change, and that for change to occur, a firm’s top executive often needs to be replaced (Tushman & Romanelli, 1985; Vancil, 1987). Executive turnover often represents a shift in the cognitive
CEO Turnover in the New Era
145
perspective, knowledge, skills, organizational experiences, and demographic background of the firm’s key decision makers, and so a change in leadership may be one of the few ways the firm can achieve adaptation to changes in the organizational context. However, the issue of whether changes in corporate leadership can be the vehicle of organizational adaptation remains an empirical question. In examining this issue, early researchers assumed that for large bureaucracies to adapt to a changing environment, executive succession would be required (Grusky, 1961, 1963). However, prior to the 1990s, business organizations rarely fired their CEOs, making it difficult to derive a statistical sample. The involuntary departure of a firm’s CEO was not typical, averaging around 13% of all CEO turnovers during the period of the 1970s5 and as shown in Exhibit A, rising to the level of 20% of all CEO turnovers during the period of the 1980s.6 Many of the early succession studies failed to distinguish involuntary from voluntary turnover, making it impossible to ascertain whether turnover served as a means of adaptation (Allen & Panian, 1982; Beatty & Zajac, 1987; Pfeffer & Salancik, 1978). A few studies examining general managers of divisions within larger organizations did find that extremely poor performance led to a greater likelihood that the manager would be replaced (Lieberson & O’Connor, 1979; Tushman et al., 1985). At the corporate level, however, performance-induced involuntary turnover was a relatively rare event (Furtado & Karan, 1990), and most studies conducted during this period could not identify whether leadership changes occurred as a result of external pressures for change. One study that examined environmental conditions, firm performance, and top team turnover did find that firm performance was not a predictor of management turnover, but that a lack of munificence, instability, and complexity in the firm’s industrial environment was a key determinant of the level of turnover with the top management team (Wiersema & Bantel, 1993). This study also found that environmental factors were more likely to predict turnover at lower levels of senior management than at the CEO and president level. The authors postulate that at the very top tier of the organization, managers are less vulnerable to environmental drivers of change.
EXTERNAL CONSTITUENTS IN THE OLD ERA: A LACK OF DIALOGUE The lack of a strong empirical record linking environmental context, poor firm performance, and leadership change in business organizations can be
146
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
Exhibit A. Succession Studies
Allen and Panian (1982) Reinganum (1985) Puffer and Weintrop (1991) Cannella and Lubatkin (1993)a Parrino (1997)b Denis and Serrano (1996)c Denis and Denis (1995) Denis and Kruse (2000)d Shen and Cannella (2002) Wiersema (2002)
History of CEO Succession Studies.
Time Period
Sample Size
Succession Incidence (%)
Dismissal as a % all Successions
1971–1980
242
91
NA
1978–1979 1982
2,500 480
27 11
NA NA
1971–1985
472
100
20e
1969–1989 1984–1989
977 98
100 37
13f 45g
1985–1988
908
22
14g
1985–1992
350
100
36h
1988–1997
387
88
22h
1996–1997
500
17
71i
a
Started with a sample of 800 firms over 15 years and found 1187 CEO succession events, of which 472 events were used as the basis for the study. b Identified based on severance of connections between a firm and the outgoing CEO. c Sample was identified on the basis of CEO turnover. d Identified based on forced resignation, conflict, and departures younger than 60 years of age with no reason reported for departure. e Started with a sample of firms that experienced corporate control activity. f Dismissals were identified based on forced resignation, poor performance, or unexpected retirements. g Started with a sample of firms with management turnovers. h Dismissals were identified based on forced resignation and departures of CEOs younger than 64 years of age. i Dismissals were identified based on articles describing the succession event.
explained by the contextual conditions of the times in which these studies were conducted. Prior to the 1980s, unless a crisis occurred, organizations were fairly immune from having to respond to changes in their external environment. To be sure, significant external macroeconomic shifts – increased global competition, changes in technology, changing demographics, and industry deregulation – occurred during this period, which placed increased pressure on the competitiveness and profitability of firms. In the automobile industry, for example, the oil shocks of the 1970s, the highly
CEO Turnover in the New Era
147
inflationary period that followed, the regulation of vehicle exhaust and fuel efficiency, and increased foreign competition represented major structural shifts for the domestic industry. The entire domestic industry suffered major declines in competitiveness as indicated by a loss in market share to foreign rivals and deteriorating financial performance. Indeed, Chrysler came to the brink of bankruptcy in 1981. Remarkably, despite the severe economic consequences as a result of changes in the competitive environment, GM, Ford, Chrysler, and American Motors were slow or almost absent in adopting significant strategic change. Even though distressed by the financial plight of the industry, the financial community was largely passive and did not exert pressures on the auto firms to adopt changes to improve their competitive position and performance. Not surprisingly, without threat of consequence, firms could ignore the environmental drivers for change. To restate this point, even though the competitive position of the U.S. automobile firms had eroded, a vocal proponent of change was lacking: there was no external constituent with whom the firm was required to dialogue. The lack of accountability on the part of the firm and its management shielded organizations from sensing and adapting to environmental change. Several theoretical perspectives help inform our understanding of this period of relative organizational inertia (Hannan & Freeman, 1989). First, the escalation of commitment literature has provided significant evidence that individuals who have been participants in past decisions determining the direction of the organization find it difficult to disentangle themselves from the courses of action set in motion by those decisions (Staw, 1981; Staw & Ross, 1978). As an example from the U.S. automobile industry, the lengthy nature and sizeable capital outlay of new product development (i.e., 5 years for a new platform), as well as the investment commitments to high levels of vertical integration, meant that decision makers in the industry had long-term cognitive commitments to a certain way of doing business. Not surprisingly, when external conditions changed in the 1970s, managers continued to adhere to past courses of action – particularly in the way vehicles were designed and manufactured – and so failed to adapt to fundamental changes in the competitive landscape.7 The domestic industry suffered from offering too much product breadth, at too high of a cost, due to an over reliance on internal manufacturing and lack of product design rationalization, but managers’ escalation of commitment to prior courses of action inhibited their ability to reverse these decisions. Thus despite competitive erosion, Old Era firms, by being committed to their prior strategic actions and capital investments, were inhibited in their ability to change their course of strategic direction.
148
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
Theories on socialization, with their focus on an individual’s conformity to the demands of the organization (Feldman, 1989), provide additional insight into why organizations in the Old Era were likely to avoid organizational change. The individual’s definition of the environment, practices, and behaviors that are valued, sanctioned, and are shaped by socialization processes within the organization, which can lead to conformity in values and beliefs (Louis, 1980; Schein, 1968; Van Maanen, 1978; Van Maanen & Schein, 1979). The internal promotion and lengthy organizational tenures of the senior executives of most large public U.S. firms during the Old Era provided lengthy organizational socialization which served to reinforce a continuation of the status quo by influencing their perspective and determining the expected response to environmental change (Fondas & Wiersema, 1997). Finally, institutional theory provides a useful framework for understanding why despite the changes in competitive landscape, firms in an industry were relatively ‘‘homogenous’’ in their lack of response. DiMaggio and Powell (1991) observe that homogeneity can best be conceptualized as a process of isomorphism that ‘‘forces one unit in a population to resemble other units that face the same set of environmental conditions’’ (p. 66). This process is, essentially, one of imitation. When firm performance fails to meet expectations, firms engage in search behavior to find a remedy for the unsatisfactory performance (Cyert & March, 1963). When faced with an uncertain and ambiguous environment, this search process is costly, and so is targeted by looking to comparable firms for clues on how to respond to environmentally derived performance shortfalls. That is, copying the response of industry peers is an easy and relatively inexpensive way to conclude this search. Since the environmental changes affect all firms in an industry equally, the response of another industry incumbent, especially the market leader, serves as a viable benchmark for the firm’s response to environmental conditions. After several iterations, the end result of this process is that all firms in the industry adopt highly similar strategies (in this case, a non-response) by which to deal with environmental change. For Old Era U.S. firms, then, the macroeconomic environmental changes of the 1970s necessarily impacted performance. Within the firm, individual level factors – escalation to commitment and socialization – slowed organizational response. But at a higher level of analysis, a mimetic process acted upon these more micro-level responses to slow overall industry reaction. In sum, despite the fact that fundamental macroeconomic changes occurred in the competitive landscape for U.S. corporations from the 1970s onwards which led to the deterioration in competition position and
CEO Turnover in the New Era
149
performance, there were contextual and theoretical reasons for why firms did not undergo significant organizational adaptation. Insights from escalation of commitment theory and socialization theory suggest that organizations and their managers were prone to inertia by their commitment to prior courses of action and their common shared perspectives, and as a result did not attend to external conditions that argued for organizational change. A process of mimetic isomorphism fed upon these inertial forces, and led to industry incumbents observing and mimicking their peers’ nonresponses. Not surprisingly, despite lackluster performance, firms did not engage in adaptation in general, and in particular, a change in leadership at the top was not commonly employed as a signal of organizational adaptation during this period.
NEW ERA OF CORPORATE ACCOUNTABILITY While global competition, technology, regulation, and demography continued to be important environmental realities impacting competitive dynamics in the U.S. corporate environment, a significant development occurred during the mid- to late-1980s, which dramatically impacted the dynamics between public firms and their external constituents. This New Era was triggered by changes in the visibility and power of the financial community in the U.S. corporate context. First, market-based mechanisms, such as the market for corporate control with its threat of hostile takeovers, emerged as a significant disciplinary force on public companies (Denis & Denis, 1995; Jensen, 1986, 1989, 1991; Mitchell & Lehn, 1990; Morck, Shleifer, & Vishny, 1989; Walsh & Kosnik, 1993). Second, political-based mechanisms, such as large blockholders, shareholder activism, institutional investors, and stock analysts began to exert their clout (Black, 1992; Pound, 1992; Gordon & Pound, 1993; Useem, 1996). For the first time, performance expectations and meeting these expectations become central to how these constituents evaluated the firm and its management. As a result, corporations adopted an increased focus on creating investor value. Finally, major changes occurred in the governance structure of firms and the expectations of the board of directors as an internal governance device. The resulting corporate context had significant implications for corporate leadership. In particular, the relationship between the board of directors, senior executives, and the financial community was forever changed. This New Era can best be characterized in terms of an emergent dialogue between the board of directors and management of U.S. publicly held firms and
150
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
various external constituents. Companies in this New Era are increasingly dealing with analysts and large institutional investors, who have taken on a more active role in monitoring the companies’ strategic and financial performance. Each party to this dialogue is deliberately aware of, and sensitive to, the other, and not surprisingly, there is evidence of increased communication, discourse, and signaling (Spence, 1974) back and forth. For example, Wall Street analysts might down grade the firm’s stock as a way of expressing concern over firm performance or institutional shareholders might sell off their shareholdings for a given firm when concerned that shareholder interests are not being fully represented. For its part, the firm might announce layoffs or downsizing programs as a way of communicating to the external constituents that it has ‘‘heard’’ their concerns and is acting in response. One of the most significant developments in this dialogue has been the board’s willingness to dismiss the CEO to appease financial community concerns over the performance and strategic direction of the firm, thereby signaling a willingness and intention to adapt through one of the most dramatic vehicles it has at its disposal: replacing the individual charged with the firm’s strategic direction and choice. In the sections that follow, we discuss how four different components of the environmental punctuation that give birth to the New Era led to the emergence of this dialogue between the firm and its key external constituents. Specifically, we consider the effects of the market for corporate control, institutional investors and shareholder activism, stock analyst coverage, and corporate governance, and regulatory oversight. In each case, we discuss how the contextual change in the environmental landscape resulted in an increased push for corporate accountability, and why firms and their board of directors become increasingly sensitive to these external demands. To better understand the implications of these contextual changes, and develop a series of propositions to guide future research on this New Era, we employ a variety of organizational theories to provide insight into how the dialogue emerges and why CEO turnover is an increasingly powerful signal of adaptation in the context of this dialogue.
Market for Corporate Control The market for corporate control (i.e., the threat of takeover) fully emerged in the late-1980s and early 1990s with the advent of the leveraged buyout and merger and acquisition boom. Mergers and acquisition activity grew from $185 billion in 1990 to $1,680 billion in 1998, breaking all prior records
CEO Turnover in the New Era
151
for merger activity.8 While many of these mergers represented industry consolidations, it also became extremely profitable for third parties to pursue poorly performing firms with the objective of turning performance around: research has shown that the probability of hostile takeovers increases with poor firm performance (Denis & Serrano, 1996; Martin & McConnell, 1991). Activity in leveraged buyouts also exploded during the 1980s, going from four deals valued at $1.7 billion in 1980 to 410 deals valued at $188 billion in 1988.9 The goal in these transactions was to use high-risk financing (i.e., ‘‘junk’’ bonds) to acquire poorly performing firms and either spin-off the assets or take the firm public again at a higher total value than the original purchase price. The potential payoff was considerable, fueling many of the deals that took place.10 Poorly performing firms were generally the target of such ‘‘plays’’ for corporate control, in that they provided the greatest opportunity for a financial turnaround for the new owner. The market for corporate control that emerged in the mid-1980s and the increased merger activity of the 1990s represent a contextual shift for public companies. It contrasts sharply with the Old Era when it was rare for hostile takeovers to occur. During the earlier period, the control and ownership of a public corporation was assumed secure, and as a result management and boards of public companies could continue on their course of action even when those actions resulted in lackluster performance. The lack of repercussions for poor performance meant that firms were not compelled to take actions, and likewise the absence of a market for corporate control meant that the financial community exerted little pressure on poorly performing firms to improve their competitive position. The more active merger market that emerged in the mid-1980s, however, presaged a New Era characterized by a fundamental shift in the relationship between public companies and the financial market. Poorly performing firms and their management became targeted for market discipline. Increased managerial discipline tended to follow takeovers, especially of companies that were performing poorly. Martin and McConnell (1991) found that the rate of managerial turnover increased significantly following corporate takeovers, and that for those firms that had managerial turnover, their pre-takeover performance was significantly worse that for acquired firms in general. Even if the eventual takeover did not occur, studies have found that firms targeted for hostile takeovers experienced twice the rate of subsequent CEO turnover than a random sample of firms (Denis & Serrano, 1996; Martin & McConnell, 1991). Why, then, this empirically observed increase in CEO turnover activity following these changes in the market for corporate control? As discussed in
152
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
the previous section, the inertial tendencies of firms in the Old Era might be partly explained by theories on the escalation of commitment (Staw, 1976; Staw & Fox, 1977; Staw & Ross, 1978): the leadership of Old Era firms grew increasingly committed cognitively to prior strategic actions and capital investments, and so were not likely to make major reversals to these prior commitments. The role of external and internal justification in protecting self-esteem, as well as norms for internal consistency, meant that individuals who had been participants in decisions determining the direction of the organization found it difficult to disentangle themselves from the courses of action set in motion by those decisions (Staw, 1981; Staw & Ross, 1978). So for an Old Era firm facing declining or poor performance, managers were prone to committing new and additional resources to past courses of action in an attempt to ‘‘turn the situation around’’ or to demonstrate the rationality of their plans (Staw, 1981). As a result, the lackluster performance and competitive erosion witnessed in many industries (e.g., the auto industry), instead of serving as a catalyst for change, served only to increase the firm’s commitment to the current strategic direction, orchestrated by the incumbent CEO. Following the change in the market for corporate control, the presence of external threat to the control and ownership of public companies served to break this cycle of escalating commitment to strategic direction. Indeed, corporate raiders and Mergers Acquisitions (M&A) specialists were wholly immune from the cognitive forces that resulted in the firm’s ongoing commitment to the existing strategic direction. Accordingly, when these external constituents began to articulate their analysis of the firm’s competitive position – that is, when they began to participate in the dialogue we describe – the firm was prompted to respond in a way that indicated a willingness to make a dramatic shift from past strategic direction. Given the CEO’s role as architect of that direction, and champion of ongoing commitment to that direction, the removal of this person sends a powerful signal about the firm’s effort to break the pattern of escalating commitment to past strategic direction. Accordingly, in more active periods of takeover activity, researchers have found that boards increasingly dismissed CEOs when performance languished, and a reduction in the degree of takeover activity in the U.S. led to less forced management turnover (Denis & Kruse, 2000; Mikelson & Partch, 1997). Huson, Parrino, and Starks (2001) in their examination of the link between governance context, the market for control, and CEO turnover found that the rate of CEO dismissal increased over time consistent with their theory that ‘‘a more active takeover market strengthens internal control mechanisms’’ (p. 2296). The existence of an active merger and
CEO Turnover in the New Era
153
acquisitions market, then, leads to a contextual change wherein public companies, especially poorly performing firms, face the loss of corporate control. Boards in their attempt to avoid loss of corporate control will increasingly use CEO turnover to signal a forthcoming change in strategic direction to the financial community. Proposition 1. The greater the amount of merger activity in the economy in a given period, the stronger the association will be between (poor) firm performance and executive turnover. Proposition 2. Companies that have been taken over (acquired) with poor pre-takeover performance will exhibit a higher rate of subsequent executive turnover than companies without poor pre-merger performance. In our overview of the Old Era, we suggested that a process of mimetic isomorphism (DiMaggio & Powell, 1983) may have fed upon the firm’s internal escalation of commitment to result in organizational inertia. We might similarly understand a mimetic process during the New Era. The emergence of the market for corporate control not only represented a major threat to management (Jensen, 1993; Morck et al., 1989), but also introduces significant uncertainty over the control, ownership, and direction of public firms and their boards. Organizational search (Cyert & March, 1963) might be understood as a predictable outcome of this uncertainty, and faced with comparable contextual conditions, boards of U.S. corporations predictably looked to each other and copied the behavioral responses – in particular, CEO turnover – they observed. The greater the merger and takeover activity in a particular industry, the greater the uncertainty for firms in that industry, and the more likely firms will be to engage in this process of mimetic isomorphism. Proposition 3. Industries with higher amounts of merger and/or takeover activity will exhibit higher rates of executive turnover. Institutional Investors and Shareholder Activism One of the most significant and enduring changes that characterizes the New Era is the increasing clout, visibility, and activism of large institutional investors. During this time period, institutional investors – pension funds, insurance companies, money managers, and mutual funds – gained considerable power (Useem, Bowman, Myatt, & Irvine, 1996). The total financial assets held by these investors increased from $6,996 billion in 1990 to $15,800 billion in 1997, with equity representing half of their portfolios.
154
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
Increasing concentration in equity holdings served to increase the profile of the institutional investors, who accounted for 60% of all corporate equity holdings by 1997.11 In addition to the increasing ownership concentration of public companies by institutional investors, the presence of large outside blockholders (typically defined as having at least 5% of the firm’s shares) have also emerged. While some companies historically had large blockholders, they were usually affiliated with management, as in the case of family or founder ownership. The presence of large blockholders that are neither affiliated nor aligned with current management were a new phenomenon that changed the investor landscape in the 1990s. The increasing concentration of equity ownership among institutional investors and independent blockholders represented a significant shift in the investor profile of public companies. This contrasts sharply with the Old Era where wider dispersion of stock ownership meant that shareholders had little voice or clout with the management of the firm or its board. Without concentrated ownership, public companies could largely ignore their shareholders. Despite the fiduciary duties of the board of directors to act in the interests of the individual stockowners, boards rarely communicated with the company’s investors nor did they necessarily act in the shareholder’s best interests. Instead, with little accountability for enhancing shareholder wealth, boards acted more as an instrument of management. Dispersed stock ownership in the Old Era meant that shareholders lacked power and importance, and in the language of the dialogue metaphor we employ here, a ‘‘voice’’ that was heard by the firm. Not surprisingly as a result, companies and their boards seldom attended to, or felt the need to be responsive to, the concerns of shareholders and did not engage in a dialogue with shareholders unless of course, the company had significant family holdings. The New Era’s emergence of independent blockholder investors and institutional investors as significant and powerful constituents changed this dynamic. Large institutional investors became more active in their management of their holdings, giving them considerably more clout on Wall Street, and research has shown that the trading activity of institutional investors can significantly impact a company’s share price (Brown & Brooke, 1993). At the same time, when faced with the poor performance of a firm whose stock they hold in their portfolios, the managers of these institutional funds have become more involved in communicating with, and influencing the management of, the firm as opposed to simply selling their position in the company (Del Guercio & Hawkins, 1999; Gillian & Starks, 1998). In particular, more active institutional investors increasingly are having one-on-one meetings with management as well as boards of directors
CEO Turnover in the New Era
155
to discuss the company’s strategy and performance (Conference Board, 1997). While the communication between institutional investors and the firm’s board of directors and top management is private and generally not shared with the public (Pound, 1992), there is sufficient evidence to suggest that these investors due to their large ownership position are able to bring about desired changes at the firms in which they invest (Carleton, Nelsen, & Weisbach, 1998). Not surprisingly, recent studies have found that a change in institutional ownership is a significant predictor of CEO dismissal. When institutional investors sell a firm’s stock, the probability that the CEO will be fired and replaced by an outsider increases significantly (Parrino, Sias, & Starks, 2003). Institutional investor holdings have also been shown to increase subsequent to a CEO firing (Parrino et al., 2003). Similarly, large outside blockholders can often obtain seats on the boards of directors and researchers have shown they can become directly involved with other board members in forcing the resignation of the CEO (Denis & Serrano, 1996). The incidence on non-routine CEO turnover was found to be significantly higher when firms had a large outsider blockholder (Denis, Denis, & Sarin, 1997). Furthermore, they found that for firms with poor performance, nonroutine CEO turnover is more likely to occur when a firm has a large blockholder than when it does not. Institutional theory (DiMaggio & Powell, 1991) can again provide insight into why boards of U.S. corporations increasingly are willing to engage in CEO turnover as part of their dialogue with the institutional investors and large blockholders. We previously described a process of mimetic isomorphism as driving CEO turnover in response to the market for corporate control, and we might identify a similar mechanism here: firms see their peer organizations responding to institutional investor dissatisfaction with CEO turnover, and so adopt a similar response. However, there is also evidence of a process of coercive isomorphism (DiMaggio & Powell, 1991) at work in the organizational response to institutional investors and shareholder activism. The coercive process is defined as resulting from ‘‘formal and informal pressures exerted on organizations by other organizations upon which they are dependent’’ (p. 67). From the foregoing discussion, it is clear that firms have become dependent upon these institutional shareholders and large blockholders: through their trading activities they have the ability to influence a company’s share price. Inasmuch as access to capital through the financial market represents an important dependency for public firms, this important constituency is positioned to exert considerable pressure for adaptation on the firm. Stated differently, this results in the ability of these
156
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
investors to apply coercive pressures as part of their participation in the dialogue with the firm. CEO dismissal, then, becomes an accepted practice in the firm’s quest for legitimacy with these important external constituents. Indeed, the coercive and mimetic processes work in tandem: the external stakeholders press for a change in strategic direction in light of poor performance, and a firm replies by engaging in CEO turnover. Another firm, searching for an appropriate response to its own institutional shareholders’ demands, observes CEO dismissal as an accepted response, and so copies this behavior. Thus, high levels of institutional ownership and/or large independent blockholders are likely to lead to greater sensitivity on the part of the board to performance concerns and their likely response to address these concerns will be to remove the incumbent CEO. When the firm’s industry is faced with higher levels of overall institutional ownership, the mimetic response to replace the CEO is likely to be stronger. Proposition 4. The greater a firm’s institutional ownership, the stronger the association will be between (poor) firm performance and executive turnover. Proposition 5. For firms experiencing a reduction in their institutional ownership, the stronger the association will be between (poor) firm performance and executive turnover. Proposition 6. For firms with large independent blockholders of shares, the stronger the association will be between (poor) firm performance and executive turnover. Proposition 7. The greater the amount of institutional ownership in a firm’s industry, the stronger the association will be between (poor) firm performance and executive turnover. Stock Analyst Coverage One of the major components of the heightened focus on earnings and shareholder wealth maximization for U.S. public firms has been the pressures exerted by Wall Street stock analysts. Analysts are security specialists who review a company and provide a rating on the stock (buy, hold, sell, etc.), thus providing investors with valuable information on the firm’s future prospects for generating shareholder wealth. Analyst coverage refers to the number of security analysts that report or follow a particular firm’s stock. Security analysts through their analysis and coverage of a firm’s financial and strategic performance provide increased scrutiny and monitoring on
CEO Turnover in the New Era
157
firm’s activities and performance. Not surprisingly, analysts’ recommendations and forecasts of the firm’s future earnings have been shown to influence demand for a firm’s stock: when added to the ‘‘buy’’ list, there is a positive impact on stock price, whereas being removed from the ‘‘buy’’ list or being added to the ‘‘sell’’ list, has been shown to have a significant negative impact on stock price (Stickel, 1995; Womack, 1996). When analysts follow a firm, there are increased resources spent on private information acquisition about the firm, thus greater analyst coverage results in more information gathered and disseminated on the firm. Since institutional investors rely on analysts’ reports, this information gathering and analysis are disseminated widely and will also increase investor scrutiny on the firm. During the Old Era, despite the considerable macroeconomic changes in the competitive environment and the associated pressures on firm profitability, financial analysts were reluctant to openly criticize a firm’s management. Indeed, analysts issued ‘‘buy’’ ratings some 90% of the time.12 Perhaps due to the same mimetic process that drives isomorphism among firms during this period, financial analysts during the 1970s and 1980s were largely silent with regard to communicating concerns regarding a firm’s competition position or financial performance. That is, since there was uncertainty in the market place concerning what organizational changes would be most appropriate to respond to the environmental changes, analysts often looked to each other and copied the behavioral responses they observed. The fact that the observed behavior focused on issuing ‘‘buy’’ ratings, was entirely acceptable, since analysts, as employees of investment banks and brokerages, had much to lose by alienating a company’s management: these firms represented potential clients for other services. Thus, despite the dependence of their livelihood on the financial performance of the public firms, which they evaluated and held shares in, there was nothing within the analyst community to foster an atmosphere of criticism. The mimetic process of isomorphism fed upon this to foster ‘‘herd behavior’’ in the investment community (Scharfstein & Stein, 1990), where the analysts did not provide the criticism that could lead to a call for firm adaptation. During the 1990s, analysts became more prominent by serving as the intermediary for highly profitable investment banking deals. In addition, analysts were increasingly rewarded based on the quality of their analysis and they became increasingly visible to the investing public via media exposure (Cole, 2001). The increased visibility and impact of their stock ratings in the New Era is the predominant method of engaging in a dialogue with the firm. Whereas historically the firm’s financial reporting served as the primary communication to the analyst community, in the New Era
158
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
firms’ investor relations campaigns became targeted to analysts (Useem, 1996). As analysts rose in both visibility and power, executives placed increased emphasis on meeting or exceeding analysts’ expectations regarding quarterly earnings (Lopez & Rees, 2002). Indeed, while we have characterized the analysts’ ratings in terms of the metaphorical dialogue we described at the start of this chapter, it is important to note that the increasing prominence and importance of stock analysts has also led quite literally to an open dialogue with the firm’s senior management: most companies now hold quarterly stock analysts calls which are intended to provide information on firm performance, but also allow for analysts to ask questions of the firm’s management. The transcripts of these calls are disseminated as part of brokerage house reports on the companies they evaluate. Taken together, this emergent communication – both metaphorical and literal – between analysts and the management of the firm has led, not surprisingly, to the widespread adoption of ‘‘earnings management’’ among U.S. corporations. Through this process, the firm endeavors to avoid reporting losses and decreases in earnings, and thus not report quarterly earnings per share that are below analysts’ earnings expectations (Burgstahler & Dichev, 1997; DeGeorge, Patel, & Zeckhauser, 1999).13 The practice of earnings management became so prevalent that when James Kilts took over as CEO of Gillette, his decision to not issue quarterly earnings guidance was notable and sent shock waves through the financial community because it was decidedly outside what had become a common practice among public U.S. firms. Because U.S. firms are dependent upon the financial community – and particularly financial analysts – for legitimation, the processes of coercive and mimetic isomorphism (DiMaggio & Powell, 1991) again illuminate our understanding of CEO turnover as a signal of adaptation in the firm’s dialogue with the financial community. Analysts can create pressures for action when performance does not meet expectations, and it is understandable how firms have necessarily adopted a common set of behaviors designed to appease these external constituents. A few examples might serve to illustrate how boards of public firms have become increasingly sensitive and committed to satisfying this important external constituent. At Hewlett-Packard, when CEO Carly Fiorina failed to exact the anticipated performance improvements from the controversial acquisition of Compaq in 2002, pressures from the external financial community were largely responsible for the board’s dismissal of her as CEO (Loomis & Ryan, 2005). The board’s commitment to appease analysts, also led to Douglas Ivestor’s forced resignation as CEO of Coca-Cola less than two years after his appointment in 1997. After a long track record of phenomenal earnings
CEO Turnover in the New Era
159
under former CEO Goizueta, Coca-Cola’s financial performance began to falter in the late 1990s.14 Many of the driving forces (i.e., global economic slowdown and changing consumer taste preferences) that contributed to Coca-Cola’s inability to continue to meet performance expectations were outside of the control of its management. Yet, despite these market realities, the financial community was unforgiving in its evaluation of the firm’s performance in general, and its leadership in particular. So when a public relations disaster in Europe due to tainted product arose, the board decided to take action by forcing CEO Ivestor to take involuntary retirement at the age of 52. The increasing visibility and importance of financial analysts has led to a literal and figurative dialogue wherein boards are increasingly sensitive of, and willing to, appease this important external constituent. When the financial analysts have lost confidence in the CEO’s ability due to repeated missed earnings forecasts, boards respond to financial community pressures in a common way – the dismissal of the CEO (Pourciau, 1993; Wiersema, 2002). As discussed earlier, boards look to each other and copy the behavioral responses they observe. Boards of companies with greater analyst scrutiny (coverage) and those with adverse analysts’ ratings are more likely to remove the incumbent CEO when firm’s performance is poor. Proposition 8. The greater the number of analysts that cover the firm and its industry, the stronger the association will be between (poor) firm performance and executive turnover. Proposition 9. The more negative the average analyst rating of the firm, the stronger the association will be between (poor) firm performance and executive turnover. Proposition 10. The more downgrades in analyst ratings of the firm, the stronger the association will be between (poor) firm performance and executive turnover. Proposition 11. Companies that have undergone acquisitions that have resulted in a subsequent dilution of shareholder value will exhibit higher rates of subsequent executive turnover than companies that have not had undergone acquisitions that have resulted in diluted shareholder value. Corporate Governance Financial community concerns over whether or not a public company is being managed with maximization of shareholder wealth as its primary
160
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
concern has led to a renewed focus on the corporate governance of the firm. It has been acknowledged that firms in the Old Era had little incentives (formal or informal) to be aligned with the interests of the shareholders (Fama & Jensen, 1983; Jensen, 1986). The lack of good internal governance has been identified as one of the key shortcomings from this era. The market for corporate control, analyst coverage, and institutional investors and large blockholders serve as external forces that impose better corporate governance. Alternatively, boards and the management of public companies can also be proactive in developing better corporate governance and oversight. The importance of corporate governance to the investment community has led to the recent development of corporate governance ratings of public firms such as Institutional Shareholder Services’ Corporate Governance Quotient (CSQ) that provides an overall assessment on how well the firm is governed. Such corporate governance ratings include evaluations of boards: their composition and structure; audit committee composition; anti-takeover provisions; executive and directors’ compensation; as well as ownership structure. As a result, most companies now have adopted provisions to improve their CSQ and thus indicate to the financial community their improved corporate governance. One area where boards have taken significant steps is on the issue of executive compensation. As concern over shareholder wealth maximization took hold, boards with the help of compensation specialists have begun to imitate each other in adopting compensation incentive plans to align executive pay with performance through the use of stock options. As heralded by Jensen and Murphy (1990), tying executive pay to stock performance resulted in aligning managerial interests with those of the shareholders. As a result, executive pay exploded as stock incentive pay (e.g., options) became the predominant component of executive compensation during the New Era.15 Overlaying this recent development on the dialogue metaphor we have used throughout this chapter, we might well observe that companies with higher governance ratings would be expected to have boards that are more likely to engage in a dialogue with financial constituents. That is, firms with high corporate governance quotients have effectively already demonstrated that they are attentive and responsive to the financial community’s demands for accountability. Accordingly, we might also expect that they would employ CEO turnover as part of their repertoire when dialoguing with these external constituents. Proposition 12. The higher the corporate governance quotient (score on aspects of corporate governance provided by third parties), the stronger
CEO Turnover in the New Era
161
the association will be between (poor) firm performance and executive turnover.
DISCUSSION AND CONCLUSION During the mid- to late-1980s a significant change occurred in the corporate context facing U.S. firms due to increasing visibility and clout of the financial community. Specifically, the monitoring and disciplinary effects of the market for corporate control, institutional investors, shareholder activism, stock analysts expectations, and corporate governance concerns changed the role and interface of the board of directors and management of public companies with constituents in the financial community. The push for accountability and the focus on shareholder wealth maximization demanded by these constituents represented an environmental punctuation and signified the beginning of a ‘‘New Era’’ for public companies. The changing dynamics led to the emergence of a dialogue between the firm (specifically, its board of directors and senior management) and key external constituents (i.e., the financial community) that has forever changed how public companies perceive and respond to external pressures for change. The increased sensitivity toward and willingness to respond to financial community concerns over the strategic direction and performance of the firm has had a significant impact on the firm’s leadership. Specifically, as the architect of the firm’s strategy and as the interface between the firm and its external constituents, the firm’s CEO has become an increasingly visible target to address performance shortcomings. This is the primary reason why CEO turnover has become one of the most powerful signals the firm can send about its intended future direction and willingness to adapt and respond to external demands for change. Our characterization of the emergence of this dialogue focused on a number of informal – albeit powerful – changes in the visibility and power of the financial community that led to a New Era. It is worthwhile to note in addition to the increasing power and influence of the financial community, a more formalized set of legal and regulatory constraints emerged in the 1990s that changed the governance structure of public companies, their financial reporting requirements, and the judicial interpretation of boards as governance bodies. Coming in the wake of numerous corporate scandals, the U.S. Congress approved The Sarbanes Oxley Act (SOX) in 2002, imposing a major set of new reporting and compliance standards on public firms enforced by the Security Exchange Commission. In particular, the executives
162
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
of the firm became accountable to put into place an internal control structure and procedures that could provide reasonable assurance that the firm’s financial statements fairly present in all material respects the financial condition and results of the firm’s operations. In addition to this greater regulatory oversight, the NYSE and NASDAQ adopted specific rules specifying board composition and structure as well the specific duties and responsibilities of various board committees. The intent was to create a more effective board by making sure that outside directors were truly independent of management, that the audit committee has sufficient accounting expertise, and that the CEO did not control the board. Finally, shareholder suits over board malfeasance have resulted in legal judgments that have served to put boards on notice by redefining what constitutes their fiduciary responsibility to the firm’s shareholders. These formal changes in governance structure, judicial interpretation, and regulatory oversight were intended to bolster financial transparency, restore confidence in the financial community, as well as change the accountability of how public firms were governed and managed. The imposition of SOX and the adoption of new listing requirements would never have occurred in the Old Era when the financial community had little say over the governance of public companies and board and management of public firms were neither receptive or subject to externally induced pressures for improved governance and accountability. Given the significant changes that occurred in the contextual conditions facing U.S. public firms during the past 15 years, researchers might well step back and ask, what are the implications for research on organizations? In this paper, we have highlighted one major consequence – mainly that external demands for performance, and the monitoring and disciplinary effects of the market for corporate control, institutional investors, shareholder activism, stock analysts expectations, and corporate governance ratings have led to the emergence of a dialogue between the firm and key external constituents (i.e., the investment community). In particular, we have advanced a theoretical framework to help direct future research on how and when responding to specific external constituents, boards are more likely to utilize CEO turnover as a signal of organizational adaptation. It is clearly evident from research on CEO turnover in the New Era (Exhibit A) that both the level of CEO turnover and the incidence of CEO dismissal rise dramatically after the mid-1980s. By the mid-1990s, CEO dismissal constitutes 71% of all CEO turnovers in large public firms (Wiersema, 2002). In addition, a significant number of studies find clear evidence that firms with management changes exhibit poor stock price performance prior to turnover (Coughlan
CEO Turnover in the New Era
163
& Schmidt, 1984; Denis & Denis, 1995; Warner, Watts, & Wruck, 1988; Weisbach, 1988), and are more often subject to active monitoring by the board, large blockholders, other shareholders, creditors, and potential acquirers (Denis & Denis, 1995; Martin & McConnell, 1991). What is evident from this research is that the contextual conditions of the New Era can foster boards to use a change in leadership to appease powerful external constituents and their concerns over firm performance. Our description of the New Era and its implications for organizations provides the basis for understanding why CEO turnover has become an increasingly popular activity. The research agenda and propositions we have developed build on how differences in contextual conditions will impact the utilization of CEO turnover as a signal for organizational adaptation. Our characterization of the New Era and the emergence of a dialogue between the firm and the financial community can enhance our understanding of why executives have become increasingly visible and why their replacement sends such a strong signal of adaptation. One particularly interesting extension of the insights we develop in this paper is highlighted by the companion piece in this volume. In their very thoughtful piece, Boone, Wezel, and van Witteloostuijn (2006) argue for integrating the organizational demography and ecology literatures in developing a set of multi-level propositions on organizational diversity. What is particularly interesting with regard to the current chapter is the fact that in many ways their theory picks up where ours leaves off. They suggest that when CEO turnover occurs, the selection of the replacement is potentially related in meaningful ways to the future performance – indeed, the survival – of the firm. Thus were we have endeavored to illuminate the signaling content of the CEO turnover event itself, they suggest the signaling content of the subsequently appointed new CEO. If, as they suggest, ‘‘homosocial reproduction’’ is associated with positive performance and organizational outcomes, then the demographic characteristics of the new (i.e., replacement) CEO should be of considerable interest to the financial community that has, as we have suggested, called for some level of organizational change. Future research, then, might well examine the alignment of the external demands for change and the replacement CEO’s demographic characteristics. While this paper has focused on U.S. based public firms, its relevance is not limited to just this population of companies. Publicly listed firms, regardless of country of origin, are subject to capital market pressures. Thus, European (e.g., Daimler-Benz, Siemens) and Asian (e.g., Sony) firms listed on the U.S. exchanges face the same contextual environment as their U.S.
164
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
counterparts. However, there are significant country differences in the governance and public company ownership (e.g., family ownership, state ownership, bank ownership, board composition, etc.) which in general serves to make companies in these countries less subject to the pressures for shareholder wealth maximization that emanate from the capital market. However, research has shown that as a country’s financial and economic development improves, firms increasingly adopt better governance, and that access to global capital markets sharpens the incentives on the part of the firm to provide better governance despite country-specific differences (Doidge, Karolyi, & Stulz, 2005). Thus non-U.S. firms are not immune from the contextual conditions characterized by the U.S. financial markets. As for the implications on leadership, poor stock performance and poor earnings have been found to lead to higher rates of managerial turnover in both Japan and Germany (Kaplan, 1994). Given the dominance of the U.S. stock exchanges in providing access to global capital, it is not surprising that for those companies that operate in the global capital market the effect has been universal. For non-public firms, the financial community may well play a less significant role. However, even private companies with significant debt-related securities are subject to the same financial reporting requirements as publicly listed companies. Moreover, for companies with aspirations to go public (IPO), the contextual conditions would not differ from that of public firms. Indeed, we might add to the characterization of the environmental punctuation we have described in this paper by observing that a contemporaneous ‘‘boom’’ occurred in the IPO market during this period. Inasmuch as non-public firms feel the effect of this change proportionally more, we suspect that they might be particularly impacted by the dynamics we have described. To be sure, when making its IPO, a firm’s Prospectus and its road show to the investment community represent some of its first forays in establishing a dialogue with the financial community. Thus, for the companies that comprise the majority of economic activity in the world, the contextual conditions that we have depicted are real and relevant, and have had an impact on how organizations are governed and managed.16 We hope that future research might build on the research agenda we have described to explore its implications and applications to non-public firms. We have depicted a New Era wherein contextual conditions have led to the emergence of a dialogue between the board, management, and important external constituents. We propose that one of the major consequences of this dialogue has been the increasing use of CEO turnover as a signal to external constituents that the board is taking action to address performance
CEO Turnover in the New Era
165
shortcoming. Whether or not the replacement of leadership actually results in adaptation on the part of the firm and improved performance remains somewhat unknown. Evidence, thus far, on the organizational consequences of CEO dismissal has not shown it to lead to improved performance (Wiersema, 2002). Given that the lackluster performance is usually not a short-run phenomenon but is indicative of serious maladjustments on the part of the firm to its competitive environment, it will not be easy, even for a new CEO, to rectify the firm’s financial position. Importantly, despite the lack of evidence that CEO turnover leads to improved performance, it has not stopped the practice. Further research is needed to examine whether the increase in CEO turnover witnessed in the New Era has led to a significant improvement in the competition position and performance of these firms. As part of this research, investigators may also want to examine the nature of the replacement, since the individual chosen may well determine how capable and successful they may be in turning around the organization’s competitive position and financial performance. For example, a productive line of inquiry in this vein might examine whether certain replacement CEO attributes yield superior post-succession performance when the firm is undertaking particular adaptations (e.g., adapting to technological vs. demographic changes). While we have focused on one principal outcome of the ramification of a dialogue for firms and their management, broader implications are also possible. One of the key outcomes of the existence of this dialogue is that organizations and their boards have become aware of and began to incorporate external demands for increased accountability. This adoption of a financial performance orientation is perhaps one of the most notable changes in how organizations are managed. While much of the organizational theory literature inherently assumes that organizations and managers are performance motivated, this assumption clearly did not hold in the Old Era when managers were immune from lackluster performance outcomes. Given the opening of a dialogue with external constituents that are largely performance driven, the theoretical assumption of performance as being a key driver for firm action is certainly more valid. The existence of a dialogue, we believe, will also have important implications in future research examining why firms pursue certain strategic actions such as portfolio decisions, expansion decision, divestitures, as well as restructuring decisions. We encourage researchers to examine how the contextual conditions of the New Era might lead organizations to signal other types of strategic actions aside from CEO dismissal to appease the financial community concerns over the firm’s strategic direction and financial performance. In such a way, we
166
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
hope the metaphor of a dialogue between public companies and the financial community might provide new insights and extend prior research on organizations and their management.
NOTES 1. For example, between 1970 and 1994, the ratio of U.S. imports of goods and non-factor services to U.S. GNP, a broad measure of overall import penetration in the U.S. market, rose almost 800% (from 1.6% to 14.3%). This increase in foreignbased competition is due in large part to reductions in trade barriers resulting from both multilateral and various bilateral trade agreements (Congressional Budget Office, 1987; Krueger, 1995). Many economists have documented that economic, political, regulatory, and technological forces radically changed the worldwide competitive environment starting in the early 1970s (Jensen, 1993). 2. The evidence of falling industry profit margins, rationalization of production, pressures for greater intra-plant efficiency, and technological developments all indicate that foreign-based competition significantly intensifies competition at the industry level (Chung, 2001; Domowitz, Hubbard, & Petersen, 1986; Ghosal, 2002; Katics & Petersen, 1995; Tybout, 2001). 3. Financial or investment community is used in this chapter to refer to both investors (e.g., institutions, money managers, insurance companies) as well as financial analysts that provide financial coverage and evaluate publicly listed firms. 4. As many financial economists have documented, the capital markets provided the incentive for an organization to change before losses generated a crisis (Jensen, 1993). 5. Calculations by the authors based on studies from that period. 6. CEO dismissal was usually coded as CEOs who left prior to attaining the age of 65 years. This may actually lead to an overstatement of the incidence of dismissals since many CEOs may have personal reasons for retiring early. 7. All of the ‘‘big three’’ automakers suffered market share erosion at the expense of foreign rivals. General Motors market share went from 45% in 1980 to 35.3% in 1991. 8. Even accounting for inflation, the merger boom of the 1990s dwarfed all other prior merger periods. Historical studies on U.S. merger activity have found takeover activity above 2–3% of GDP is unusual. Historical comparisons indicate that takeover activity in the 1980s is historically high and the activity in the late 1990s is extraordinary – at more than 10% of GDP (Golbe & White, 1988; Holmstrom & Kaplan, 2001). 9. Leveraged Buyouts and the Pot of Gold: Trends, Public Policy, and Case Studies. Economics Division of the Congressional Research Service. U.S. House of Representatives, 1987. 10. As an example, within 18 months after the leveraged buyout of the Beatrice Company by Kohlberg, Kravits, & Roberts (KKR) in 1986 for $6.25 billion, KKR sold off most of the assets (businesses) of Beatrice for $10.8 billion. 11. Institutional ownership of public company stock went from 40% in the 1980s and leveled out at 60% by the late 1990s. Institutional Investment Report, The
167
CEO Turnover in the New Era
Conference Board’s Global Corporate Governance Research Center, New York, NY. 12. Zacks Historical database provides analyst coverage and recommendations. Many studies have found the analyst ratings were biased upwards (see Hong & Kubik, 2003; Womack, 1996). 13. The accounting literature has written extensively about the nature of earnings management (see DeGeorge et al., 1999; Payne & Robb, 2000, as examples). This practice became widespread and was one of the major factors that led to the adoption of Sarbanes–Oxley. Additionally, SEC commission chairman Arthur Levitt publicly has expressed concerns about the use of earnings management to meet Wall Street earnings expectations set by analysts’ forecasts. 14. Coca-Cola’s stock price appreciated by 4000% during the 16 year tenure of prior CEO Roberto Goizueta. 15. Average total CEO compensation for S&P 500 firms was $850,000 in 1970 and reached $14 million in 2000 with base salaries accounting for only 17% of average pay in 2000 (Jensen & Murphy, 2004). 16. While many companies have publicly complained about the cost of abiding to the new governance and transparency requirements of the New Era, very few firms have actually exercised the option of going private. Even for firms that have gone private (St. John, for example), the requirements on their debt necessitate public filings similar to those of a publicly listed firm.
REFERENCES Allen, M. P., & Cohen, S. (1969). Information flows in R and D labs. Administrative Science Quarterly, 1, 12–19. Allen, M. P., & Panian, S. K. (1982). Power, performance, and succession in the large corporation. Administrative Science Quarterly, 27, 538–547. Andrews, K. (1971). The concept of corporate strategy. Homewood, IL: Dow Jones-Irwin. Astley, W. G. (1985). The two ecologies: Population and community perspectives on organizational evolution. Administrative Science Quarterly, 30, 224–241. Barnard, C. (1938). The functions of the executive. Cambridge, MA: Harvard University Press. Barr, P., Stimpert, J., & Huff, A. (1992). Cognitive change, strategic action, and organizational renewal. Strategic Management Journal, 13, 15–32. Beatty, R., & Zajac, E. (1987). CEO change and firm performance in large corporations: Succession effects and manager effect. Strategic Management Journal, 8, 305–318. Black, B. (1992). Institutional investors and corporate governance: The case of institutional voice. Journal of Applied Corporate Finance, 4, 19–32. Boeker, W. (1997). Strategic change: The influence of management characteristics and organizational growth. Academy of Management Journal, 40, 152–170. Boone, C., Wezel, F. C., & van Witteloostuijn, A. (2006). Top management team composition and organizational ecology: A nested hierarchical selection theory of team reproduction. In: J. A. C. Baum, S. D. Dobrev & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (pp. 103–136). Oxford, UK: JAI/Elsevier. Bower, J. L. (1970). Managing the resource allocation process. Boston, MA: Division of Research, Harvard Business School.
168
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
Brady, G., & Helmich, D. (1984). Executive succession. Englewood Cliffs, NJ: Prentice-Hall. Brown, K., & Brooke, B. (1993). Institutional demand and security price pressure. Financial Analysts Journal, 49, 53–63. Burgstahler, D., & Dichev, I. (1997). Earnings management to avoid earnings decreases and losses. Journal of Accounting and Economics, 24, 99–126. Cannella, A., & Lubatkin, M. (1993). Succession as a sociopolitical process: Internal impediments to outsider selection. Academy of Management Journal, 36, 763–793. Carleton, W., Nelson, J., & Weisbach, M. (1998). The influence of institutions on corporate governance through private negotiations: Evidence from TIAA-CREF. Journal of Finance, 4, 1335–1362. Carroll, G. R. (1987). Publish or perish: The organizational ecology of newspaper industries. Greenwich, CT: JAI Press. Chaganti, R., & Sambharya, R. (1987). Strategic orientations and characteristics of upper management. Strategic Management Journal, 8, 393–401. Child, J. (1972). Organization structure, environment, and performance: The role of strategic choice. Sociology, 6, 1–22. Chung, W. (2001). Mode, size, and location of foreign direct investment and industry markups. Journal of Economic Behavior and Organization, 45, 185–211. Cole, B. M. (2001). The pied piers of Wall Street: How analysts sell you down the river. Princeton, NJ: Bloomberg Press. Coughlan, A., & Schmidt, R. (1984). Executive compensation, management turnover, and firm performance: An empirical investigation. Journal of Accounting and Economics, 7, 43–66. Cyert, R. M., & March, J. G. (1963). A behavioral theory of the firm (2nd ed.). Englewood Cliffs, NJ: Blackwell Publishers, Inc. DeGeorge, F., Patel, J., & Zeckhauser, R. (1999). Earnings management to exceed thresholds. The Journal of Business, 72(1), 1–33. Del Guercio, D., & Hawkins, J. (1999). The motivation and impact of pension fund activism. Journal of Financial Economics, 52, 293–340. Denis, D., & Denis, D. (1995). Performance changes following top management dismissals. Journal of Finance, 1, 1029–1057. Denis, D. J., Denis, D. K., & Sarin, A. (1997). Ownership structure and top executive turnover. Journal of Financial Economics, 45, 192–331. Denis, D., & Kruse, T. (2000). Managerial discipline and corporate restructuring following performance declines. Journal of Financial Economics, 55, 391–424. Denis, D., & Serrano, J. (1996). Active investors and management turnover following unsuccessful control contexts. Journal of Financial Economics, 40, 239–266. DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48, 147–160. DiMaggio, P. J., & Powell, W. W. (1991). The iron cage revisited. In: W. W. Powell & P. J. DiMaggio (Eds), The new institutionalism in organizational analysis. Chicago, IL: University of Chicago Press. Doidge, C., Karolyi, G., & Stulz, R. (2005). Why do countries matter so much for corporate governance. Working Paper, Fisher College of Business, Ohio State University. Domowitz, I., Hubbard, R. G., & Petersen, B. C. (1986). Business cycles and the relationship between concentration and price-cost margins. Rand Journal of Economics, 17, 1–17. Dubin, S. (1972). Professional obsolescence. Lexington, MA: Lexington Books, DC Health.
CEO Turnover in the New Era
169
Dutton, J., Fahey, L., & Narayanan, V. (1983). Toward understanding strategic issue diagnosis. Strategic Management Journal, 4, 307–323. Fama, E. F., & Jensen, M. (1983). The separation of ownership and control. Journal of Law and Economics, 26, 235–301. Feldman, D. C. (1989). Socialization, resocialization, and training: Reframing the research agenda. In: I. L. Goldstein & Associates (Eds), Training and development in organizations. San Francisco, CA: Jossey-Bass. Finkelstein, S., & Hambrick, D. (1990). Top-management-team tenure and organizational outcomes: The moderating role of managerial discretion. Administrative Science Quarterly, 35, 484–503. Finkelstein, S., & Hambrick, D. (1996). Strategic leadership: Top executives and their effects on organizations. West, Minneapolis/St. Paul: West Publishing. Fiske, S. T., & Taylor, S. E. (1984). Social cognition. Reading, MA: Addison-Wesley. Fondas, N., & Wiersema, M. F. (1997). Changing of the guard: The influence of CEO socialization on strategic change. Journal of Management Studies, 34, 561–583. Furtado, E., & Karan, V. (1990). Causes, consequences, and shareholder wealth effects of management turnover: A review of the empirical evidence. Financial Management, 19, 60–75. Ghosal, V. (2002). Potential foreign competition in U.S. manufacturing. International Journal of Industrial Organization, 20, 1461–1489. Golbe, D., & White, L. (1988). A time-series analysis of mergers and acquisitions in the U.S. economy. In: A. Auerbach (Ed.), Corporate takeovers: Causes and consequences. Chicago, IL: The University of Chicago Press. Gordon, L., & Pound, J. (1993). Information, ownership structure, and shareholder voting: Evidence from shareholder-sponsored corporate governance proposals. Journal of Finance, 2, 697–718. Govindarajan, V. (1989). Implementing competitive strategies at the business unit level: Implications of matching managers to strategies. Strategic Management Journal, 10, 251–269. Grusky, O. (1961). Corporate size, bureaucratization and managerial succession. American Journal of Sociology, 67, 263–267. Grusky, O. (1963). Managerial succession and organizational effectiveness. American Journal of Sociology, 69, 21–31. Gupta, A. (1984). Contingency linkages between strategy and general manager characteristics: A conceptual examination. Academy of Management Review, 9, 399–412. Gupta, A., & Govindarajan, V. (1984). Business unit strategy, managerial characteristics, and business unit effectiveness at strategy implementation. Academy of Management Journal, 27, 25–42. Hambrick, D., & Mason, P. (1984). Upper echelons: The organization as a reflection of its top managers. Academy of Management Review, 9, 193–206. Hambrick, D. C., Geletkanycz, M., & Fredrickson, J. (1993). Top executive commitment to the status quo: Some tests of its determinants. Strategic Management Journal, 14, 401–418. Hannan, M. T., & Freeman, J. (1989). Organizational ecology. Cambridge, MA: Harvard University Press. Haveman, H. A. (1992). Between a rock and a hard place: Organizational change and performance under conditions of fundamental environmental transformation. Administrative Science Quarterly, 37, 48–75.
170
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
Haveman, H. A., Russo, M. V., & Meyer, A. D. (2001). Organizational environment in flux: The impact of regulatory punctuations on organizational domains, CEO succession, and performance. Organization Science, 12, 253–273. Helmich, D., & Brown, W. (1972). Successor type and organizational change in the corporate enterprise. Administrative Science Quarterly, 17, 371–381. Holmstrom, B., & Kaplan, S. N. (2001). Corporate governance and merger activity in the United States: Making sense of the 1980s and 1990s. Journal of Economic Perspectives, 15, 121–144. Hong, H., & Kubik, J. D. (2003). Analyzing the analysts: Career concerns and biased earnings forecasts. Journal of Finance, 58, 313–351. Huson, M. R., Parrino, R., & Starks, L. (2001). Internal monitoring mechanisms and CEO turnover: A long-term perspective. The Journal of Finance, 56, 2265–2297. Jensen, M. (1986). Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review, 76, 323–339. Jensen, M. (1989). Eclipse of the public corporation. Harvard Business Review, 67, 61–74. Jensen, M. (1991). Corporate control and the politics of finance. Journal of Applied Corporate Finance, 4, 13–33. Jensen, M. (1993). The modern industrial revolution, exit, and the failure of internal control systems. Journal of Finance, 48, 831–880. Jensen, M., & Murphy, K. (1990). CEO Incentives – It’s not how much you pay, but how. Harvard Business Review, 68, 138–153. Jensen, M., & Murphy, K. J. (2004). Remuneration: Where we’ve been, how we got to here, what are the problems, and how to fix them. ECGI Working Paper Series in Finance, July 2004. Kaplan, S. N. (1994). Top executives, turnover, and firm performance in Germany. Journal of Law, Economics, and Organization, 10, 142–160. Katics, M. M., & Petersen, B. C. (1995). The effect of rising import competition on market power: A panel data study of U.S. manufacturing. The Journal of Industrial Economics, 42, 277–286. Katz, R. (1982). The effects of group longevity on project communication and performance. Administrative Science Quarterly, 27, 81–104. Keats, B. W., & Hitt, M. A. (1988). A causal model of linkages among environmental dimensions, macro organizational characteristics, and performance. Academy of Management Journal, 31, 570–598. Kelly, D., & Amburgey, T. L. (1991). Organizational inertia and momentum: A dynamic model of strategic change. Academy of Management Journal, 34, 591–613. Krueger, A. (1995). US trade policy and the GATT review. The World Economy: Global Trade Policy, 18, 65–79. Lawrence, P., & Lorsch, J. (1967). Organization and its environment. Cambridge, MA: Harvard University Press. Levinthal, D. (1994). Surviving schumpeterian environments: An evolutionary perspective. In: J. A. C. Baum & J. V. Singh (Eds), Evolutionary dynamics of organizations (pp. 167–178). New York: Oxford Univeristy Press. Levinthal, D., & March, J. G. (1981). A model of adaptive organizational search. Journal of Economic Behavior & Organization, 2, 307–333. Lieberson, S., & O’Connor, J. (1979). Leadership and organization performance: A study of large corporations. American Sociological Review, 37, 117–130. Loomis, C. J., & Ryan, O. (2005). Fortune (Europe), 3/7/2005, 151(4), 69–71.
CEO Turnover in the New Era
171
Lopez, T. J., & Rees, L. (2002). The effect of beating and missing analysts’ forecasts on the information content of unexpected earnings. Journal of Accounting, Auditing & Finance, 17, 155–184. Louis, M. R. (1980). Surprise and sense making: What newcomers experience in entering unfamiliar organizational settings. Administrative Science Quarterly, 25, 226–251. Lyles, M. A. (1981). Formulating strategic problems: Empirical analysis and model development. Strategic Management Journal, 2, 61–76. March, J. G., & Simon, H. (1958). Organizations. New York, NY: Wiley. Martin, K., & McConnell, J. (1991). Corporate performance, corporate takeovers, and management turnover. Journal of Finance, 46, 671–687. McCall, M. W., & Kalpan, R. E. (1985). Whatever it takes: Decision makers at work. Englewood Cliffs: Prentice-Hall. Meyer, A. D. (1982). Adapting to environmental jolts. Administrative Science Quarterly, 27, 515–537. Meyer, A. D., Brooks, G. R., & Goes, J. B. (1990). Environmental jolts and industry revolutions: Organizational responses to discontinuous change. Strategic Management Journal, 11, 93–110. Mikelson, W. H., & Partch, M. M. (1997). The decline of takeovers and disciplinary managerial turnover. Journal of Financial Economics, 44, 205–228. Miles, R. E., & Snow, C. C. (1978). Organizational strategy, structure, and process. New York, NY: McGraw-Hill. Miller, D. (1991). Stale in the saddle: CEO tenure and the match between organization and environment. Management Science, 37, 34–52. Mitchell, M., & Lehn, K. (1990). Do bad bidders become good targets? Journal of Political Economy, 98, 372–398. Morck, R., Shleifer, A., & Vishny, R. (1989). Alternative mechanisms for corporate control. The American Economic Review, 79, 842–852. Parrino, R. (1997). CEO turnover and outside succession: A cross-sectional analysis. Journal of Financial Economics, 46, 165–197. Parrino, R., Sias, R. W., & Starks, L. T. (2003). Voting with their feet: Institutional ownership changes around forced CEO turnover. Journal of Financial Economics, 68, 3–46. Payne, J., & Robb, S. (2000). Earnings management: The effect of ex ante earnings expectations. Journal of Accounting, Auditing & Finance, 15(4), 371–392. Pelz, D., & Andrews, F. (1966). Scientists in organizations. New York, NY: Wiley. Pfeffer, J., & Salancik, G. (1978). The external control of organizations. New York, NY: Harper & Row. Porter, M. (1980). Competitive strategy. New York, NY: Free Press. Pound, J. (1992). Raiders, targets, and politics: The history and future of American corporate control. Journal of Applied Corporate Finance, 4, 6–18. Pourciau, S. (1993). Earnings management and non-routine executive changes. Journal of Accounting and Economics, 16, 317–336. Priem, R. L. (1994). Executive judgment, organizational congruence, and firm performance. Organization Science, 5, 421–438. Quinn, J. B. (1980). Managing strategic change. Sloan Management Review, 21, 3–20. Reger, R. K., & Huff, A. S. (1993). Strategic groups: A cognitive perspective. Strategic Management Journal, 14, 103–124.
172
MARGARETHE F. WIERSEMA AND THOMAS P. MOLITERNO
Romanelli, E. (1989). Organizational birth and population variety: A community perspective on origins. In: L. Cummings & B. Staw (Eds), Research in organizational behavior (pp. 35–65). Greenwich, CT: JAI Press. Romanelli, E., & Tushman, M. (1988). Executive leadership and organizational outcomes: An evolutionary perspective. In: D. Hambrick (Ed.), The executive effect: Concepts and methods for studying top management (pp. 211–247). Greenwich, CT: JAI Press. Scharfstein, D. S., & Stein, J. C. (1990). Herd behavior and investment. The American Economic Review, 80, 165–479. Schein, E. (1968). Organizational socialization and the profession of management. Industrial Management Review, 9, 1–16. Schwenk, C. R. (1988). The cognitive perspective on strategic decision making. Journal of Management Studies, 25(1), 41–55. Selznick, P. (1957). Leadership in administration. Evanston, IL: Harper & Row. Sims, H. P., Jr., & Gioia, D. A. (1986). Cognition-behavior connections: Attribution and verbal behavior in leader-subordinate interactions. Organizational Behavior and Human Decision Processes, 37, 197–230. Snow, C. C., & Hrebiniak, L. G. (1980). Industry differences in environmental uncertainty and organizational characteristics related to uncertainty. Academy of Management Journal, 23, 750–760. Spence, A. M. (1974). Market signaling. Cambridge, MA: Harvard University Press. Starbuck, W., & Milliken, F. (1988). Executives perceptual filters: What they notice and how they make sense. In: D. Hambrick (Ed.), The executive effect: Concepts and methods for studying top management (pp. 35–65). Greenwich, CT: JAI Press. Staw, B. M. (1976). Knee-deep in the big muddy: A study of escalating commitment to a chosen course of action. Organizational Behavior and Human Performance, 16, 27–44. Staw, B. M. (1981). The escalation of commitment to a course of action. Academy of Management Review, 6(4), 577–587. Staw, B. M., & Fox, F. V. (1977). Some determinants of commitment to a previously chosen course of action. Human Relations, 30, 431–450. Staw, B. M., & Ross, J. (1978). Commitment to policy decision: A multi-theoretical perspective. Administrative Science Quarterly, 23, 40–64. Stickel, S. E. (1995). The anatomy of the performance of buy and sell recommendations. Financial Analysts Journal, 51, 24–39. Thomas, A. S., Litschert, R. J., & Ramaswamy, K. (1991). The performance impact of strategy manager coalignment: An empirical examination. Strategic Management Journal, 12, 509–523. Thompson, J. (1967). Organizations in action. New York, NY: McGraw-Hill. Tushman, M., & Romanelli, E. (1985). Organizational evolution: A metamorphosis model of convergence and reorientation. Research in Organizational Behavior, 7, 171–222. Tushman, M., Virany, B., & Romanelli, E. (1985). Executive succession, strategy reorientation, and organizational evolution. Technology in Society, 7, 297–314. Tybout, J. R. (2001). Plant and firm-level evidence on ‘‘new’’ trade theories. Working Paper no. 8418. National Bureau of Economic Research. Useem, M. (1996). Investor capitalism: How money managers are changing the face of corporate America. New York, NY: Basic Books. Useem, M., Bowman, E., Myatt, J., & Irvine, C. (1996). U.S. institutional investors look at corporate governance in the 1990s. European Management Journal, 11, 175–189.
CEO Turnover in the New Era
173
Van Maanen, J. (1978). People processing: Strategies of organizational socialization. Organizational Dynamics, 7, 19–36. Van Maanen, J., & Schein, E. H. (1979). Toward a theory of organizational socialization. Research in Organizational Behavior, 1, 209–264. Vancil, R. F. (1987). Passing the baton: Managing the process of CEO succession. New York, NY: Harvard Business School Press. Walsh, J. P., & Kosnik, R. D. (1993). Corporate raiders and their disciplinary role in the market for corporate control. Academy of Management Journal, 36, 671–700. Warner, J., Watts, R., & Wruck, K. (1988). Stock prices and top management changes. Journal of Financial Economics, 20, 461–492. Weisbach, M. S. (1988). Outside directors and CEO turnover. Journal of Financial Economics, 20, 431–459. Wiersema, M. F. (1992). Strategic consequences of executive succession within diversified firms. Journal of Management Studies, 29, 72–94. Wiersema, M. F. (2002). Holes at the top: Why CEO firings backfire. Harvard Business Review, 80, 70–77. Wiersema, M. F., & Bantel, K. (1993). Top management team turnover as an adaptation mechanism: The role of the environment. Strategic Management Journal, 14, 485–504. Womack, K. L. (1996). Do brokerage analysts’ recommendations have investment value? Journal of Finance, 51, 137–167.
This page intentionally left blank
174
PART III: ORGANIZATIONAL CHANGE
175
176
ECOLOGY, STRATEGY AND ORGANIZATIONAL CHANGE Jitendra V. Singh ABSTRACT One area in which strategy and organizational ecology converge is organizational change. This essay weaves together salient themes in my (and my co-authors’) various writings on organizational change, and is anchored in the research literature of the last twenty years. Among other ideas developed here, I point out that there is now a convergence of agendas in strategy and ecology, with an important role being played by intraorganizational ecology. I develop the distinction between strong and weak selection approaches to organizational ecology. While the strong selection view does not find empirical support, there is stronger support for the weak selection view. I lay out some key features of an emerging evolutionary synthesis for the study of strategy and organization, and develop an evolutionary approach to organizational change.
INTRODUCTION Since their original paper outlining an initial research agenda was published by Hannan and Freeman (1977), organizational ecology has proved to be perhaps the most fecund of several competing theoretical perspectives that Ecology and Strategy Advances in Strategic Management, Volume 23, 177–214 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23006-7
177
178
JITENDRA V. SINGH
originated at about that time. A large number of empirical studies have been published, and several books – both edited compilations and original works – have appeared. Some initial theoretical positions have been pruned or modified, new theoretical arguments have been proposed, and, as the evidence suggests, the research program has thrived (see Baum & Shipilov, forthcoming, 2006, for a recent review and related chapters by Crossan & Hurst, 2006 and Boone & van Witteloostuijn, 2006 in this volume). However, the broader influence of organizational ecology in the contiguous field of strategy and their mutual relationship has, in some ways, been juxtaposed to this success. In the early years, the stereotypical argument made by some strategists was, to paraphrase: ‘‘(population) ecology denies strategic choice by managers and does not admit the possibility of organizational change; therefore, ecological thinking is fundamentally incompatible with strategy.’’ As I will argue below, this initial view articulated by some strategy researchers was overstated and inaccurate. It is also worth noting that even at the height of such criticism an extant minority opinion suggested the inherent complementarity of strategy and ecology (Burgelman & Singh, 1987) and called for an integration of their research agendas. Fortunately, almost two decades on, this position has largely been modified and there is a greater convergence of agendas between organizational ecology and strategy. While this convergence is multifaceted, one important area in which the two perspectives have come together is organizational change.1 In this essay, I will focus on organizational change. While I will review below some key recent works that have addressed organizational change, this is not intended primarily as a review essay. Instead, what follows below is a weaving together of my (and my co-authors’) writings on some salient issues related to organizational change, anchored in over the last twenty years of research and framed in the broader context of the convergence of ecology and strategy. Put differently, the underlying aesthetic metaphor is much less that of a single melody than a collage. I aim in this essay to raise some new questions, energize some provocative discussion, and more generally add to the emerging rapprochement between organizational ecology and strategy. This essay is organized as follows: the next section of this paper focuses on the shifting role of organizational change within organizational ecology. Whereas the focus in Hannan and Freeman (1977) was on an absolute notion of inertia, and the core logic asserted that organizations were largely inert; Hannan and Freeman (1984) shifted the emphasis to relative inertia. Whereas organizational ecology started with a primary focus on populations of organizations, over time intraorganizational ecology has emerged
Ecology, Strategy and Organizational Change
179
as another important focus (Baum & Singh, 1994a, 1994b). This permits a quite natural interest in questions of change in individual organizations (see, for example, Singh, House, & Tucker, 1986a, 1986b; Tucker, Singh, & Meinhard, 1990, 1991). The third section presents a brief, selective review of the recent empirical evidence on change in individual organizations and change in organizational forms, although there is considerably less research on the latter. Building upon Baum and Shipilov (forthcoming, 2006), I summarize the main findings related to two main predictions of structural inertia theory (Hannan & Freeman, 1984): the age and size dependence of rates of organizational change and the consequences of organizational change for survival rates. This section concludes with a review of some theory and empirical evidence on the creation of new organizational forms, how they change over time and their extinction. In the fourth section, I emphasize the distinction between strong and weak selection arguments for how change takes place in organizational populations, an argument I had separately made earlier (Bryce & Singh, 2001). The strong selection argument, much in the spirit of Hannan and Freeman (1977), posits that since organizations are inert, change in organizational populations takes place primarily through selection processes; the weak selection argument takes the view that, as a general matter, although selection forces clearly shape organizational populations, so too does individual organizational change. The burden of the empirical evidence of a couple of decades seems clear: a large number of empirical studies do not support the strong selection view, based on which evidence this position should be ruled out; the weak selection view, however, finds much stronger empirical support. The fifth section builds upon Bryce and Singh (2001) and presents an overview of an emerging evolutionary synthesis in the study of strategy and organization. I lay out the main features of this synthesis, and specify several key shifts in the research literature which characterize it. I also note that Hannan and his colleagues (Carroll, Polos, and others) have recently taken an identity-based interest in organizational forms, a more genetic view of organizational forms compared with the more structural view adopted earlier. Moreover, perhaps echoing the success of the weak selection view over strong selection, they are now explicitly interested in individual organizational change. I conclude this section with the broad-scoped claim that most leading theories of organization can be reframed as evolutionary theories that specify the different roles that selection forces play in the evolutionary trajectories of organizational populations.
180
JITENDRA V. SINGH
In the sixth section, I start from some features of the evolutionary synthesis in the previous section and raise some questions oriented to an evolutionary approach to organizational change. In doing so, I connect with earlier work on the quantum, gestalt and configuration views of strategy, structure, and organizational transition (Mintzberg, 1981, 1992; Miller & Friesen, 1984) as well as the punctuated equilibrium approach to organizational change (Tushman & Romanelli, 1985). The next section continues along a similar vein and raises several open questions related to organizational change that deserve further research. The penultimate section, building on earlier developments in the paper, suggests an emerging synthesis between organizational ecology and organizational strategy, with a crucial role being played by the study of organizational change at the intraorganizational level. Burgelman and Singh (1987) had first made this claim almost two decades ago, but, in hindsight, that idea may have come well before its time, and well before the field was ready to embrace it. Interestingly, in the years since then, that is pretty much where the research literature has ended up. Finally, I summarize some of the main ideas and conclude this paper.
ORGANIZATIONAL ECOLOGY AND ORGANIZATIONAL CHANGE Prior to the publication of Hannan and Freeman’s (1977) paper ‘‘The population ecology of organizations,’’ the dominant perspective in the study of organizations was an adaptation view. The typical argument went thus: organizations are led by leaders, or dominant coalitions, who scan the environment for threats and opportunities, consider simultaneously the strengths and weaknesses of their organizations, and, based on their analyses, chart a path for their firms that leverages its strengths to exploit opportunities. This was almost an unquestioned article of faith for the mainstream of organization theory at the time. The provocation, and, to no small extent, the eventual influence of Hannan and Freeman (1977) lay in challenging this article of faith head on. Relatedly, I have always been intrigued by why organizational ecology has had so much influence, both among its supporters and critics, who, after all, have expended considerable energy attacking these ideas. I believe Murray Davis’ (1971) provocative essay on why some ideas are viewed as more interesting than others provides an initial answer. Davis argued that ideas
Ecology, Strategy and Organizational Change
181
that successfully challenge weakly held assumptions are viewed as interesting; ideas that challenge strongly held assumptions generally get ignored and even labeled irrelevant. Relatedly, one might ask if the adaptation perspective was a weakly or strongly held assumption among organization theorists. I think a credible case can be made that adaptation was a weakly held assumption. Unlike a field like Microeconomics, with its strongly held assumptions about optimization and rationality in human decision-making, organization theory has (and, perhaps, did so even more strongly during the 1970s) a less developed paradigm, with fewer strongly held assumptions. Viewed in this light, the success of organizational ecology likely reflects how it challenged successfully the weakly held adaptation assumption in the field, and, hence, was seen by many as interesting for that reason.2 Hannan and Freeman (1977) argued that, as a general matter, organizations are inert for a variety of internal and external reasons, and it is difficult for them to change radically.3 Consequently, there is reason to believe that change in populations of organizations occurs less through adaptive change in individual organizations and more through selection processes. Specifically, selection processes shape organizational populations through the growth and decline of organizational forms themselves and through the differential entries and exits by individual organizations of a specific form. Broadly speaking, this initial logic went: inertia, therefore selection.4 Even as critics of Hannan and Freeman (1977) took up arms against their initial position, the authors modified and refined their theory in Hannan and Freeman (1984). In an important departure, their argument went from an absolute to a relative notion of inertia. Whereas absolute inertia denied organizational change, relative inertia examined the rate of change in organizational features relative to the rate of change in the environment. Thus, an organization merely making incremental changes in a static environment could be relatively plastic or adaptive, while an organization changing significantly in an even more rapidly changing environment could be relatively inert. This shift to relative inertia accommodated organizational change, while still maintaining the foundations for the operation of selection processes. As long as organizations were relatively inert, change in organizational populations would still occur more because of selection rather than adaptation processes. The other significant shift by Hannan and Freeman (1984) was a shift in the logic of their argument: inertia was the consequence of, not the causal precursor to selection processes.5 In modern societies, they argued, selection forces favor organizations with reproducible structures. This reproducibility
182
JITENDRA V. SINGH
of organizations comes, in part, from reliability and accountability, both of which are favored by selection forces. Reproducibility is usually accompanied by high levels of relative inertia. This modified argument went: selection, therefore inertia.6 One interesting, though odd, omission has been that some critics of organizational ecology seem to not have progressed beyond the original argument (Hannan & Freeman, 1977). It is almost as if, having taken a strong position against the original statement, they were not obligated to and, indeed, did not, read the later modified theoretical statements. One unfortunate historical consequence of this inattention is an inadequately convergent conversation between ecology and its critics, some of whom were strategists (for example, see Astley & Van de Ven, 1983). To be fair, however, at roughly the same time, several research papers were published broadly within the organizational ecology framework that dealt quite directly with organizational change. Singh et al. (1986a), in a study of the processes underlying the liability of newness, examined the impact of internal organizational changes on mortality rates of voluntary social service organizations and found that the lack of external legitimacy was one main reason for the high mortality of young organizations. In a companion study, Singh et al. (1986b) tested competing theoretical perspectives on the impact of organizational change on mortality rates. Tucker, Singh, and Meinhard (1990, 1991) studied rates of change directly and found empirical support for an imprinting argument: founding conditions influenced change rates in various features of voluntary social service organizations.7 As I will point out in the section below, while such studies were unusual, even rare, twenty years ago, the literature has since moved in such a direction that studies of organizational change are not as rare any more; indeed, they have come to occupy an important niche in the research literature. Next, I want to turn to some themes that have emerged in the recent ecological literature on organizational change. Before turning to that question, however, I want to make note of one significant development. The initial interest in organizational ecology focused primarily at the population level (Hannan & Freeman, 1977), with some interest in the community level at which populations interact with each other. Over time, this interest broadened to the organizational level, with the emphasis on the study of entry and exit rates in density dependence theory (Hannan & Carroll, 1992). However, while such organizational demography-related questions were studied at the organizational level, the overarching interest was still in population change. On the other hand, studying questions at the intraorganizational level, as
Ecology, Strategy and Organizational Change
183
evinced by an interest in individual organizational change, was historically not present within organizational ecology. Yet, in an early departure from the mainstream of organizational ecology, Singh and Lumsden (1990, pp. 179–182), in a review essay on organizational ecology, devoted attention to various theoretical points of view on organizational change and their empirical tests. In the years since, intraorganizational ecology had emerged as a distinct and important level of analysis in ecological inquiry, and was integrally built into the framework of Baum and Singh’s (1994) edited book, which focused simultaneously on the intraorganizational, organizational, population, and community levels of analysis. Burgelman (1991, 1994, 1996, 2002) has been, and continues to be, one of the more articulate writers adopting an evolutionary perspective at this intraorganizational level, with his interest in the intraorganizational ecology within which complex strategic decisions get made in large corporations.8 Since then, Baum (2002) has recently edited an ambitious volume, Companion to Organizations, which uses, in part, Baum and Singh’s (1994a, 1994b) organizing device of separating the intraorganizational, organizational, and interorganizational levels of analysis, with a final section devoted to epistemology and research methods. Two chapters from that book relevant to this discussion focus respectively on intraorganizational ecology (Galunic & Weeks, 2002) and intraorganizational evolution (Warglien, 2002). Baum and Rao (2004) have elaborated the Baum and Singh (1994a, 1994b) framework of ecological and genealogical hierarchies by integrating variation, selection, and retention into their ideas. Intraorganizational ecology has emerged as a thriving new level of analysis lending itself to ecological inquiry. As I will argue toward the end of this essay, this is most felicitous, because this linkage provides one natural causeway between the literatures in organizational ecology and organizational strategy.
ECOLOGICAL LITERATURE ON ORGANIZATIONAL CHANGE: A BRIEF REVIEW9 As I pointed out in the previous section, the relative emphasis on organizational change in the ecological literature has shifted over time. At the broadest level, here two themes are important: the first deals with organizational forms, and the second focuses on individual organizations of a specific form. It is noteworthy that relatively little theoretical and empirical work has dealt with change in organizational forms (but see Lumsden &
184
JITENDRA V. SINGH
Singh, 1990; Ruef, 2000, 2004). Nevertheless, though it started as a trickle to begin with (see, for example, Singh, Tucker, & Meinhard, 1991), there are by now a large number of empirical studies of individual organizational change. An earlier review paper summarizing research on organizational change (Barnett & Carroll, 1995) had made the distinction between content (what changes) and process (how it changes) changes in organizations and pointed to the promise of studies that emphasized both content and process. Recently, a comprehensive review of organizational ecology (Baum & Shipilov, forthcoming, 2006) has surveyed the empirical literature on organizational change and offered several broad conclusions. In this section, I will build upon Baum and Shipilov’s review and highlight some key issues related to both these themes. A good starting point to delve into the ecological literature on organizational change is structural inertia theory, articulated initially by Hannan and Freeman (1984). Whereas organization theorists have traditionally focused on content of organizational changes (see, for example, Miller & Friesen, 1984), structural inertia theory focuses on the process of organizational change and asks two main questions: how changeable are organizations and does change benefit organizational survival?10 Baum and Shipilov lay out pictorially (refer to their Fig. 1) the main hypotheses of structural inertia theory, underscoring the complexity of the arguments. Clearly, the management theory literature has come a long way from taking the adaptive properties of organizational change for granted, and this contribution has significantly been enabled by developments in organizational ecology. Hannan and Freeman (1984) refer in structural inertia theory primarily to core structural elements: the goals, forms of authority, core technologies, and market strategies of organizations. Compared to peripheral structural features – number and size of subunits, number of hierarchical levels, spans of control, communication patterns, and buffering mechanisms – core features have higher structural inertia. Inertia here is defined in relative terms: when the rate of change in core features is lower than the rate of environmental change, organizations are relatively inert. However, very few studies have explored the notion of relative inertia empirically (see Ruef, 1997, for a notable exception); most studies examine absolute rates of change. Some key predictions derived from structural inertia theory state that rates of change in core structural features of organizations decline with organizational age and size. Baum and Shipilov review over 41 empirical studies ranging across a number of different populations (business periodicals, semiconductor producers,
Ecology, Strategy and Organizational Change
185
airlines, voluntary social service organizations, day care centers, wineries, bank-holding companies, radio stations, hospitals, automanufacturers, to name just a few) and find mixed results from these studies for the age and size dependence of rates of organizational change. They conclude that these studies do not uniformly support the predictions of structural inertia theory. Further, invoking the core–periphery distinction suggested as a possible explanation by Singh and Lumsden (1990) for the discrepant findings does not appear to help much in clarifying these results. Yet other theoretical arguments suggest that rates of change may increase with age. Baum and Shipilov conclude that this fluidity of aging argument (Singh, Tucker, & Meinhard, 1988) is supported by the empirical evidence about as often as the rigidity of aging thesis coming from structural inertia theory. A different series of papers focuses on repetitive momentum, which refers to how an organization’s competency at executing a specific change increases with the number of times it has engaged in that change earlier, which, in turn, increases the propensity to engage in that change (Amburgey & Kelly, 1985; Amburgey, Kelly, & Barnett, 1993; Amburgey & Miner, 1992; Kelly & Amburgey, 1991). Baum and Shipilov conclude that there is strong empirical support for repetitive momentum. They note that studies that control for prior changes often support negative age dependence of change rates, supportive of structural inertia theory, and suggest, quite plausibly, that fluidity of aging might reflect accumulated experience with specific changes rather than true positive age dependence. Baum and Shipilov also review the empirical evidence for tests of predictions from structural inertia theory about the consequences of organizational change for survival rates. While the detailed argument is nuanced, the main prediction is that core changes are disruptive (see, for example, Dobrev, Kim, & Carroll, 2003) in the short run, although, if the organization manages to overcome the hazards associated with this disruption, change may sometimes prove to be adaptive in the long run. In reviewing 21 different empirical studies, Baum and Shipilov conclude that the results are mixed. As a general matter, while organizations do not necessarily fail due to these changes, they do not much improve their survival chances either. Despite the large number of studies that have already been executed, they suggest that it is premature to draw final conclusions, particularly in light of several methodological complexities such as left and right censoring, the usual absence in such studies of measures of ongoing organizational performance, the unmeasured variation in organizational susceptibility to risks of change, and the lack of attention to within change-type differences and
186
JITENDRA V. SINGH
their survival implications. Notably, on this last point, Baum and Singh (1996) had found that the effects of market domain changes, both expansion and contraction, on the failure rate in a population of day care centers depended crucially on how the changes influenced the intensity of competition. If changes lowered the intensity of competition, survival chances improved, but if changes increased the intensity of competition, survival chances were lowered. To summarize, Baum and Shipilov’s review suggests several broad conclusions. Most of the empirical research has examined absolute rates of change in organizations as opposed to relative inertia, with few exceptions. The pattern of results from several quite diverse populations suggests that organizations change quite frequently. As I will discuss in the next section below, this has significant implications for the initial position taken by Hannan and Freeman (1977) about the importance of adaptation vs. selection in explaining change in organizational populations. Despite strong evidence of these frequent changes, however, the findings are mixed both for the age and size dependence of change rates and the survival consequences of organizational change. There does not seem to be strong, uniform support for the predictions from structural inertia theory, although there is some evidence that if organizational changes reduce competitive intensity, they result in better survival chances, and organizational changes that increase competitive intensity result in lowered survival chances. However, there appears to be strong support for the existence of repetitive momentum in processes of change, broadly consistent with learning ideas. While the final resolution of these questions may have to await further research, it appears that the research findings rule out elegant, parsimonious conclusions like strong support for structural inertia theory or strong evidence against it. Indeed, the empirical evidence seems broadly consistent with views anchored in the bounded rationality of decision makers, which suggest that while organizational changes take place routinely and continuously, their consequences often are not as intended, sometimes resulting in surprises (March, 1981; Starbuck, 1983). As mentioned above, another theme concerns organizational forms, the processes of their creation, their change over time, and their ultimate demise (see Singh & Lumsden, 1990, p. 189, for one early statement). One might even argue that this was an important question central in spirit to Hannan and Freeman (1977), yet it has received relatively scant attention in the research literature. There were, of course, some exceptions to this general pattern. Lumsden and Singh (1990) theorized about organizational speciation, the emergence of the first instance of a new organizational form, as a
Ecology, Strategy and Organizational Change
187
process of search by entrepreneurs in an opportunity space. Zucker (1989, p. 542, footnote 2; also see Singh & Lumsden, 1990, pp. 188–189), citing the example of the fur industry in the Netherlands, argued that the decline in the number of organizations of a particular form may reflect not just increasing competition as posited by density dependence theory (Hannan & Carroll, 1992), but decreasing legitimacy. Implicit in her argument was a particular view of how organizational forms may decline, eventually leading to the extinction of an organizational population. Perhaps the recent best work on these questions has been done by Ruef (2000, 2004). Ruef (2000) took a community ecology approach to the emergence of organizational forms in the health care sector and examined identity positioning in the context of the population. The results showed that the probability of form emergence increased up to a point with the density and size of organizations with similar identities (cross-form legitimation) but that more saturated regions of the identity space were less inviting to new form creation (cross-form competition). Ruef (2004) studied the disappearance of the Southern plantation in the years following the American Civil War using a perspective based in organizational ecology and the social movement literature. While there was limited support for exogenous explanations of plantation demise, ecological dynamics played a greater role in the form of challenges from alternative forms of labor organization and interdependence with mid-size farms as well as the decisions made by laborers with respect to incentive structures and reconstruction of their social networks. Further investigation of organizational forms continues to be an area of great promise for future research efforts.
STRONG SELECTION VS. WEAK SELECTION The discussion of the two previous sections – the role of organizational change within organizational ecology and a brief review of literature on organizational change – helps reinforce a quite important theoretical distinction between strong selection vs. weak selection arguments for how change takes place in organizational populations. Although this distinction has earlier been made elsewhere (see Bryce & Singh, 2001, p. 164), here I will develop the idea more fully. The strong selection argument, the best instantiation of which is Hannan and Freeman (1977), takes the position that, as a general matter, due to a variety of internal and external factors that contribute to inertia, organizations do not change. Consequently, change in organizational populations
188
JITENDRA V. SINGH
occurs primarily through selection processes. This argument applies both to individual organizations and to organizational forms. Thus, organizational forms – for example, Southern plantations – are created within a specific sociocultural or socioeconomic milieu, and, over time, while the number of organizations of the specific form grow or decline in abundance, due to inertial forces, the organizational form does not undergo significant change. In time, as contextual conditions change, selection forces lead to the decline of the organizational form and, ultimately, its extinction. The weak selection argument, on the other hand, takes the view that while selection forces clearly shape organizational populations though a number of processes, so too does individual organizational change. There is a parallel process at the level of the organizational form itself, by which organizational forms also change over time in response to contextual pressures, or, alternatively, are shaped both by selection and adaptation forces. Indeed, at a general level, change in organizational populations occurs both through change in individual organizations and through the operation of selection forces on individual organizations (manifested mainly in differential entry and exit rates of organizations), but also through change in organizational forms and through the operation of selection forces on organizational forms themselves (manifested in differential creation and extinction rates of organizational forms), although, arguably, these latter rates may be more modest in magnitude than individual organizational change. More significant than this theoretical distinction, however, is an empirical question: what position is better supported by the overall body of empirical evidence? As outlined in the previous section, I believe the answer is rather clear, especially at the level of individual organizations. The burden of the empirical evidence is that the strong selection view is not supported by research.11 While this is amply clear at the level of individual organizations, there is some evidence that this may also be the case at the level of organizational forms. However, while the modest evidence that does exist is broadly consistent with this view, we need the accumulation of more evidence to arrive at a more definitive position. On the other hand, the overwhelming burden of the empirical evidence suggests much stronger support for the weak selection argument. While it is clear that selection forces operate both at the level of individual organizations (ample evidence) and organizational forms (modest evidence), it is equally clear that organizations change all the time and that these changes can have significant survival implications.12 There is some evidence that organizational forms themselves are subject to forces of change (Ruef,
Ecology, Strategy and Organizational Change
189
2004), although it is likely that, in general, this is a story in which adaptation and selection will be difficult to fully sort out. Whereas the strong selection view poses adaptation and selection as competing views, the weak selection view sees the fundamental complementarity of adaptation and selection. Indeed, here the foundations lie in the sometimes intricate ways in which selection and adaptation processes jointly shape the trajectories of organizational evolution (Baum & Singh, 1994a, 1994b; Bryce & Singh, 2001; Baum & Shipilov, forthcoming, 2006; Levinthal, 1991; March, 1991; Singh et al., 1986a, 1986b; Singh & Lumsden, 1990). Several implications flow from the conclusion above. One, I would hope that we can finally rule out the strong selection version of organizational ecology, which has found little empirical support over almost the last three decades. Clearly, as a general statement about organizations, it is an inadequate, even invalid, description. Two, if we embrace the weak selection view as a more satisfactory and general descriptive statement about organizations, not only does this put paid to the usual criticisms about the incompatibility of ecology and strategy as mentioned above, it suggests the clear complementarity of the two agendas. Fundamentally, both selection and adaptation processes help bring about change in organizational populations, both at the levels of individual organizations and organizational forms.13 At the broadest level, organizational populations change through the creation of new organizational forms and the decline or demise of extant forms, the differential entries and exits of organizations of a particular form, and change in individual organizations and organizational forms. The study of all these questions in their myriad incarnations is at the heart of an evolutionary perspective on organizations that can provide the building blocks both for organizational ecology and for organizational strategy.14
THE EMERGING EVOLUTIONARY SYNTHESIS15 In an earlier essay, my co-author and I (Bryce & Singh, 2001, pp. 163–165) argued that, over the years, an evolutionary synthesis had emerged related to the study of organizations, and we outlined some of the main features of this synthesis. In this section, I draw upon those earlier observations and suggest, building upon Baum and Singh (1994a, 1994b, pp. 3–20), that a broader view of organizational evolution comprising a dual focus on both ecological and genealogical hierarchies is the appropriate framework for future research in strategy and organization. While most early work in organizational ecology (Hannan & Freeman, 1977, 1984, 1989) had taken a
190
JITENDRA V. SINGH
structural view of organizational form, more recent developments have made some headway with a more genetic view (see, for example, Bruderer & Singh, 1996 and Ethiraj & Singh, 2005, among others). Here it is worth noting that Hannan and Freeman (1977, pp. 934–936) had chosen explicitly to stay away from a genetic conceptualization of organizational form (note their discussion of Marschak & Radner’s, 1972, notion of ‘form as blueprint’) by focusing instead on the formal structure, patterns of activity, and the normative order within organizations. This evolutionary synthesis, particularly when combined with the weak selection argument discussed above, provides a broad framework to guide future developments in strategy and organization theory. Bryce and Singh (2001, pp. 163–165) wrote: Evolutionary thinking about organizations has moved in the last twenty-five years into the mainstream of organizational thinking (see, for example, Nelson & Winter, 1982; Hannan & Freeman, 1989; Baum & Singh, 1994a, 1994b; Aldrich, 1999). However, it is still true that there is not widespread agreement on what an evolutionary perspective on organizations entails. Therefore, we lay down below some of the main features of such an evolutionary synthesis. The overview below builds upon ideas developed elsewhere (Baum & Singh, 1994a, 1994b; Rao & Singh, 2001a, 2001b) y.
We conceptualize organizational evolution as a process that has the following features: – it involves dynamic change over time; – the evolving entities are subject to selection pressures broadly in inverse proportion to their fitness to the selection environment; – the evolutionary trajectory is history or path dependent, i.e., future steps depend on previous states attained; – selection processes involve the interplay of ecological and genealogical entities (Baum & Singh, 1994a, 1994b); – and selection processes involve multiple nested levels of analysis such that different nested entities are simultaneously subject to selective influence. We suggest that although questions of organizational evolution have drawn more attention in recent years with the relative fecundity of research in organizational ecology, their scope is considerably broader. Organizational evolution is a central question in the study of organizations more generally. Indeed, various leading theories of organization – contingency theory, resource dependence theory, institutional theory, organizational ecology, organizational learning, evolutionary economics, and punctuated equilibria – may all be reframed as specifying the detailed processes by which organizational evolution takes place.
Ecology, Strategy and Organizational Change
191
The emergent evolutionary synthesis involves several key shifts in the research literature. While it is likely that there may not be universal agreement on these shifts, we believe some of these shifts are – – a shift from strong to weak selection: whereas the earliest theoretical formulations stated that selection was the predominant mode of change, the burden of empirical evidence from numerous studies supports better a view in which selection plays an important role, but occurs together with adaptive change in organizations. Thus, adaptation and selection are not only not opposing forces, they act in complementary ways; – a shift from a focus primarily on Darwinian mechanisms to a focus on Lamarckian mechanisms, such that organizational learning can be transmitted within and between organizations over time. Whereas it is clear that Darwinian selection does operate, it is increasingly evident that Lamarckian selection is often a better descriptor of intraorganizational and higher order change; – a shift from organizational populations being the primary level of analysis to the study of selection at multiple levels of analysis, in particular their nested (one level contained within the next higher level), hierarchical nature; – a shift from an examination of entities in their respective ecological contexts, to a simultaneous interest in their genealogical lines of descent; – a shift from viewing environmental context as exogenous to endogenous environmental contexts, leading to co-evolution; – a shift from focusing primarily on macrostructure (vital rates affecting the population and higher levels, as a whole) of organizational populations to a simultaneous focus on macro- (primarily at the lower of the nested, hierarchical levels) and microstructure, involving the simultaneous study of nested levels of analysis; – a shift from viewing institutional and selection views as antithetical, to the view that institutional contexts – whether the state and its actions or changing social beliefs and values – often provide the selection context, highlighting the complementarity of the two views. I believe the passage of a few years since making the above statements has only strengthened their validity. However, some elaborations and additions are appropriate that take better account of research published since then. First, Michael Hannan and his colleagues (prominently Glenn Carroll and Laszlo Polos, among others) have embarked upon a quite ambitious research program aimed at a unified logical formalization of all of organizational ecology (see, for example, Hannan, Carroll, & Polos, 2003a, 2003b,
192
JITENDRA V. SINGH
2003c; Hannan, Polos, & Carroll, 2004; Po´los, Hannan, & Carroll, 2002), perhaps energized by Peli, Bruggeman, Masuch, and O’Nuallain’s (1994) formalization of structural inertia theory in first-order logic and an examination of the logical properties of the formalized theory. The scale of ambition of this exercise is admirable. While a comprehensive review of this body of work is beyond the scope of this essay, there are two important points of departure embodied in this new work that here are noteworthy. One concerns the interest of Hannan and colleagues in a more genetic conceptualization of organizational form, typically expressed in this work through a focus on organizational identity. The other point of departure is a more direct and explicit interest in individual organizational change. As more related papers appear, a more coherent picture should emerge of how this new thrust fits overall with the earlier more empirically driven research agenda of organizational ecology. We should also, in time, be better able to understand the potential of this work in shaping the future agenda for the study of strategy and organization. Finally, I want briefly to elaborate the somewhat cryptic assertion made by Bryce and Singh (2001), as quoted above, that many leading theories of organization can be recast as evolutionary theories. The purpose below is not to lay out a fully fleshed out theoretical perspective, but to outline in broad brushstrokes just how such an approach might look. It is necessary here to deal with simpler versions of each theoretical argument, without going into all the details, nuances, and exceptions. Below I will tackle several perspectives one by one and outline broadly the role of selection processes in each theory. Contingency theory (see, for example, Lawrence & Lorsch, 1969) posited that the structural arrangements within an organization needed to match the environmental demands as well as the demands of size and technology. Organizations that had the appropriate fitting structures exhibited better performance. The evolutionary version of this argument focuses on the dynamic process as opposed to the static, as was common in the cross-sectional studies typical at the time. Organizations whose structures match the demands of size, environment, and technology will perform better over time than other organizations that do not have the appropriate structures in place. Such latter organizations will be subject to stronger selection pressures and more likely to fail over time. The central statement of resource dependence theory (Pfeffer & Salancik, 1978) was that organizations implemented various interorganizational strategies such as mergers, diversification, interlocking directorates, among others, as a way to manage critical interdependencies and uncertainty.
Ecology, Strategy and Organizational Change
193
Interestingly, though, to the best of my knowledge, most empirical studies of resource dependence, while they showed that firms were more likely to use these strategies at the highest level of uncertainty, usually did not take the next step and asked whether resorting to these strategies did in fact reduce problematic uncertainty and interdependence, resulting in performance differentials. This direction, which directly ties the appropriate interorganizational actions to reduction of uncertainty and interdependence, thereby enhancing performance, takes resource dependence toward an evolutionary approach. Thus, firms not following appropriate interorganizational strategies will be subject to greater selection pressures and more likely to get selected out. Resource dependence theory also had an important intraorganizational component pertaining to internal power distribution within organizations. Here again, an evolutionary version of the argument emphasizes the selection advantages to having internal power distributions that are flexible enough to shift in response to changing environmental contingencies as translated into novel critical problems for organizations. Such nimble organizations will be favored by selection forces over organizations with more rigid, non-responsive, and ossified power distributions. While institutional theory comes in variants that differ somewhat from each other (Scott, 1987; DiMaggio & Powell, 1983), for the sake of simplicity, here I focus primarily on the influential statement by Meyer and Rowan (1977). The central argument in their essay on institutional theory was an interesting contingency statement. As institutionalized myths and beliefs in the cultural environment of organizations change over time, they have to respond with corresponding changes in formal structure, policies, and practices that reflect these changing beliefs. Organizations that successfully achieve this became more legitimate, which provides them greater access to societal resources, resulting in enhanced performance. Organizations that do not make such changes suffer reduced legitimacy and subsequent lower performance. Once their theory is articulated in this fashion, the evolutionary component becomes clearer, with less legitimate organizations being subject to stronger selection pressures (also see Singh et al., 1986a, 1986b) and more likely to be selected out over time. Of all the different theoretical perspectives, the links of organizational ecology and evolutionary economics to an evolutionary perspective perhaps are the clearest. However, there are some divergences between the two, in that, among other differences, evolutionary economics (Nelson & Winter, 1982) was avowedly Lamarckian in flavor while Hannan and Freeman (1989) took a more Darwinian approach. Also, the focus in evolutionary
194
JITENDRA V. SINGH
economics was more intraorganizational, whereas the focus in organizational ecology, at least in the early years, was more at the population level. Organizational learning (see Levitt & March, 1988, for a review) also has clear connections to a selection narrative. Indeed, organizational learning processes can be rethought as selection among a population of behaviors, with some behaviors being reinforced by positive outcomes and other, nonreinforced behaviors becoming extinguished over time. Indeed, the role played by learning in reconciling adaptation and selection perspectives to organization is by now well understood (Levinthal, 1991). Of all the different leading theoretical perspectives, perhaps the transactions costs approach (Williamson, 1975, 1979, 1985, 1996) would appear, prima facie, the most difficult to reconcile with a selection perspective. However, I believe selection does play a role, perhaps even an important one, though it has not been explicated nearly enough in the theory. While this may not be the only, or, for that matter, the best way to integrate a selection view with transactions costs economics, I outline below one such mechanism. A key proposition in Williamson’s approach is that depending on transactions costs considerations, activities will shift from markets to organizations and vice versa, making the optimum choice of firm boundaries an important focus of this theory. Curiously absent is the mechanism by which this movement of transactions takes place between firms and markets. I believe this is precisely where a selection argument fits well with Williamsonian logic. Whether concretely manifested in market forces or through managerial agency, selection based on transactions costs efficiencies provides the missing mechanism in the transactions costs approach through which firm boundaries change.16 Having reiterated the emergence of an evolutionary synthesis as noted earlier by Bryce and Singh (2001), and having pointed out how several leading theories can be recast as evolutionary theories, I turn my attention in the next section to some observations on an evolutionary perspective on organizational change.
AN EVOLUTIONARY PERSPECTIVE ON ORGANIZATIONAL CHANGE I summarized in the section above Bryce and Singh’s (2001) statement of an evolutionary perspective on organizations. In this section, I will develop some implications of such an evolutionary perspective for organizational
Ecology, Strategy and Organizational Change
195
change. I will elaborate some extant views on organizational change and draw them closer to an evolutionary approach, building upon some earlier work (Amburgey & Singh, 2002; Moldoveanu & Singh, 2003). The starting point is to examine three separate though conjugate bodies of work and highlight their relationships: the quantum or configuration view (Miller & Friesen, 1984), the gestalt view in the study of strategy and structure (Mintzberg, 1981, 1992), and the punctuated equilibria view of organizational change (Tushman & Romanelli, 1985). Although doing so runs the risk of simplifying overly much every one of these theoretical positions, I begin with broad statements of each perspective. The quantum or configuration view, posited in the context of contingency theory that had by then likely peaked in its influence and was on the wane, treated organizations ‘‘as complex entities whose elements of structure, strategy and environment have a natural tendency to coalesce into quantum states or configurations. These configurations are composed of tightly interdependent and mutually supportive elements such that the importance of each element can best be understood by making reference to the whole configuration y. Configurations may represent common organizational structures, common scenarios of strategy making in context, and even common developmental or transitional sequences’’ (Miller & Friesen, 1984, p. 1). The gestalt view, attributed primarily to Mintzberg (1981, 1992), in related vein to the configuration approach, argued that organizational structure and strategy occur in certain natural gestalts, and that other hybrid arrangements do not perform as well. The punctuated equilibria view17 holds that organizational change occurs, for the most part, in incremental steps over long periods of time. However, from time to time, this incremental process is punctuated by episodes of very high rates of change. These three theoretical views are hardly identical. Yet, despite their differences, when taken together, they do constitute a broad, mutually sympathetic worldview. Central to this worldview is the position that the tempo of change in organizations is incremental over long periods of time, punctuated by episodic bursts of high rates of change. Moreover, structure, strategy, even organizational changes (transitions) are more likely to occur in certain configurations than in others. However, these approaches do not, in my view, address nearly fully enough the evolutionary aspects of these phenomena. Specifically, they do not ask how structure, strategy, and patterns of organizational change got to be that way. I believe exploring this question will be quite informative for the study of organizational change. I believe that while it is useful to show empirically that structure, strategy, or organizational changes are more likely to occur in certain configurations,
196
JITENDRA V. SINGH
this says relatively little about how they got to be that way. First of all, there are some methodological complexities involved here. Specifically, imagine, as theorists of this genre would almost certainly agree, that certain patterns of organizational strategy, structure, and change are more conducive to better organizational performance. As an organizational population evolves over time, it seems reasonable to assert that organizations that pursue other suboptimal patterns of strategy, structure, or organizational change will face stronger selection pressures, and eventually be selected out of the population, leaving behind the organizations that adopted the more adaptive patterns.18 For the purposes of this argument, it is not even necessary that firms make these choices consciously; ad extremum, even random choices made by firms would result, over time, from the sheer winnowing action of selection processes, in survivors showing the better adapted patterns (configurations) of strategy, structure, or change. More realistically, of course, it is very likely that when faced with uncertainty, some (or many) firms will imitate other apparently successful firms, making this convergence to the gestalts or configurations occur even faster. The key point is that starting such empirical investigations from an extant set of organizations is likely to overstate the frequency in the population with which these adaptive configurations of strategy, structure, or change occur, precisely because of survivor bias problems, tellingly noted by organizational ecologists in earlier published work. If asking the how question (focusing on the process by which this evolution occurred) is important, adopting a prospective design rather than a retrospective design, as is standard in organizational ecology studies, is crucial to avoiding problems of survivor bias. To the best of my knowledge, such studies have not yet been done when it comes to investigating patterns of organizational structure, strategy, and change. While this is a challenge on the one hand, it is also a research direction with considerable future promise on the other.
SOME OPEN QUESTIONS RELATED TO ORGANIZATIONAL CHANGE More generally, I believe there is much untapped potential in viewing these questions about patterns of organizational structure, strategy, and change through the lens of the evolutionary perspective outlined in the section above. Some of the questions that will get raised are, I believe, novel and their pursuit fruitful. For instance, precisely how do Darwinian and Lamarckian
Ecology, Strategy and Organizational Change
197
processes play out in patterns of organizational strategy, structure, or change? How does the view of organizations as nested, hierarchical entities raise different questions about patterns of structure, strategy, and change? How do changes in the institutional context – as in the recent examples of the transition from different versions of state-centered socialism to market economies in former Eastern European countries like Hungary, Poland, or the Czech Republic, or even China and India – alter the dynamic patterns of strategy, structure, and change, and how do patterns adapted to the old institutional context get selected out? What specific ecological and genealogical mechanisms are involved in the spin-off phenomenon and what role does history play here? While these are just broad-brush questions at this point, I fully believe that, in time, answers to these and other related questions will give rise to a body of knowledge about organizational change that is more truly evolutionary in nature. More specifically, in pursuit of a more comprehensive evolutionary approach to organizational change, I believe there are several specific research directions that have potential. First, to begin with, at its broadest level, organizational ecology has had to address in the aggregate six key questions: how new organizational forms are created and how they sometimes die, how change in organizational forms comes about, how change in organizational populations is brought about by processes of entry and exit by individual organizations, and how change occurs in individual organizations. An overall assessment of the many achievements by organizational ecology as a field suggests that, despite some exceptions, the center of gravity of the empirical research has been around the study of entry and exit rates across an impressive array of populations (although, as reviewed in an earlier section, an exception to this generalization is the studies of change in individual organizations as well). In service of the broader goal, the creation of a more comprehensive evolutionary approach to organizational change, we need to address these other questions about the creation, change, and extinction of organizational forms more fully. Below, I suggest some initial starting points that will take us in this desired direction. This is not intended to be comprehensive or exhaustive; it is a selective view of some questions that I believe are important. Secondly, related to the discussion above about genetic views of organizational forms, the study of organizational capabilities and their evolutionary implications is, I believe, a fruitful area of study.19 Whereas the concept of organizational capabilities (Barnard, 1938 [1975]; Simon, 1957; Selznick, 1957) has been invoked in its various incarnations for almost fifty years (a recent addition to this list being the notion of core competence, Hamel &
198
JITENDRA V. SINGH
Prahalad, 1994), relatively little systematic research has been done that asks directly whether and how capabilities influence firm performance, although several studies have addressed the question indirectly (Brush & Artz, 1999; Henderson & Cockburn, 1994; Makadok & Walker, 2000; McGrath, MacMillan, & Venkatraman, 1995; Schroder, Bates, & Juntila, 2002; Silverman, 1999). A modest start in this direction has recently been made by Ethiraj, Kale, Krishnan, and Singh (2005) in an empirical study of the influence of capabilities on performance at the project level in a software services firm. However, much more is needed to more fully unpack the evolutionary role of capabilities. For instance, it is quite plausible that selection forces operate directly on organizational capabilities at the subunit level (Anderson, 1995), whereas adaptation is observed at the organizational level. A related question is: where do organizational capabilities reside? Clearly, there exist both organizational components at the systemic level and other components that are carried by individuals. This latter position is supported by casual empiricism: some firms spawn multiple entrepreneurial ventures, presumably because hopeful entrepreneurs leave these parent firms carrying with them blueprints of capabilities that enable their future entrepreneurial success. As has been noted in the literature, executive mobility has an important role to play here (Saxenian, 1994). This question has productively been studied recently in the context of Silicon Valley law firms to examine how the transfer of resources and routines between a parent organization and its progeny reduce the life chances for the parent firm and increase them for the progeny (Philips, 2002) and the genealogical persistence of gender inequality through the blueprints that founders transfer from their parent firms to their new firms (Philips, 2005).20 Third, the creation and demise of organizational forms has not received much attention until recently (though see, for exceptions, Lumsden & Singh’s, 1990, conceptual discussion of organizational speciation for one early paper, the more recent work by Ruef, 2000, 2004, as discussed above, and McKendrick, Jaffee, Carroll, & Khessina, 2003, on the rise of disk array producers as a new organizational form), perhaps because such events are uncommon, even rare. And yet they do occur. Arguably, the recent creation of firms like Amazon and Ebay are good candidates for new organizational forms. This very question begs a deeper discussion of precisely what constitutes organizational forms (see Herber, Singh, & Useem, 2000 for a discussion in the context of emerging technologies). Rao and Singh (2001a, 2001b) made some headway on this issue by discussing the role of institutional factors in the speciation of the automobile and biotechnology industries. Bryce and Singh (2001) argued that for new organizational forms,
Ecology, Strategy and Organizational Change
199
the usual forces of competitive selection may be less important than institutional selection. ‘‘Whereas the selection pressures of the institutional context are clearest in the extreme case of legal proscription of organizational forms with certain features, the phenomenon has more generality. Thus, organizational forms can decline because of the erosion of the institutional context or because of discontinuous legal changes. A global change in values over a period of time can also lead to delegitimization of a form’’ (Bryce & Singh, 2001, pp. 170–171). I believe that a productive context in which to raise such questions about the growth and decline of organizational forms is in economies that have in recent years made the transition from socialism to varying degrees of free market economies. Both countries in Eastern Europe (Poland, Russia, Hungary, the former East Germany, and the Czech Republic, among others) and Asia (China and India, among others) come to mind. Whereas in mature market economies like the United States, we do not readily discern the very foundations of selection environments, these transitional economies are naturally occurring experiments that have greater potential to illuminate such questions. Recent work by Stark (1996, 2001) on the creation of new organizational forms in Hungary and the Czech Republic is one example of novel work on this theme. Some other papers that are broadly relevant to the theme here are Dobrev’s (1999, 2001) empirical studies of the newspaper industry in Bulgaria and the emergence of an independent press. These above examples of nation states making the transition to free market economies in recent years are examples of settings in which seismic shifts in institutional contexts have led to the large-scale destruction and creation of organizational forms, a virtual Cambrian explosion. Of course, it must be noted that changes in the institutional contexts, while they shake the very foundations of selection environments (see Bryce & Singh, 2001, pp. 170–176), are but one way in which new forms can come about, another important source being technological innovation, particularly radical technological discontinuities. Several other observations are noteworthy in this regard. One, the empirical study of these instances of new form creation poses some serious demands for research design and data gathering. The research designs, of necessity, have to be longitudinal. But it is also necessary to develop a fine-grained understanding of the institutional contexts both before and after the transition. Also, clearly, these transitions are not point events but much more process like, and the details differ from one country to another. Two, these very differences of detail, however, provide a research opportunity to get at the institutional foundations of organizational form creation and destruction through comparative research across
200
JITENDRA V. SINGH
different countries. Such research can lead to more general theoretical ideas about the creation and destruction of organizational forms. Three, a related question that Bryce and Singh (2001, pp. 180–184) discussed earlier concerns whether there is a global convergence or divergence of organizational forms. It is worth reiterating their conclusion here: while there are forces both for the convergence and divergence of organizational forms, the divergences are sufficiently strong that convergence to a common form is not likely in the near future. Finally, the previous point has deeper implications for the global spread of capitalism: it seems likely that there will not be just one form of free market capitalism across the world; what is already well under way is the development of several variants of free market capitalism, as evident in China, India, Russia, Poland, Hungary, and more developed countries like Korea or Taiwan, which, while they may share some common features, also differ from each other in significant ways. Finally, related to the point raised above about change in organizational forms, I believe that, at a general level, organizational forms are continuously being influenced both by competitive and institutional selection forces. Arguably, in a contemporary development of far reaching consequences, the modern global corporation is itself being reconfigured and transformed (relatedly, see Friedman, 2005, for a lucid and influential popular account of the ‘flattening’ of the global economy and the implications for global organizations). The core argument follows below in somewhat simplified form. In the last fifty, or even a hundred years, the global business landscape has been dominated by a specific organizational form: the large, vertically integrated, multidivisional corporation. While the story of the rise of this organizational form is clearly a complex and multifaceted one, I want here to zone in on two factors that appear to have played key roles in the recent unbundling (Hagel & Singer, 1999) process: the role of information and communication technologies and globalization, particularly of labor markets. These broad changes have been noted earlier by some writers focusing on network forms of organization (see, for example, Powell, 1990, 2001), but, with the passage of time, I believe several nuances have become clearer than before. Some of the more incisive thinking about the rise of vertically integrated firm and the multidivisional corporation is evident in Williamson’s (1975, 1979, 1985, 1996) body of work21 on transaction cost economics. One important reason why firms chose vertical integration over resorting to market transactions was due to problems (‘transactions costs’) exacerbated by information asymmetries. However, the use of information and communications
Ecology, Strategy and Organizational Change
201
technologies can also ameliorate information asymmetries in such market transactions (Clemons, 1991). Since the middle 1990s, dramatic improvements in information and communications technologies, including the explosive growth in the use of the Internet and the World Wide Web, have made communication and interaction over long distances relatively inexpensive. This has led to the unbundling of the modern corporation (Hagel & Singer, 1999; Hagel, 2002). It is against this backdrop that the rapid growth of the Indian IT and software services industry over the last few decades (see Ethiraj et al., 2005, for a description of the industry background), and, more recently, business process outsourcing (see Aron & Singh, 2005) by US and European firms are best apprehended. This new emerging global corporation is narrower in scope at its core, with many activities that were formerly carried out within firm boundaries now being done contractually through a network of market relationships in distant countries like India and China (Hagel & Brown, 2005).
ORGANIZATIONAL ECOLOGY AND ORGANIZATIONAL STRATEGY: AN EMERGING SYNTHESIS? At the 1987 annual meetings of the Academy of Management, myself and Robert Burgelman had led a symposium titled Strategy and Ecology: Steps to a Rapprochement in which we presented a paper (Burgelman & Singh, 1987) in which we asserted that, contrary to the view of some scholars at the time that strategy and ecology were fundamentally opposing perspectives, the two bodies of work were inherently complementary. Specifically, we suggested that together with organizational populations and communities, there was another important level of analysis in organizational ecology – intraorganizational ecology. In this context, the body of work by Burgelman (1991, 1994, 1996, 2002) on the emergence of strategies within organizations can best be reframed as discussions about the intraorganizational ecology of strategic decision-making. While the symposium attracted some attention, it is worth noting that it did not immediately set the world on fire. In retrospect, I think we were on the right track in highlighting this rapprochement; our timing, however, was off by a couple of decades. Back in 1988, neither academic theorists in organizational ecology, nor in strategy were ready substantially to embrace this idea. On the other hand, to be fair, much of the work that has been accomplished in the years since, the kind of
202
JITENDRA V. SINGH
research that Burgelman and Singh were calling for at the time, was not yet in existence. While some may argue that this synthesis is still not in the theoretical mainstream, today a somewhat clearer picture has emerged. Moving forward, even as some parts of organizational ecology and strategy become more closely integrated, it seems likely that both literatures will maintain somewhat separate identities, with some questions unique to each subfield, even as increasingly more questions are shared between them, providing growing convergence of the two. As I had noted in the introductory section above, one important way in which the convergence of agendas has occurred between organizational ecology and organizational strategy is in the study of organizational change. Yet this is not the only area where the agendas of the two literatures converge. In this section, I want to highlight some other aspects of this convergence. To begin with, there is a history that has had a shaping influence on the present which deserves some acknowledgement. As far back as a decade and a half ago, individual organizational change was seen by some as a question of importance in organizational ecology (Singh & Lumsden, 1990; Singh, 1990). Baum and Singh (1996) did a fine-grained empirical analysis of how changes in organizational niches were influenced by competitive dynamics. Baum and Singh (1994a, 1994b) proposed a theoretical framework for the study of the evolutionary dynamics of organizations that had a dual emphasis. On the one hand, it emphasized the nested, hierarchical nature of organizations, and, on the other, it focused on both ecological and genealogical hierarchies. Whereas the ecological hierarchy focused on the shaping influence of the ecological context at each level of analysis through selection processes, the genealogical hierarchy focused on the entities themselves that were being transmitted over time. This framework was broad enough to accommodate in unified form many research questions from strategy and organizational ecology. Building upon Baum and Singh, Bryce and Singh (2001) summarized key elements of an evolutionary perspective on organizations. Baum (2002) has recently edited an ambitious volume, Companion to Organizations, which includes comprehensive review articles at the intraorganizational, organizational, and interorganizational levels, with a final part of the book devoted to research methods. Baum and Shipilov’s (forthcoming, 2006) review of organizational ecology focuses explicitly on organizational change and reviews empirical studies of rates of organizational change and the survival consequences of change. Even as this convergent agenda has continued to grow and ramify, a new research journal, Strategic Organization! has successfully been launched with a novel
Ecology, Strategy and Organizational Change
203
mission quite sympathetic to this growing convergence, as the name of the journal itself suggests. It appears that the literature has come a long way since 1988 in the direction that Burgelman and I had indicated. Indeed, as noted earlier, some recent work by Hannan and colleagues on considerations of organizational identities and organizational change has also begun to move in broadly consistent directions. Yet, much more work remains undone. While, as a general matter, I believe it is difficult to accurately predict exactly where the most leverage may come from to further this emerging convergence, I want to underscore some broad directions that hold promise. Clearly, the ecological and genealogical dynamics of organizational change still remains an important theme. Among the other promising directions, some in particular stand out and below I touch upon them briefly. First, a theme with much unexplored potential concerns the ecological and genealogical dynamics of creation and demise of organizational forms. While some modest progress has been made, we need a more comprehensive investigation of this area, including the role of institutional and technological factors. Second, pursuing genetic conceptualizations of organizational form also holds significant potential.22 Indeed, at the intraorganizational level, we can view strategies themselves as discrete entities and study their evolutionary paths, and the myriad ways in which adaptation and selection processes come together as strategies unfold. Third, an impressive cumulative body of knowledge now exists on the competitive dynamics within organizational populations in the organizational ecology literature. While competition has also been a subject of much interest within strategy, the organizational ecology literature brings to bear an interest in mutualism as well, echoing some recent treatments of competition and collaboration, or co-opetition (Brandenberger & Nalebuff, 1996). Fourth, computational models of evolutionary dynamics of organizations hold great potential. As indicated above, the emergent evolutionary synthesis highlights, among other issues, the importance of nested, hierarchical levels, the dual focus on ecological and genealogical hierarchies, path dependence and the presence of many agents. Studying such questions using the usual methods of empirical analysis, closed form analytical modeling, or narrative theorizing quickly runs into limitations, given the complexities involved. Under these conditions, computer-based simulations are a powerful tool, and their potential is increasingly being harnessed (see, for example, Bruderer & Singh, 1996; Levinthal, 1997). An added advantage of using computational models is that it forces theorists to formalize propositions that are executable: this heightened logical precision and internal consistency is very much in the spirit of recent axiomatization efforts in
204
JITENDRA V. SINGH
organizational ecology by Carroll, Hannan, and Polos, as I have discussed above.23 Finally, for many years, Porter’s (1980) influential reframing of research from industrial organization economics has been a dominant presence on the curricular agenda of the strategy field. In the meantime, an impressive body of research focused on evolutionary processes in strategy and organization has progressively accumulated. I wonder if the time is not ripe for rewriting strategic management in a way that takes more seriously the evolutionary nature of strategy-making processes and the importance of organizational context. Whereas Porter was more interested in the comparative statics of the performance characteristics of different strategies, such an evolutionary approach would focus more on where strategies come from. Some features of this approach to strategy would highlight how the historical trajectory of an organization, where it has been in the past, has an important influence on which strategies it can adopt next. Put another way, if one thinks of evolutionary processes as a branching, ramifying tree, it is not possible suddenly to jump from one branch of the tree to another distant branch; neighborhood moves are more likely to occur and to succeed. Of course, this exciting challenge resides largely in the future.
SUMMARY I began this review essay by noting the changing role of organizational change in organizational ecology. Over time, not only has organizational ecology modified its position on organizational change, a new level of analysis has been added to organizational ecology that deals with intraorganizational ecology. The agendas of organizational strategy and organizational ecology have converged, at least in part, due to the interest in organizational change in both literatures. I reviewed the broad pattern of empirical findings related to organizational change. The burden of the empirical evidence ruled out strong selection arguments in favor of weak selection arguments, which posit the joint role of selection and adaptation forces in shaping the trajectories of organizational populations. I noted the emergence of an evolutionary synthesis in the literature and elaborated some of the main features of this synthesis. I then developed some of the implications of this evolutionary synthesis for the study of organizational change. Finally, I concluded this essay by touching upon some open research questions, and pointing to some questions that instantiate the convergence of agendas between organizational ecology and organizational strategy. This emergent
Ecology, Strategy and Organizational Change
205
synthesis provides the foundation for what will surely be some very exciting developments in strategy and ecology in the years ahead, but more broadly in management theory as well.
NOTES 1. Interestingly, the companion paper on organizational change in this volume, an essay on family controlled businesses (Miller & Le Breton-Miller, 2006), which explores how governance affects the capacity for change in organizations, contrasts the differences between the strategic and ecological perspectives on organizational change. They suggest that the strategic view focuses on constant learning and adaptation, emphasizing flexibility, whereas the ecological focuses on inertia in organizations and their inability to change (reflecting the strong selection view from organizational ecology). As I will point out later in this essay, the current state of affairs in the literature is somewhat more nuanced, and the two views are more complementary than competing. 2. It is worth noting that one of the editors, an economist by training, has tried to integrate relevant parts of IO economics with organizational ecology (see Boone & van Witteloostuijn, 1995; van Witteloostuijn & Boone, 2006). He believes that the attacks by some organizational ecologists on maximization and equilibrium thinking in economics are ill-advised. In his view, both concepts are extremely flexible, suggesting possibilities for integration. In addition to these conceptual opportunities, there are interesting opportunities in applying the modeling apparatus from economics, especially game theory, to organizational ecology (see van Witteloostuijn, 1998; van Witteloostuijn, Boone, & van Lier, 2003), by deriving a game theoretic model in which organizational inertia is performance enhancing. However, this work has not received adequate attention, perhaps due to paradigmatic differences between the fields. 3. A nuance is worth mentioning here. Early critics of organizational ecology, as I pointed out above, argued that ecologists took the position that organizations were inert and did not change. Since it was possible, argued the critics, to think of specific counter examples of organizations that had changed, or even made radical changes in some cases, this approach was obviously wrong. A subtler, more generous interpretation of Hannan and Freeman (1977) takes a different tack. While it is possible to think of specific organizational change anecdotes, the deeper issue is whether, as a general matter, organizations change routinely or all the time, (a position taken, for example, by March, 1981). Hannan and Freeman (1977) challenged this more general view of organizational adaptation. 4. Hannan and Freeman (1977, pp. 930–931) state: ‘‘Clearly, leaders of organizations do formulate strategies and organizations do adapt to environmental contingencies. As a result, at least some of the relationship between structure and environment must reflect adaptive behavior or learning. But there is no reason to presume that the great structural variability among organizations reflects only or even primarily adaptation. There are a number of obvious limitations on the ability of organizations to adapt. That is, there are a number of processes that generate structural inertia. The
206
JITENDRA V. SINGH
stronger the pressures, the lower the organizations’ adaptive flexibility and the more likely that the logic of environmental selection is appropriate.’’ 5. Hannan and Freeman (1984, p. 149) state: ‘‘This paper goes beyond our earlier theory in acknowledging that organizational changes of some kinds occur frequently and that organizations sometimes even manage to make radical changes in strategies and structures. Nevertheless, we argue that selection processes tend to favor organizations whose structures are difficult to change. That is, we claim that high levels of structural inertia in organizational populations can be explained as an outcome of an ecological–evolutionary process.’’ 6. At least one of the editors of this volume took issue with my interpretation of the two Hannan and Freeman papers and argued: ‘‘y in our view (the causal direction flowing from selection to inertia) was not necessarily meant to supplant the earlier argument (inertia therefore selection). An alternative interpretation is that the two papers produce a cyclical model in which inert organizations are first selected favorably (due to higher reliability and accountability) but subsequently this very inertia also exposes them to higher selection pressures if rapid exogenous change mandates significant internal reorganization. In short, selection leads to inertia in relatively stable environments and inertia leads to selection in relatively dynamic environments.’’ (I have added the italicized text to clarify the meaning of the editorial comment.) I believe this is an interesting, even creative synthesis of some of the ideas from the two Hannan and Freeman papers and one that deserves more attention. However, I do not think attributing this interpretation to Hannan and Freeman can be justified based on the original texts. 7. There is significant unexplored potential behind the notion that founding conditions can influence future rates of change through imprinting effects. One way in which this might potentially occur is through mode of industry entry. For instance, Dobrev, Kim, and Hannan, 2001, found that de alio vs. de novo entry modes influenced the likelihood of change. This avenue of research merits further investigation. 8. Several interesting contributions to this intraorganizational focus have also come from economists. Czernich (2003) has worked on the ecology of projects within a large multinational enterprise. Boone, van Olffen, van Witteloostuijn, and de Brabander, 2004, in an ecology-inspired study of executive mobility, have focused on the ecology of selecting top managers, and the implications for firm-level strategy and population-level selection. 9. Quite felicitously, at the time of writing of this paper, a comprehensive review of organizational ecology has just been completed (Baum & Shipilov, forthcoming, 2006). Since Baum and Shipilov surveyed comprehensively the empirical literature on organizational change, I build upon their efforts in this section. 10. Ruef (2006), in one of the chapters in this volume, develops the notion of entrepreneurial inertia, the lag time exhibited by entrepreneurs or investors in entering a market niche, and boom-and-bust cycles in industries and financial markets. 11. I believe the use of strong inference (Platt, 1964) is valuable in research. According to this view, progress comes more readily if we conclude from empirical studies which theoretical positions are not supported by the data. In an ideal world, once enough such evidence has accumulated, researchers can spend their time and energy more productively by addressing other open questions.
Ecology, Strategy and Organizational Change
207
12. It is worth noting that the weak selection view does not suggest that change in organizations is always adaptive in its consequences. Indeed, consistent with March (1982), while change occurs routinely and all the time, its consequences are often difficult to know ex ante and sometimes surprising. 13. The complementarity of adaptation and selection processes is not a novel observation. It has been noted earlier by Singh et al. (1986a, 1986b) and Levinthal (1991), among others. 14. A recent paper (Hannan, Polos, & Carroll, 2003c) has examined how architectural change in organizations can often trigger change in other subordinate units, leading to cascades of change. Given the limited foresight of change initiators about the interconnectedness of units and the cultural restrictions on architectural features, they underestimate the costs of change and how long it will take, leading to them undertaking changes with adverse consequences. The editors of this volume suggested that one way to introduce change into ecological theory may be to limit the most adverse consequences for cascades of change, whereas single changes may occur more frequently with fewer adverse consequences. This may, in time, prove to be an interesting avenue. For present purposes, however, I am not aware of any large-scale empirical studies of cascades of change; indeed, the research demands would, prima facie, appear daunting. Baum and Shipilov’s review does not focus on cascades of change either. Hence, this issue is somewhat beyond the scope of this essay and will need to await further empirical study. 15. This section draws upon some of my earlier writings on this topic (see Bryce & Singh, 2001, in Paul DiMaggio’s edited book, The Twenty First Century Firm: Changing Economic Organization in International Perspective). 16. While I have known Oliver Williamson professionally for some years, a few years back I ran into him one more time at a conference in the spectacular surroundings of the Asilomar Conference Center in Pacific Grove, CA. I put my conjecture to him as a way to bring selection processes into the transactions costs approach, while simultaneously providing a mechanism currently missing in the perspective. Williamson thought it was a direction that made a good deal of sense and promptly suggested that I should write a paper on the topic! 17. The term ‘punctuated equilibria’ originated in evolutionary biology (see Eldredge & Gould, 1972) and dealt with questions of macroevolution, the creation and disappearance of species in large numbers from time to time in the fossil record, an issue of great importance to paleobiologists. While Tushman and Romanelli appropriated the terminology of Eldredge and Gould, it is worth noting that their use of this term is somewhat inconsistent in spirit with Eldredge and Gould’s usage. While some may dismiss this point as aesthetic quibbling, I believe it is nevertheless important to highlight. As used in the context of organizational change, punctuated equilibria clearly deals with microevolutionary questions. But in a notable exception, Astley (1985) used punctuated equilibria to refer to macroevolutionary questions. Perhaps because of his premature death a few years later, this line of thinking did not develop significantly in later years. 18. To be fair, Miller and Friesen (1984) were not unaware of the implications of the selection logic for the existence of configurations. As they pointed out, one argument for why configurations exist was that ‘‘y Darwinian forces may encourage only relatively few organizational forms to survive in the same setting, their variety
208
JITENDRA V. SINGH
and number circumscribed by the dictates of population ecology. The Darwinian argument is, of course, very tentative, and we hesitate to make too much of it. It does seem, however, that a type of organizational Darwinism could very well limit the number of viable forms by selecting out those that are relatively weak and those that fail to achieve internal complementarities’’ (p. 21). However, perhaps because evolutionary thinking had not made much progress at that time twenty years ago, they did not pursue this line of inquiry very far. Clearly, some of the methodological issues raised above were not central to their research and did not get adequate attention. 19. For a paper that reviews differing perspectives on capabilities developed by ecologists, IO economists and consultants, see Dobrev, Kim, and Solari (2004). Their argument focuses on the managerial limitations to shifting capabilities. 20. For a different though broadly related study of how the organizational context of an individual either accelerates or retards the likelihood of entrepreneurship, the decision by an individual to found a new organization, see Dobrev and Barnett (2005). 21. While Williamson’s body of work on transactions costs economics has been seminal, he has not paid much systematic attention to the role played by information and communications technologies. Perhaps this was an accurate reflection of the corporations of the 1960s, 1970s, even the 1980s. But, clearly, for the early 21st century corporation, information and communications technologies are the very backbone on which the systemic reconfiguring is taking place. This role of information and communications technologies in considerations of transactions costs is a fertile domain for future research efforts. 22. Organizational ecologists have in recent years taken a greater interest in more genetic, as opposed to structural notions of organizational forms. Po´los et al. (2002), using tools from formal logic and set theory, lay out a general approach to the study of organizational forms, defining forms as a type of socially coded identity. 23. I am grateful to Felipe Csaszar for reminding me of the importance of computational models as a fruitful direction.
ACKNOWLEDGMENT I thank the editors for their many helpful comments on earlier drafts of this paper, and for their patient support along the way while this paper was in progress, and to Felipe Csaszar, for several important suggestions. Thanks are also due to all the attendees of the MOBS conference at the Kellogg Graduate School of Management, Northwestern University where I presented an earlier version of these ideas. This research was completed while I was on a sabbatical from the Wharton School at the Marcel DeSautels Center for Integrative Thinking, Rotman School of Management at the University of Toronto, Canada, during fall 2005.
Ecology, Strategy and Organizational Change
209
REFERENCES Aldrich, H. E. (1999). Organizations evolving. Thousand Oaks, CA: Sage. Amburgey, T. L., & Kelly, D. (1985). Adaptation and selection in organizational populations: A competing risk models. Paper presented at the Academy of Management meetings, San Diego, CA. Amburgey, T. L., Kelly, D., & Barnett, W. P. (1993). Resetting the clock: The dynamics of organizational change and failure. Administrative Science Quarterly, 38, 51–73. Amburgey, T. L., & Miner, A. S. (1992). Strategic momentum: The effects of repetitive, positional, and contextual momentum on merger activity. Strategic Management Journal, 13, 335–348. Amburgey, T. L., & Singh, J. V. (2002). Organizational evolution. In: J. A. C. Baum (Ed.), Companion to organizations (pp. 327–343). Oxford, UK: Blackwell. Anderson, P. (1995). Microcomputer manufacturers. In: G. R. Carroll, & M. T. Hannan (Eds), Organizations in industry (pp. 37–58). New York, NY: Oxford University Press. Aron, R., & Singh, J. V. (2005). Rightsourcing services: Make, migrate or outsource. Harvard Business Review, 83(12), 135–143. Astley, W. G. (1985). The two ecologies: Population and community perspectives on organizational evolution. Administrative Science Quarterly, 30, 224–241. Astley, W. G., & Van de Ven, A. H. (1983). Central perspectives and debates in organization theory. Administrative Science Quarterly, 28, 245–273. Barnard, C. (1938) [1975). The functions of the executive. Cambridge, MA: Harvard University Press. Barnett, W. P., & Carroll, G. R. (1995). Modeling internal organizational change. Annual Review of Sociology, 21, 217–236. Baum, J. A. C. (Ed.). (2002). Companion to organizations. Oxford, UK: Blackwell. Baum, J. A. C., & Rao, H. (2004). Evolutionary dynamics of organizational populations and communities. In: M. S. Poole, & A. H. Van deVen (Eds), Handbook of organizational change and innovation (pp. 212–258). New YorkL: Oxford University Press. Baum, J. A. C., & Shipilov, A. V. (2006). Ecological approaches to organizations. In: S. R. Clegg, C. Hardy, W. R. Nord, & T. Lawrence (Eds), Handbook of organization studies, (2nd ed.). London: Sage. Baum, J. A. C., & Singh, J.V. (Eds). (1994a). Evolutionary dynamics of organizations. New York: Oxford University Press. Baum, J. A. C., & Singh, J. V. (1994b). Organizational hierarchies and the evolutionary process: Some reflections on a theory of organizational evolution. In: J. A. C. Baum, & J. V. Singh (Eds), Evolutionary dynamics of organizations. New York: Oxford University Press. Baum, J. A. C., & Singh, J. V. (1996). The dynamics of organizational responses to competition. Social Forces, 74(4), 1261–1297. Boone, C., van Olffen, W., van Witteloostuijn, A., & de Brabander, B. (2004). The genesis of top management team diversity: Selective turnover among top management teams in Dutch newspaper publishing, 1970–94. Academy of Management Journal, 47(5), 633–656. Boone, C., & van Witteloostuijn, A. (1995). Industrial organization and organizational ecology: The potentials for cross-fertilization. Organization Studies, 16(2), 265–298.
210
JITENDRA V. SINGH
Boone, C., & van Witteloostuijn, A. (2006). Top management team composition and organizational ecology: A nested hierarchical selection theory of team and organization diversity. In: J. A. C. Baum, S. D. Dobrev & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 103–135). Oxford, UK: JAI/Elsevier. Brandenberger, A. M., & Nalebuff, B. J. (1996). Co-opetition. New York: Currency Doubleday. Bruderer, E., & Singh, J. V. (1996). Organizational evolution, learning and selection: A genetic algorithm based model. Academy of Management Journal, 39(5), 1322–1349. Brush, T. H., & Artz, K. W. (1999). Toward a contingent resource based theory: The impact of information asymmetry on the value of capabilities in veterinary medicine. Strategic Management Journal, 20(3), 223–250. Bryce, D. J., & Singh, J. V. (2001). The future of the firm from an evolutionary perspective. In: P. DiMaggio (Ed.), The twenty first century firm: Changing economic organization in international perspective (pp. 161–185). Princeton, NJ: Princeton University Press. Burgelman, R. A. (1991). Intraorganizational ecology of strategy making and organizational adaptation: Theory and field research. Organization Science, 2(3), 239–262. Burgelman, R. A. (1994). Fading memories: A process theory of strategic business exit in dynamic environments. Administrative Science Quarterly, 39, 24–56. Burgelman, R. A. (1996). A process model of strategic business exit: Implications for an evolutionary perspective on strategy. Strategic Management Journal, 17, 193–214. Burgelman, R. A. (2002). Strategy as vector and the inertia of co-evolutionary lock-in. Administrative Science Quarterly, 47, 325–357. Burgelman, R. A., & Singh, J. V. (1987). Strategy and organization: An evolutionary approach. Presented in Singh, J. V., & Burgelman, R. A., Ecology and strategy: Steps toward a rapprochement. Joint Symposium, Business Policy and Planning and Organization and Management Theory Divisions, Academy of Management National Meeting, New Orleans, LA. Clemons, E. K. (1991). Corporate strategies for information technology: A resource based approach. IEEE Computer, 24(11), 23–32. Crossan, M. M., & Hurst, D. (2006). Strategic renewal as improvisation: Reconciling the tension between exploration and exploitation. In: J. A. C. Baum, S. D. Dobrev & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 273–298). Oxford, UK: JAI/Elsevier. Czernich, C. H. (2003). When do ideas survive in organizations? Relatedness and framing of the intrapreneur. Paper presented at the Business Policy and Strategy Division, Academy of Management annual meeting, Seattle, WA. Davis, M. S. (1971). That’s interesting: Towards a phenomenology of sociology and a sociology of phenomenology. Philosophy of Social Sciences, 1, 309–344. DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48, 147–160. Dobrev, S. D. (1999). The dynamics of the Bulgarian newspaper industry in a period of transition: Organizational adaptation, structural inertia, and political change. Industrial and Corporate Change, 8(3), 573–605. Dobrev, S. D. (2001). Revisiting organizational legitimation: Cognitive diffusion and sociopolitical factors in the evolution of Bulgarian newspaper enterprises, 1846–1992. Organization Studies, 22(3), 419–444. Dobrev, S. D., & Barnett, W. P. (2005). Organizational roles and transition to entrepreneurship. Academy of Management Journal, 48(3), 433–449.
Ecology, Strategy and Organizational Change
211
Dobrev, S. D., Kim, T. Y., & Hannan, M. T. (2001). Dynamics of niche width and resource partitioning. American Journal of Sociology, 106(5), 1299–1337. Dobrev, S. D., Kim, T. Y., & Carroll, G. R. (2003). Shifting gears, shifting niches: Organizational inertia and change in the evolution of the U.S. automobile industry, 1885–1981. Organization Science, 14(3), 264–282. Dobrev, S. D., Kim, T. Y., & Solari, L. (2004). The two sides of the coin: Core competence as capability and obsolescence. In: J. A. C. Baum, & A. M. McGahan, (Eds), Business strategy over the industry life cycle: Advances in strategic management (Vol. 21, pp. 255–285). Oxford, UK: JAI/Elsevier. Eldredge, N., & Gould, S. J. (1972). Punctuated equilibria: An alternative to phyletic gradualism. In: T. J. M. Schopf (Ed.), Models in paleobiology (pp. 82–115). San Francisco: Freeman Cooper. Ethiraj, S. K., Kale, P., Krishnan, M. S., & Singh, J. V. (2005). Where do capabilities come from and how do they matter? A study in the software services industry. Strategic Management Journal, 26, 25–45. Ethiraj, S. K., & Singh, J. V. (2005). Exploration, exploitation and the adaptive potential of organizations. Unpublished paper. Department of Management, The Wharton School, University of Pennsylvania. Friedman, T. L. (2005). The world is flat: A brief history of the twenty-first century. New York: Farrar, Strauss and Giroux. Galunic, D. C., & Weeks, J. R. (2002). Intraorganizational ecology. In: J. A. C. Baum (Ed.), Companion to organizations (pp. 75–97). Oxford, UK: Blackwell. Hagel III, J. (2002). Out of the box: Strategies for achieving profits today and growth tomorrow. Boston, MA: Harvard Business School Press. Hagel III, J., & Brown, J. S. (2005). The only sustainable edge: Why business strategy depends on productive friction and dynamic specialization. Boston, MA: Harvard Business School Press. Hagel III, J., & Singer, M. (1999). Unbundling the corporation. Harvard Business Review, 77(2), 133–141. Hamel, G., & Prahalad, C. K. (1994). Competing for the future: Breakthrough strategies for seizing control of your industry and creating the markets of tomorrow. Boston, MA: Harvard Business School Press. Hannan, M. T., & Carroll, G. R. (1992). Dynamics of organizational populations: Density, competition and legitimation. New York: Oxford University Press. Hannan, M. T., Carroll, G. R., & Po´los, L. (2003a). The organizational niche. Sociological Theory, 21, 309–340. Hannan, M. T., Carroll, G. R., & Po´los, L. (2003b). A formal theory of resource partitioning. Working Paper. Graduate School of Business, Stanford University. Hannan, M. T., Carroll, G. R., & Po´los, L. (2003c). The fog of change: Opacity and asperity in organizations. Administrative Science Quarterly, 48, 399–432. Hannan, M. T., Carroll, G. R., & Po´los, L. (2004). The evolution of inertia. Industrial and Corporate Change, 13, 213–242. Hannan, M. T., & Freeman, J. H. (1977). The population ecology of organizations. American Journal of Sociology, 83, 929–984. Hannan, M. T., & Freeman, J. H. (1984). Structural inertia and organizational change. American Sociological Review, 49, 149–164. Hannan, M. T., & Freeman, J. H. (1989). Organizational ecology. Cambridge, MA: Harvard University Press.
212
JITENDRA V. SINGH
Henderson, R. M., & Cockburn, I. (1994). Measuring competence? Exploring firm effects in pharmaceutical research. Strategic Management Journal, 15(Winter special issue), 63–84. Herber, J., Singh, J. V., & Useem, M. (2000). The design of new organizational forms. In: G. Day, & P. J. H. Schoemaker (Eds), Wharton on managing emerging technologies (pp. 376–392). New York: Wiley. Kelly, D., & Amburgey, T. L. (1991). Organizational inertia and momentum: A dynamic model of strategic change. Academy of Management Journal, 34, 591–612. Lawrence, P., & Lorsch, J. W. (1969). Organization and environment. Homewood, IL: Richard Irwin. Levinthal, D. A. (1991). Organizational adaptation and environmental selection – interrelated processes of change. Organization Science, 2, 140–145. Levinthal, D. A. (1997). Adaptation on rugged landscapes. Management Science, 43(7), 934–950. Levitt, B., & March, J. G. (1988). Organizational learning. Annual Review of Sociology, 14, 319–340. Lumsden, C. J., & Singh, J. V. (1990). The dynamics of organizational speciation. In: J. V. Singh (Ed.), Organizational evolution: New directions (pp. 145–163). Newbury Park, CA: Sage. Makadok, R., & Walker, G. (2000). Identifying a distinctive competence: Forecasting ability in the money fund industry. Strategic Management Journal, 21(8), 853–864. March, J. G. (1981). Footnotes to organizational change. Administrative Science Quarterly, 26, 563–577. March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2, 71–87. Marschak, J., & Radner, R. (1972). Economic theory of teams. New Haven, CT: Yale University Press. McGrath, R. G., MacMillan, I. C., & Venkatraman, S. (1995). Defining and developing a competence: A strategic process paradigm. Strategic Management Journal, 16(4), 251–275. McKendrick, D. G., Jaffee, J., Carroll, G. R., & Khessina, O. M. (2003). In the bud? Disk array producers as a (possibly) emergent organizational form. Administrative Science Quarterly, 48, 60–93. Meyer, J. W., & Rowan, B. (1977). Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83, 340–363. Miller, D., & Friesen, P. H. (1984). Organizations: A quantum view. Englewood Cliffs, NJ: Prentice-Hall. Miller, D., & Le Breton-Miller, I. (2006). The best of both worlds: Exploitation and exploration in successful family businesses. In: J. A. C. Baum, S. D. Dobrev & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 215–240). Oxford, UK: JAI/Elsevier. Mintzberg, H. (1981). Organization design: Fashion or fit. Harvard Business Review, 59(1), 103–116. Mintzberg, H. (1992). Structure in fives: Designing effective organizations. Englewood Cliffs, NJ: Prentice-Hall. Moldoveanu, M., & Singh, J. V. (2003). The use of evolutionary logic as an explanation generating engine for the study of strategic organization. Strategic Organization, 1(4), 439–449. Nelson, R. R., & Winter, S. G. (1982). An evolutionary theory of economic change. Cambridge, MA: Harvard University Press. Platt, J. R. (1964). Strong inference. Science, 146(3642), 347–353.
Ecology, Strategy and Organizational Change
213
Peli, G., Bruggeman, J., Masuch, M., & O’Nuallain, B. (1994). A logical approach to formalizing organizational ecology. American Sociological Review, 59, 571–593. Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations. New York: Harper & Row. Philips, D. (2002). A genealogical approach to organizational life chances: The parent–progeny transfer among Silicon Valley law firms, 1946–1996. Administrative Science Quarterly, 47, 474–506. Philips, D. (2005). Organizational genealogies and the persistence of gender inequality: The case of Silicon Valley law firms. Administrative Science Quarterly, 50(3), 440–472. Po´los, L., Hannan, M. T., & Carroll, G. R. (2002). Foundations of a theory of social forms. Industrial and Corporate Change, 11(1), 85–115. Porter, M. (1980). Competitive strategy. New York: Free Press. Powell, W. W. (1990). Neither market nor hierarchy: Network forms of organization. In: B. Staw, & L. L. Cummings (Eds), Research in organizational behavior (pp. 295–336). Greenwich, CT: JAI Press. Powell, W. W. (2001). The capitalist firm in the twenty-first century: Emerging patterns in Western enterprise. In: P. DiMaggio (Ed.), The twenty first century firm: Changing economic organization in international perspective (pp. 33–68). Princeton, NJ: Princeton University Press. Rao, H., & Singh, J. V. (2001a). Sources of variation in organizational populations: The speciation of new organizational forms. In: J. A. C. Baum, & B. McKelvey (Eds), Variations in organizational science: In: Honor of Donald T. Campbell (pp. 63–91). Newbury Park, CA: Sage. Rao, H., & Singh, J. V. (2001b). Organizational speciation as new path creation: Institutionbuilding activity in the early automobile and biotechnology industries. In: R. Garud, & P. Karnoe (Eds), Path as progress (pp. 243–267). Mahwah, NJ: Lawrence Erlbaum. Ruef, M. (1997). Assessing organizational fitness on a dynamic landscape: An empirical test of the relative inertia thesis. Strategic Management Journal, 18, 837–853. Ruef, M. (2000). The emergence of organizational forms: A community ecology approach. American Journal of Sociology, 106(3), 658–714. Ruef, M. (2004). The demise of an organizational form: Emancipation and plantation agriculture in the American South, 1860–1880. American Journal of Sociology, 109(6), 1365–1410. Ruef, M. (2006). Boom and bust: The effect of entrepreneurial inertia on organizational populations. In: J. A. C. Baum, S. D. Dobrev & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 29–72). Oxford, UK: JAI/Elsevier. Saxenian, A. (1994). Regional advantage: Culture and competition in Silicon Valley and route 128. Cambridge, MA: Harvard University Press. Schroder, R. G., Bates, K. A., & Juntila, M. A. (2002). A resource based view of manufacturing strategy and the relationship to manufacturing performance. Strategic Management Journal, 23(2), 105–117. Scott, W. R. (1987). The adolescence of institutional theory. Administrative Science Quarterly, 32, 493–511. Selznick, P. (1957). Leadership in administration. New York: Harper & Row. Silverman, B. S. (1999). Technological resources and the direction of corporate diversification: Toward an integration of the resource based view and transaction cost economics. Management Science, 45(8), 1109–1124.
214
JITENDRA V. SINGH
Simon, H. A. (1957). Administrative behavior (2nd ed.). New York: Macmillan. Singh, J. V. (1990). Organizational evolution: New directions. Newbury Park, CA: Sage. Singh, J. V., & Lumsden, C. J. (1990). Theory and research in organizational ecology. Annual Review of Sociology, 16, 161–195. Singh, J. V., House, R. J., & Tucker, D. J. (1986a). Organizational change and organizational mortality. Administrative Science Quarterly, 31, 587–611. Singh, J. V., Tucker, D. J., & House, R. J. (1986b). Organizational legitimacy and the liability of newness. Administrative Science Quarterly, 31, 171–193. Singh, J. V., Tucker, D. J., & Meinhard, A. G. (1988). Are voluntary social service organizations structurally inert? Exploring an assumption in organizational ecology. Academy of Management annual meeting, Anaheim, CA. Singh, J. V., Tucker, D. J., & Meinhard, A. G. (1991). Institutional change and ecological dynamics, In: W. W. Powell, & P. J. DiMaggio (Eds), The new institutionalism in organizational analysis (pp. 390–422). Chicago, IL: University of Chicago Press. Starbuck, W. H. (1983). Organizations as action generators. American Sociological Review, 48(1), 91–102. Stark, D. (1996). Recombinant property in east European capitalism. American Journal of Sociology, 101(4), 993–1027. Stark, D. (2001). Ambiguous assets for uncertain environments: Heterarchy in post-socialist firms. In: P. DiMaggio (Ed.), The twenty first century firm: Changing economic organization in international perspective (pp. 69–104). Princeton, NJ: Princeton University Press. Tucker, D. J., Singh, J. V., & Meinhard, A. G. (1991). Founding characteristics, imprinting, and organizational change. In: J. V. Singh (Ed.), Organizational evolution: New directions (pp. 182–200). Newbury Park, CA: Sage. Tushman, M. L., & Romanelli, E. (1985). Organizational evolution: A metamorphosis model of convergence and reorientation. Research in organizational behavior (Vol. 7, pp. 171–222). Greenwich, CT: JAI Press. van Witteloostuijn, A. (1998). Bridging behavioral and economic theories of decline: Organizational inertia, strategic competition, and chronic failure. Management Science, 44(4), 501–519. van Witteloostuijn, A., & Boone, C. (2006). A resource-based theory of market structure and organizational form. Academy of Management Review, 31(2), 409–426. van Witteloostuijn, A., Boone, C., & van Lier, A. (2003). Toward a game theory of organizational ecology: Production adjustment costs and managerial growth preferences. Strategic Organization, 1(3), 259–300. Warglien, M. (2002). Intraorganizational evolution. In: J. A. C. Baum (Ed.), Companion to organizations (pp. 98–118). Oxford, UK: Blackwell. Williamson, O. E. (1975). Markets and hierarchies: Analysis and antitrust implications. New York: Free Press. Williamson, O. E. (1979). Transaction cost economics: The governance of contractual relations. Journal of Law and Economics, 22, 3–61. Williamson, O. E. (1985). The economic institutions of capitalism. New York: Free Press. Williamson, O. E. (1996). The mechanisms of governance. New York: Oxford University Press. Zucker, L. G. (1989). Combining institutional theory and population ecology: No legitimacy, no history. American Sociological Review, 54, 542–545.
THE BEST OF BOTH WORLDS: EXPLOITATION AND EXPLORATION IN SUCCESSFUL FAMILY BUSINESSES Danny Miller and Isabelle Le Breton-Miller ABSTRACT Family controlled businesses (FCBs) have been found to out-survive and out-earn non-FCBs, and their market valuations reflect that. This edge may be attributed in part to the agency- and stewardship-related consequences of ownership – consequences that via organization governance and design allow many family businesses not only to reap advantages of continuity and focus (‘‘exploitation’’), but also to reorient themselves when needed (‘‘exploration’’). These capacities rest on qualities such as owners’ discretion, knowledge and incentives, and their stewardship over the mission, core capabilities, people, and external relationships of the firm. We suggest a research agenda to investigate these issues.
This paper analyzes how the capacity for organizational change is affected by the governance of organizations, specifically, through the agency and stewardship effects of family governance and their impact on the ability of
Ecology and Strategy Advances in Strategic Management, Volume 23, 215–240 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23007-9
215
216
DANNY MILLER AND ISABELLE LE BRETON-MILLER
organizations to exploit core competencies and explore new ones. Our arguments will pull together and build on the literatures in these disparate areas, and apply them to the domain of large family-controlled enterprises (FCBs). This context, we believe, is an especially informative one as many FCBs, due to their governance and designs, appear to have an edge in exploiting distinctive capabilities and in undertaking substantial reorientations. We will thus focus on successful exemplars, not to generalize about the FCB species, but to illustrate our arguments. For purposes of this analysis, we define an FCB as a firm in which multiple members of the same family (as opposed to a single entrepreneur) have de facto control of the company and plan to remain active in the business. It is only then that many of the agency and stewardship factors we will discuss to take on relevance – factors such as owner–manager discretion, and a desire to build a business to provide security and career opportunities for later generations. Moreover, given the complex modes of strategy making, organization design, exploitation, and exploration we will be discussing, our analysis will pertain to large organizations, not small ‘‘Mom and Pop’’ operations. It will also concentrate on American and European examples. FCBs based, for example, in South America and Asia, are embedded in very different cultures and subject to very different regulatory and governance constraints. These, therefore, are beyond the scope of our analysis. In our later discussions of the boundary conditions of our arguments, we will specify more precisely the governance conditions under which we believe change-facilitating agency and stewardship effects are most apt to apply.
ORGANIZATION AND SCOPE OF THE PAPER We start with a review of some popular theories in the literature to determine the normative requirements of organizational change. Then in the two major sections of the paper, we assess how agency and stewardship aspects, respectively, acting within FCBs can address those requirements. The agency advantages of some types of family business are derived from the discretion, monitoring capabilities, and motivations of owners. We describe the resource and capability consequences of those qualities for both superior exploitation and exploration. Specific FCB governance conditions are defined under which these agency advantages are most and least likely to occur. The next major section of the paper introduces three stewardship tendencies of many family owners that shape how FCBs allocate the resources
217
The Best of Both Worlds: Exploitation and Exploration
generated by reduced agency costs. These include stewardship over the business and its mission, over employees, and over relationships with outside stakeholders. All of these forms elicit particular attributes and types of long-term investments that facilitate both exploitation and exploration, but again, only under conditions that we will specify. Figure 1 presents a schematic summary of the above relationships. It also incorporates direct relationships of agency and stewardship with performance. However, as these have been discussed exhaustively in the literature (see the reviews by Ang, Cole, & Lin, 2000, and Davis, Schoorman, & Donaldson, 1997, respectively), they are beyond our scope.
REQUISITES FOR EFFECTIVE ORGANIZATIONAL CHANGE Two common perspectives of organization change are the strategic and the ecological. The former suggests that companies must be in a constant state of learning, and that they need to remain flexible enough to adapt. It argues
Exploitation Advantages FCB Agency Effects Owner power, knowledge & incentives
• Agency: • Knowledge to deepen capability • Careful, generous investment • Stewardship: • Substantive mission & capabilities • Cohesive, stable workforce • Partnering along value chain
Performance Longevity & LT returns
Exploration Advantages FCB Stewardship Effects Over mission, employees & external stakeholders
Fig. 1.
• Agency: • Freedom to act • Knowledge to change • Slack resources for change • Stewardship: • Broad domain • Flexible, collaborative org. • Access to external resources
Agency, Stewardship, Change and Performance in FCBs.
218
DANNY MILLER AND ISABELLE LE BRETON-MILLER
that managers need to keep seizing new opportunities, finding new market niches, and renewing their competencies (see review by Finkelstein & Hambrick, 1996). The entrepreneurship literature, for example, stresses adaptability as the soul of survival (Zahra, 2003). The literature on competitive analysis too advocates that firms constantly analyze their environments to discover niches they can defend against emerging competition (Porter, 1996). And the resource-based and dynamic capability perspectives claim that firms need to build a core capability that they can keep developing faster than any competitor (Barney, 1991; Miller, 2003; Teece, Pisano, & Shuen, 1997). All these perspectives imply that firms thrive when their strategies constitute a mix of exploiting existing competencies and exploring for new ones. The ecological view is very different. It suggests that inertia is good for organizations – that organizational routines, specialization, focus on known core skills, and stable relationships inside and outside the organization are all highly functional and economical (Hannan & Freeman, 1984). Indeed, ecologists believe that the inertia of the firm typically exceeds that of the environment. They believe that executives and their organizations have a very limited capacity to change as fast as the conditions and demands of the environment. There are cognitive and motivational limitations, resource restrictions, and inevitable uncertainties in the environment, that make it unlikely that firms will be able to adapt successfully – indeed that managers will even recognize the need to change (Barnett & Carroll, 1995; Baum, 1996). In fact, given the costs entailed by the process of change, by the time cognitive and political obstacles to change are overcome, firm resources may be inadequate to the challenge of change (Baum & Singh, 1994; Miller & Friesen, 1984). And even if change were initiated in a timely fashion there remain potentially catastrophic disruptions with layoffs, brutal reorganizations, and new skills and alliance requirements. In short, ecologists argue that firms are often limited to exploitation of their current competencies. The empirical literature on change suggests that there is truth to both strategic and ecological views: some firms do change in a useful and significant way, but this is indeed a risky process. It is noted that firms often recognize the need for change too late, after their situations have deteriorated significantly (Barnett & Carroll, 1995; Miller, 1990; Singh, 2006). So punctuated equilibrium models of change have organizations alternating between very long periods of capability exploitation or inertia and rarer, often tardy, periods of exploratory reorientation (Singh, 2006). They also find that reorienting exploration is associated with a high risk of failure
The Best of Both Worlds: Exploitation and Exploration
219
(Miller & Friesen, 1984). Indeed, even less dramatic changes can be problematic. Faddish expedients such as quality programs and downsizing frequently involve costs that exceed benefits, while acquisitions often destroy vital complementarities. In short, change, while sometimes salutary, is a high-risk endeavor. Although the strategic and ecological schools disagree about an organization’s capacity to change, they do not diverge substantially in their view of the requisites for effective change: namely that a firm’s controlling owners or their managerial agents have the discretion, knowledge, incentives, and resources to be able to adapt effectively. As we will argue, family-controlled businesses often have an edge in satisfying all these conditions. FCBs have been shown in emerging research to enjoy greatly superior longevity: 24 years on average, versus 12 years for the typical non-family company (de Geus, 1997; Mackie, 2001; Miller & Le Breton-Miller, 2005). They also enjoy higher returns (Anderson & Reeb, 2003; Bornstein, 1983; McConaughy, Walker, Henderson, & Mishra, 1998; McConaughy, Matthews, & Fialco, 2001; Villalonga & Amit, in press). We will argue that this healthy longevity accrues to family firms because, subject to conditions we shall be discussing, some are able to do two things especially well – ‘‘exploit’’ effectively and extend an existing capability, and move toward a viable new capability when and if such ‘‘exploration’’ is required. By exploitation we mean the profitable pursuit of a stable core capability: typically by sustaining and deepening a focal competency, and leveraging it across related opportunity sets. By exploration we mean the significant reorientation of a company’s capabilities to embrace new competencies and sources of competitive advantage (March, 1991). Our thesis is that certain kinds of FCBs, due to the agency and stewardship consequences of their modes of governance, take a long-term view of the business that facilitates both exploitation and exploration. First, because of the reduced information asymmetries and better alignment of incentives between owners and managers, some FCBs minimize agency costs and opportunistic resource allocations. This frees up resources to build and adapt the business. Second, in numerous FCBs, those in control are good stewards – they care deeply about subsequent generations, and therefore about the future. So they are willing to invest in the resources, capabilities, employees, and external relationships that will sustain a company for years to come. Many of those investments and the resulting capabilities, as we will see, are Janus-faced, in the sense of being able to extend and reorient capabilities.
220
DANNY MILLER AND ISABELLE LE BRETON-MILLER
Agency Advantages in Exploitation and Exploration The most common agency problem in public companies is that managers, using their superior or ‘‘asymmetric’’ information, are able to expropriate resources and therefore rents from the owners (Demsetz, 1988; Ang et al., 2000). These agency costs can be especially high where ownership is diffuse and shareholders do not have the power to police their managerial agents (Jensen & Meckling, 1976). Agency considerations often manifest in top managers being more concerned with feathering their nests than securing the long-term future of their organizations (Jacobs, 1991). The problem has been exacerbated as CEO tenures have fallen in the last 15 years from almost 10 to less than 4 years (Lansberg, 1999). Also, stock options based on short-run performance now account for the lion’s share of top management compensation (Khurana, 2002). The result is that the leaders of many public companies are fixated on quarterly earnings, and to boost returns they favor expedients such as acquisitions and downsizing gestures that earn quick benefits but incur steep long-run costs (James, 1999). Agency arguments suggest that this behavior may be rarer in some kinds of family businesses (Miller & Le Breton-Miller, 2005). There, concentrated ownership allows a family to control enough votes or shares to appoint or serve as directors or top managers, and thus gives them a great deal of discretion – specifically, the power to police their managerial agents (Demsetz, 1988; Jensen & Meckling, 1976).1 Many family owners, especially a founder and his long-apprenticed kin, are also more apt to have significant depth of knowledge about the business (Ward, 2004), so information asymmetries between agents and owners are reduced or eliminated; this too improves monitoring capability (Ang et al., 2000). Finally, the incentives of family managers may differ from those of remote shareholders. The former typically have more invested – both financially and personally in the business, and thus are motivated to act for its long-run success. We will argue that these three qualities – family owner discretion, knowledge, and incentive – reduce agency costs, thereby generating surplus resources, and contributing both to the capacities to exploit core competencies, and to explore for new ones (see Table 1). The next section on stewardship will say more about the uses of these surplus resources. Owner Power and Discretion Family owners who control a commanding share of a business have enormous sway. Some assume the position of chairman or CEO, while others
221
The Best of Both Worlds: Exploitation and Exploration
Table 1. Agency Advantages in FCBs. Exploitation
Exploration
Theme: Concentration of ownership & owner knowledge of business reduce waste and monitoring costs; free up resources, allow courageous decisions Power of family owners: Major owners who control a firm can act as or put in place secure, long-tenured managers whose extended time horizons allow them to implement and thoroughly exploit a strategic capability. Long tenures reap the fruits of stability and exploitation.
Powerful owners and their managers can act boldly and quickly when needed, to make courageous decisions to renew the firm (without career jeopardy)
Reduced information asymmetries and monitoring costs: Owners deeply understand and often run the business – this lowers ‘‘monitoring costs’’, thereby generating superior resources vis-a`-vis firms run by small, remote shareholders. Additional resources allow more investment in the enterprise, while owner knowledge increases the appropriateness of investment.
Slack resources available when needed to fund periods of upheaval. Knowledgeable owners are also in better position to effectively monitor change initiatives (change intelligently to be able to survive)
Owner incentives: Family owners also have their fortunes, reputations and future generations tied to the soundness of the business. Thus, they may have an incentive to invest for the long run and avoid wasteful fads that often entice and entrap careerist managers. Resources are usually focussed on specific industries and capabilities
Some owners are willing to invest more deeply in the long-run future of the business, and are meticulous in their change endeavors
appoint agents who closely pursue their interests. The power held by these major shareholders enables them to keep their managerial agents in line. This reduces agency costs that come in the form of managerial side-payments, inefficiency, or opportunistic initiatives that are good for managers, but harmful to the long-run interests of the firm (Demsetz, 1988; Jensen & Meckling, 1976). Lower agency costs, in effect, free up resources that can be invested to enhance company capabilities (Hoopes & Miller, in press; Makadok, 2003). Their ample discretion also allows family owners or their executives to remain in office as executives or directors for a long time – typically, 3 to 5 times longer than at non-FCBs (Ward, 2004). Aspirations for intergenerational continuity within many family businesses reinforce such long tenures.
222
DANNY MILLER AND ISABELLE LE BRETON-MILLER
Although there can be disadvantages such as stagnation and insularity associated with long executive tenures, one common positive consequence of these tenures is longer business and investment time horizons (Miller & Shamsie, 2001; Singh, 2006; Ruef, 2006). By contrast, at many public nonFCBs, less secure executives with far shorter tenures often avoid long-term investments in order to augment quarterly earnings and please impatient shareholders (Jacobs, 1991). Surplus resources coupled with longer investment time horizons and the discretion and willingness to invest in the business can facilitate both capability exploitation and exploration. Exploitation benefits from long-term investment and the ability of long-tenured executives to stay focussed on developing and leveraging a core capability that is near and dear to them (Miller & Le Breton-Miller, 2005). Such focus can be maintained as fewer transfers of power minimize the likelihood of corporate shake-ups (Wiersema & Bantel, 1992). Also there is less need to please short-sighted shareholders by engaging in competency-eroding distractions such as downsizing or unrelated acquisitions (Morck, Shleifer, & Vishny, 1990). Another consequence of family CEO discretion at FCBs is that leaders have the freedom to explore courageously to renew the firm. They can act quickly and in unconventional ways to explore new paths for developing and reorienting the company, unlike many CEOs at public companies who are forced to get board approval for major changes and focus more on matters of the moment.2
Reduced Information Asymmetries and Monitoring Costs Unlike the small shareholders of public companies, major family owners tend to know a good deal about their businesses. Many are company founders or have grown up in the business, having had ample apprenticeships in the ways of the firm and its industry (Lansberg, 1999; Ward, 2004). The resulting knowledge enables such owners to closely monitor their managerial agents and ensure that their behavior aligns with the long-term interests of the business (Demsetz, 1988; Jensen & Meckling, 1976; Ang et al., 2000). Again, the implication is that knowledgeable monitoring decreases agency costs of managerial opportunism, shirking, and waste (Demsetz & Lehn, 1985; Claessens, Djankov, Fan, & Lang, 2002). As noted, reduced agency costs free up company resources, so that, ceteris paribus, many FCBs will be more resource rich than their less astutely monitored non-FCB rivals, and thus more able to exploit their capabilities (Hoopes & Miller, in press).
The Best of Both Worlds: Exploitation and Exploration
223
Opportunistic exploration initiatives that change a firm merely for the benefit of its managers (e.g. via dangerous acquisitions or downsizing) are less apt to occur in the presence of profound owner knowledge. Indeed, FCBs have been shown to be especially resistant to fleeting management fads and wasteful change efforts (Miller & Le Breton-Miller, 2005). Moreover, a combination of superior owner and managerial knowledge may reduce the uncertainty inherent in most exploration initiatives. Motivation and Incentives Family owners have a lot at stake in their business: they have much of their fortunes invested there, as well as their reputations, and sometimes, too, the futures of their children. Again this suggests that the resources generated by lower agency costs are unlikely to be squandered (Hoopes & Miller, in press; Ward, 2004). As noted, many family owners, strive for the long run success of their enterprises. Thus, FCBs are left not simply with the surplus resources discussed above, but those that will be invested to strengthen core competencies for years to come (Dieryckx & Cool, 1989; James, 1999; Miller, 2003; Teece et al., 1997). These motivations not only foster generous investment in a firm’s health and capabilities, they also assure greater meticulousness in any change endeavors. Qualifications Some FCB governance arrangements are not at all conducive to agency advantages. For example, where ownership is divided among feuding family factions, or where family managers have too few shares to exercise much discretion, the CEO may have inadequate discretion to manage (Ward, 2004). This becomes more likely in second and later generations of the business as ownership becomes more dispersed (Gersick, Davis, Hampton, & Lansberg, 1997; Ward, 2004). Agency advantages also may fade with too much CEO discretion – when family CEOs have the status and control to do whatever they want because of their dominant status or ownership. If such leaders are rash or myopic in how they run the business, it is hard for anyone else to set things straight. Parties with unchallenged control and little chance of opposition can abuse that power by taking resources out of the business at the expense of other shareholders (Claessens et al., 2002; Schulze, Lubatkin, Dino, & Buchholtz, 2001; Schulze, Lubatkin, & Dino, 2003). This is most apt to happen when
224
DANNY MILLER AND ISABELLE LE BRETON-MILLER
there are no influential outside shareholders or directors (Anderson & Reeb, 2004). A special class of agency problems arises when family incentives are not aligned with those of minority shareholders (Claessens et al., 2002), for example, when the family controls the firm with minimal investment and financial devices such as super-shares, and a pyramidal hierarchy of holding companies (Morck & Yeung, 2003). Firms at the top of the pyramid can control those at the bottom with little ownership, giving them a strong incentive to transfer costs down the pyramid, and revenues upwards to those companies they have the most invested in. Finally, FCB agency advantages tend to weaken when the CEO is an outsider. The latter may not have as much incentive, knowledge or freedom to realize the agency advantages we have discussed.
Summary Agency advantages can accrue when a family has the power, incentive, and knowledge to closely monitor the business. Superior monitoring prevents opportunistic behavior by managerial agents, thereby creating a potential resource advantage. Subject to the important qualifications noted above, the motivations of family owners often make it more likely that surplus resources will be invested for the long-run benefit of the business (see Table 1). As we will see in the next section, such resources and the capabilities they engender can be used both to build and exploit a stable set of capabilities; and for exploratory renewal.
STEWARDSHIP ADVANTAGES IN EXPLOITATION AND EXPLORATION Agency theory embraces assumptions of self-interest and economic rationality to derive its conclusions. But a very different paradigm, stewardship theory, posits that many company executives and owners are driven instead by motivations such as altruism, fairness, and pride in one’s work (Davis et al., 1997). These differences notwithstanding, we will argue that stewardship theory, like agency theory, suggests potential exploration and exploitation advantages for certain types of family businesses. It also helps to
The Best of Both Worlds: Exploitation and Exploration
225
clarify just how the resources generated and invested for the long-run benefit of the firm are most apt to be allocated. In other words, whereas agency theory suggests how and why FCBs will generate excess resources for exploitation and exploration, stewardship theory helps to explain in more detail how these resources will be allocated.
Three Stewardship Advantages at some FCBs The primary assumption of stewardship theorists is that people do not simply behave according to narrow self interest, but are motivated to make a substantive social contribution – to their family, company mission, co-workers, and society at large (Davis et al., 1997; Donaldson & Davis, 1991). Examples of stewardship behavior by owners and their managers include (1) cherishing a socially valuable mission and doing an exceptional job of fulfilling it – e.g., in quality, design, or service delivery; (2) looking after one’s employees and fostering their fruitful development; and (3) dealing generously with external stakeholders such as customers, business partners, and the community. We expect that such stewardship behavior will be especially prevalent where firms are owned by those who are personally connected to the organization and therefore have it and its stakeholders’ best long-term interests at heart. Therefore, values tend to be important and shared; thus the work is more meaningful, and social relationships are more substantive and enduring. Exactly these conditions obtain in many long-thriving family businesses where the family is deeply involved (Miller & Le Breton-Miller, 2005). First, a family is often strongly committed to the enterprise and its mission, which, as noted, is closely tied to its economic well-being and reputation. So it is apt to have generous investment in the mission of the business and the capabilities needed to attain it. Second, because the family so values the business and mission, it takes care to hire, train, and nurture employees with the right values, talents, and attitudes, and who will use them for the benefit of the company. And these people are treated well. Finally, stewardship can show up in relationships with outside stakeholders – customers who are the lifeblood of the business, suppliers who provide key inputs, and partners who help leverage competencies. All of these assets: core capabilities, a talented corps of personnel, and enduring relationships with outside stakeholders, can facilitate both exploitation and exploration. Stewardship, however, is by no means always present in family businesses, and again we will take pains to qualify our discussion by defining the conditions under which it applies.
226
DANNY MILLER AND ISABELLE LE BRETON-MILLER
Stewardship over Mission and Capabilities Many families care deeply about the mission of the business and what it contributes to the world. It is their values, their reputation on the line. And they feel that in their bones. The Mercks, Johnsons, and Lillys in pharmaceuticals, the Timkens in steel bearings, and the Sulzbergers and Grahams in publishing, for example, were in business not only to make money but to leave a positive mark on the world. The Mercks invested deeply in research projects and commercialization efforts that addressed the most urgent medical problems – and they speeded products to market only after the most exacting testing and enormously steep investments (Collins & Porras, 1995). The Timkens had as their objective to overcome mechanical friction and accelerate the technological development of the century (Pruitt, 1999). And the Sulzbergers wanted to publish a newspaper that would serve as the pillar of an enlightened electorate – nothing less (Tifft & Jones, 1999). These missions drove deep investment in related capabilities, and long periods of financial frugality, sacrifice, and ploughing back of earnings (Miller, 2003). Roots Many family business owners see the business as an extension of themselves, a reflection of their values, and a source of personal pride and reputation (Ward, 2004). They also see it as a pillar of the family future. It is natural then for them to invest generously and patiently in the company’s mission and its required core competencies (Kang, 1998). And because reputation is at stake and personal ethics drive these businesses, the values of leading family members figure prominently (Gersick et al., 1997). In the companies just mentioned, founder values and a desire to make an important social contribution were far more important determinants of the strategies of the company than any financial target. Economic performance was viewed mostly as the means to a more substantive end. It may be as well that in family businesses, it is easier than elsewhere for a founder to imprint his or her values upon the organization. There is frequently a close personal connection with the upcoming generation of leaders, who tend to be family members. Lengthy apprenticeships served under a related family founder can deeply engrain and so propagate company values, cultures, and routines. Effects on Exploitation Owners’ higher levels of commitment give many of these companies an enormous competitive edge. Owners are willing to make investments that are (a) generous, (b) focussed on a core capability tied to the mission, and
The Best of Both Worlds: Exploitation and Exploration
227
(c) intended for long-run success. Empirical research suggests that family businesses do indeed outshine their non-family counterparts in all of these respects. Kang (1998) found that FCBs out-invest non-FCBs in modern capital equipment and plant. Weber et al. (2003) have shown that FCBs outinvest their counterparts in R&D. And Amihud and Lev (1999), Khanna and Palepu (2000) and Morck et al. (1990) have shown that concentratedownership businesses (such as most FCBs) are more focussed in their acquisition and growth trajectories and avoid being side-tracked into unrelated ventures. The literature on dynamic competency creation and core capability emphasizes the importance of exactly such investment patterns of concerted, focussed capability building (Dieryckx & Cool, 1989; Miller, 2003; Teece et al., 1997). It suggests that companies can best achieve competitive advantage when they pursue a persistent, dedicated and cumulative investment trajectory to build capabilities. This keeps the firm ahead of its rivals, and the depth of the resulting capability allows firms to leverage their special talents by exploiting them in different products or market niches. Timken is a good example. It invested generously in improving the quality of its bearings, spending 2–5 times the industry average on product and process R&D. It achieved a 16-fold increase in bearings durability in 20 years, and was able to manufacture its steel, the finest in the world, using one-third of the labor hours per ton of the best Japanese mills of the day (Bagsarian, 1997; Pruitt, 1999). But the firm was highly focussed in its investments, spending relatively little on administrative overhead and nonessential functions, and ploughing resources instead into capability building. Core capabilities were then leveraged across every conceivable sort of tapered bearing, from those of a few ounces, to those of over a ton, for virtually every industry, and for most countries in the world. Timken also sold its perfect steel to its most demanding competitors (Bagsarian, 1997; Vinas, 2002). Effects on Exploration Deep investment in the capabilities tied to a cherished mission can also facilitate exploration and renewal. Because concern is with long-run performance, the emphasis is not on specific products or markets but missionrelated capabilities – which apply to a wide variety of markets and products. Indeed, generous long-term investment, particularly in process or technological R&D, often pulls companies into propitious new markets. Timken’s metallurgical R&D for its bearings design led it to alloys and processes for making better steel and seamless tubing, which was ideal for the emerging jet
228
DANNY MILLER AND ISABELLE LE BRETON-MILLER
propulsion business. Corning’s profound research into glass enabled it to build on its light bulb expertise to create radio tubes, and then television picture tubes. Its other discoveries in glass helped it create the first fibreoptic cable, and its expertise there led to the photonics business for gene manipulation. Although all investments centered on Corning’s core glassrelated capabilities, they have pulled the company into the markets of the future for more than a century (Graham & Shuldiner, 2001). Many FCBs also derive exploratory advantage simply by building slack resources within the company, thereby creating a pool to fund new projects and survive periods of upheaval (Dreux, 1990; Gallo & Vilaseca, 1996). Qualifications Not all family businesses embrace their mission or invest generously in capabilities. Later generations of a family may not identify closely with the business or mission, especially if they are less involved in the company and are desirous of generous dividends (Lansberg, 1999). And when ownership is divided among family members who disagree about the future of the business and how profits should be allocated, there is less scope for making longterm investments (Lubatkin, Ling, & Schulze, 2002). In short, stewardship over the firm and its capabilities is most likely to be manifested when the founder is still present and plans to keep the firm in the family, when family owners agree about the mission, and when family members are close enough to the business to serve as caring directors and managers (Villalonga & Amit, in press). People and Organizations Where owners and managers are committed to the mission and long-term success of their businesses, they are also apt to be especially solicitous of the quality of their human resources. They will take care to recruit those who share their values and can be trusted to contribute fully to the company. So hiring practices at many family firms are extremely selective, and great emphasis is placed on employee socialization and training. Nordstrom and Timken, for example, hired only one out of every 100 applicants, and put them through training programs that were 3 to 4 times longer than industry norms (Spector & McCarthy, 1995; Pruitt, 1999). This behavior is not uncommon among FCBs. Systematic comparative studies have shown that FCBs invest more intensively than other organizations in their people: in training, selective hiring, and long-term benefits (Allouche & Amann, 1997;
The Best of Both Worlds: Exploitation and Exploration
229
Reid & Harris, 2002). To preserve morale, these businesses also minimize the gap between the highest paid executives and line-level workers (Allouche & Amann, 1997). Other comparative studies found that family businesses are more value-driven than non-FCBs – guided more by norms and socialization than financial incentives or bureaucratic controls (Guzzo & Abbott, 1990). With motivated, well-trained employees, companies can avoid the sluggishness, de-motivation, and costs of bureaucracy, as well as the gamesmanship and fragmentation caused by an over-reliance on financial incentives. Thus FCBs, as a group, have been shown to employ flatter organization structures, broader job definitions, and greater reliance on employee initiative (Beehr, Drexler, & Faulkner, 1997; Donckels & Frohlich, 1991; Goffee & Scase, 1985; Guzzo & Abbott, 1990; Ward, 2004). Less bureaucracy and fragmentation also makes for better collaboration among people from across functions and divisions (Pfeffer, 1998). Roots There is often an emotional connection between a family and the employees who safeguard its most valued assets: company, mission, and reputation. First, the family wishes to recruit, train, and motivate people who will build the business with their diligence and initiative. They want to nurture employees who will augment the performance of their enterprise through better quality work and service. Second, the family often feels gratitude and responsibility toward its employees and so treats them well. That treatment invites reciprocation. Finally, some family owners reside in a community that identifies them with their businesses. So they strive to be exemplary employers (numerous generations-old FCBs such as S.C. Johnson have never laid anyone off (Miller & Le Breton-Miller, 2005). Effects on Exploitation Well-trained and well-treated employees tend to be more productive, and careful selection and socialization ensure that superior talents can be applied to building and leveraging core capabilities. A motivated and loyal workforce is less subject to turnover, thereby preserving and extending the tacit knowledge that resides in organizational teams (Knott, Bryce, & Posen, 2003).3 Better training translates into deeper knowledge and understanding, and an ability to use that knowledge to augment the capabilities of the firm and so improve processes and products (Miller, 2003). Better treatment gives people an incentive to use their talent to help the firm (Pfeffer, 1998).
230
DANNY MILLER AND ISABELLE LE BRETON-MILLER
At Timken’s Faircrest plant, every employee could do every job in the plant: self-directed work teams and project teams were charged with particular improvement initiatives. The team members rotated across six role positions in which they assumed responsibilities that once had been those of the supervisors: equipment, quality, cost, safety, etc. Decision-making authority was with front line workers. So anyone with an idea could mention it to the supervisor and put it into effect immediately, without going up the chain of command. The result was that within 18 months of the plant’s opening, the employees themselves had formed teams that reduced costs by 25%, customer rejects by 62%, and downtime from 600 to 175 hours per year. Worker teams also increased production by 38%. All of this was accomplished without a penny of new capital expenditures (Sheridan, 1992; Lund, Bishop, Newman, & Salzman, 1993). The result was that Timken remained a quality leader in its industry for decade after decade, while at the same time it was one of the most efficient manufacturers in the business – ‘‘exploiters par excellence’’. Effects on Exploration Assiduous hiring, training, and socialization enable firms to function in a way that recalls more ‘‘organic’’ forms of organization (Burns & Stalker, 1961). Specifically, the skills and motivations of employees allow firms to define jobs more broadly and flexibly, employ flatter hierarchies with fewer bureaucratic constraints, and accord more discretion to people at all levels. The resulting organization constitutes an ideal climate for innovation and discovery as it affords more scope for initiative, and facilitates easier lateral communication and collaboration. Initiative promotes innovation at all levels, while the lateral organization facilitates the collaboration across functions and units that is so critical to commercialization (Galbraith, 2000; Lawrence & Lorsch, 1967). Exploration and organizational renewal are thus greatly facilitated. W.L. Gore & Company, the maker of gore-tex and myriad related products, selects its employees according to their expertise as well as their entrepreneurial personalities and social skills – it then nurtures and mentors them carefully. The organization is very flat, and new hires are encouraged to define their own very broad roles and projects. Each staff member is put together with a mentor with whom to work closely and learn about company culture, values, and behavioral expectations. The mentor also provides personal contacts to help the new person to network and get things done. New hires are then expected to launch a new product or marketing project, and must attract experts in complementary functions to the project by
The Best of Both Worlds: Exploitation and Exploration
231
convincing them of its merits. The majority of employees consider themselves ‘‘leaders’’, and all work in units of less than 200 people that contain all the major business functions. For decades now, the level of new product and new market innovation has been outstanding (Dominguez, 1998; Harrington, 2003). Qualifications Some FCBs do not treat their employees especially well, and that can limit their exploration capabilities. This, indeed, is often the case at companies that are efficiency- and low-cost driven, and where employees, for the most part, perform very routine functions and do not interact with clients. Examples are Tyson Foods and Wal-Mart, family businesses that are hardly generous places to work (Miller & Le Breton-Miller, 2005). Similarly, families running holding companies, or who are remote from the business and its mission will tend to be less solicitous of their workers. We should point out, as well, that care for employees has a downside: it is costly, certainly in the short run, and it limits an organization’s ability to engage in radical cutting measures when organizations face crises.
Relationships with External Parties Attitudes of stewardship over the long-term success and reputation of the business and the family drive some FCB owners and managers to invest deeply in relationships with outside stakeholders: customers, alliance partners, suppliers, government, and even the community at large. Relationships are to be contrasted with transactions: the former are more open-ended, evolving, enduring, and trust-based, and less opportunistic and shortsighted (Das & Teng, 2001). Miller and Le Breton-Miller (2005) have found that successful family businesses were unusually generous in investing in long-term relationships with suppliers and customers. Some dedicated special resources to individual clients to cement client loyalty; others would invest in the operations of suppliers to make them better providers. Others still had ongoing joint ventures, which were over 60 years old. Roots Many family owners want to form strong enduring relationships with outside parties and stakeholders to support the business during the long haul and increase its soundness and resilience for subsequent generations. They want, in short, to build social capital: preferential treatment or access to
232
DANNY MILLER AND ISABELLE LE BRETON-MILLER
resources based on positive prior interactions or reputation (Adler & Kwon, 2002; Barney & Hansen, 1994; Bourdieu, 1980; Gomez-Mejia, NunezNickel, & Gutierrez, 2001). Family owners also have special strengths in being able to develop such capital. First, outsiders know where the buck stops – family leaders tend to be there for a long time, are profoundly empowered, and care about the reputations of the companies. This makes them especially credible in negotiating long-term, open-ended deals (Fukuyama, 1995; Palmer & Barber, 2001). Second, many FCBs are future-driven – so they invest more generously early in a relationship, for example in customizing products or dedicating facilities for a major client (Bubolz, 2001; Naphiet & Ghoshal, 1998). Third, many family companies are private. Where these businesses operate in sensitive areas, they have an edge in preserving secrecy (McCartney, 1988). Finally, FCBs can transfer the social capital and relationships of one generation to the next (Braudel, 1984). In short, many FCBs not only have a strong incentive to embark on long-term relationships, they also have a real edge in developing these relationships.
Effects on Exploitation Enduring relationships such as those with long-term clients give firms advantages in such quintessentially exploitative activities as cross selling, supply chain-integration, and partnerships for customizing products. Broader, more trusting relationships with suppliers enable buyer and seller to work together to find areas of complementarity for building on and complementing one another’s skills. For example, Sam Walton at Wal-Mart got supplier P&G to form dedicated cross-product and cross-functional teams that would determine stocking strategies for each store – for all products (including non-P&G products) in a product category. Trust between the two parties allowed P&G to share with Wal-Mart its data on demand demographics for its products, and link them to store by store point-of-sale data. Together, the firms also developed automatic stocking and logistics systems that shaved inventory costs by 10%. Wal-Mart could exploit P&G’s logistical expertise, while P&G could expand its business with the retailer. Each accelerated and benefited from the others’ learning. FCBs also form relationships with customers: Este´e Lauder’s sales representatives make every effort to document the preferences of their 100 best customers, and contact these people on a regular basis to get feedback, help re-supply items, and inform them of new offerings and sales. Nordstrom and L.L.Bean associates keep detailed files on their better customers, again to broaden and extend the relationship with the client by having a more
The Best of Both Worlds: Exploitation and Exploration
233
responsive approach. Such relationships are leveraged across a broad range of products. Effects on Exploration If rich, enduring relationships drive exploitation, they also aid in exploration. They provide revealing market contacts and privileged access to the knowledge, capital, and expertise needed to recognize new opportunities, adapt, and change. Wal-Mart’s ties with P&G provided ideas for new approaches and processes. The close customer contacts of Este´e Lauder’s sales associates and executives kept them in touch with the shifting demands of the market, and helped to guide the changes in its product lines and marketing strategies. Some alliances help more radical forms of innovation and reorientation. The Houghton family, who controlled Corning for over 100 years, stimulated key innovation partnerships with companies such as Owens, Dow, etc., to pioneer new categories of products such as glass panel displays and fiber optic technology. Corning today has dozens of innovation partnerships that not only allow companies to pool their competencies and develop new ones, but also to reduce the risks of innovation by combining resources. Qualifications FCBs are less apt to pursue long-run relationships with stakeholders when they care little about building social capital or are not concerned about the business’s ability to support a future generation of family members. Unstable ownership and management also thwart long-term relationships, as outsiders have no one with clout and reliability to depend on. Finally, strategy and type of business can figure in the pursuit of long-term relationships. Companies like Wal-Mart that strive to excel via value chain coupling benefit from strong outsourcing relationships. Innovators such as Corning seek alliances with technology companies and universities. And companies like Bechtel, Tyson Foods, and Bombardier that pursue very large and one-of-a-kind projects, desire enduing relationships with major clients. But not all strategies are so relationship dependent. Table 2 summarizes our stewardship arguments.
Alternative Explanations of FCB Performance It might be suggested that it is not so much agency and stewardship advantages that make FCBs more long-lived and adaptive – but rather their small
234
DANNY MILLER AND ISABELLE LE BRETON-MILLER
Table 2.
Stewardship Advantages of FCBs.
Exploitation
Exploration
Stewardship: Incentive of owners and executives to protect firm, heirs, and reputation. Therefore have long investment time horizons for capabilities, people, and partnerships Mission and related capability: Care for substantive mission, focus, careful stewardship over resources
Broad domain definition – skills and learning, not products or markets; exploration based on core firm capabilities
People and organization: Stable work force, efficient, low turnover, well-trained – many disciples of mission and core competency
Talent and training; flat, org. structures, job flexibility; staff motivation to use initiative & spot opportunities, good collaboration in change endeavors
Relationships with external parties: Favor relationships over transactions; unusually generous. Reap customer loyalty, preferred access to suppliers, old partnerships and social capital
Ability to get resources, keep in close touch with environment, share the risks and resources of reorientation
size. Small firms may favor informality and simplicity, and eschew bureaucracy and hierarchy, making them less subject to inertia. The limited resources of such firms, however, and their tendency to be young and subject to liabilities of newness may make significant change very risky. Also, the best research demonstrating the superior longevity and performance of family businesses is on larger FCBs – those on the Fortune 500 list (Anderson & Reeb, 2003; Villalonga & Amit, in press; Miller et al., 2005). Here, small size is by no means a driver of superiority. An alternative argument for the greater longevity and superior performance of FCBs is that they grow less quickly and take fewer risks. That notion is belied, in part, by the fact that FCBs seem to show higher market valuations than non-FCBs on the Fortune 500 (see Anderson & Reeb, 2003; Villalonga & Amit, in press). Economic theory would suggest that it is the faster growing companies that have the highest market valuations (Tobin’s q) (Demsetz, 1988). As for risk taking, Miller and Le Breton-Miller (2005) found that their sample of high performing FCBs did take significant risks – in the form of long-term investments in capabilities, research, and infrastructure with uncertain payoffs – but they often did so with a relatively conservative capital structure. Risk, in other words, is a multifaceted notion. While some aspects of FCB conservatism might slow growth, others would tend to accelerate it.
The Best of Both Worlds: Exploitation and Exploration
235
CONCLUSION In our introduction we noted that although the strategic and ecological paradigms disagreed about an organization’s ability to change, they both agreed that effective change would require of any company the following: the incentives, discretion, and knowledge to change, as well as the resources to do so. Our discussion of agency has suggested that, subject to the qualifications we have outlined, the owners and leaders of many family businesses will indeed meet these requirements. And in our discussion of stewardship we have argued that firms will be able to develop the core capabilities, people and organizations, and external relationships that improve their abilities at both exploitation and exploration. Long-term investments in capabilities may not only be used for today’s products and markets but for capabilities that will determine tomorrow’s products and markets. Cohesive, energized organizations permit not only efficient operations and continuous improvement, but flexibility that flows from broad job definitions, flat structures, informality, and the freedom to pursue experimental initiatives. Close relationships with outside stakeholders not only afford access to resources but also to information about important challenges and opportunities, and partnerships that allow those opportunities to be seized more surely. These potential advantages of exploitation and exploration (see Table 3) may help to explain the superior Table 3.
Summary of Exploitation and Exploration Advantages at some FCBs.
Basis for Exploitation Advantages
Basis for Exploration Advantage
(1) Discretion to resist dangerous, resource depleting, short-sighted moves (fads etc.), and to act for the long run (a, s)a (2) Focus on the core competencies and capability building, tied most closely to their mission (s); (3) Deep investment in corporate cultures to get a stable and loyal workforce (a, s); (4) Invest for the long run in stable relationships with outside parties (s)
(1) Discretion to act fast when they have to (a) (2) Do not stray far from their competencies when it is time to change (a, s) (3) Slack resources (due to agency advantages and stewardship concerns) for experimentation and learning (a, s) (4) Informal, loosely coupled structures and cultures with initiative and motivation (s) (5) Stay in touch with environment by forming close links with external stakeholders (s)
a
a, s refer to agency and stewardship sources of advantage.
236
DANNY MILLER AND ISABELLE LE BRETON-MILLER
longevity of many family firms. We need to be clear, however, that given the different requirements of exploration and exploitation, FCBs are not apt to be able to concentrate on both initiatives at once. Moreover, we expect only firms that satisfy the governance requirements we laid out to enjoy the advantages we have described. Our analysis here is certainly speculative, and based mostly on anecdotal evidence. To take this research to the next step, scholars have a variety of directions they may wish to explore. First, it would be useful to determine how differences in agency-related aspects of governance – level of family ownership, presence of founder, family management, intention to keep the family in the business, can influence the nature and time horizons of company investments, the boldness and speed of executive action, and the level of accretion of slack resources. It would also be of interest to see how such governance dimensions might influence stewardship – over the mission, over employees, and over external relationships of the firm. Is family ownership or management apt, for example, to be associated with more substantive, less financially driven missions, more generous, long-term and flexible human resources policies, or longer and more cohesive external partnerships? Such outcomes in turn can be related to the nature of an organization’s exploration and exploitation efforts – as gauged, say, by investments in core capabilities, in product and market renewal, and in change efforts. Finally, these outcomes can be examined for their association with long-run financial returns, company longevity, and, where relevant, market valuations. The proposed research, qualitative or quantitative, would have to incorporate adequate variation in family ownership and management and control for industry effects. It would also employ a comparison sample of non-family businesses.
NOTES 1. Indeed, where an owner-CEO actually runs the company the interests of the owners and the chief agents converge. 2. Discretion may even be distributed within top management teams consisting of multiple, long-tenured family members, who trust one another and work well together. 3. The implicit contrast here is with organizations that rely heavily on bureaucratic controls or financial incentives, and have less depth of talent.
REFERENCES Adler, P. S., & Kwon, S. W. (2002). Social capital: Prospects for a new concept. Academy of Management Review, 27(1), 17–40.
The Best of Both Worlds: Exploitation and Exploration
237
Allouche, J., & Amann, B. (1997). Le retour du capitalisme familial. L’expansion: Management Review, 85, 92–99. Amihud, Y., & Lev, B. (1999). Does corporate ownership structure affect corporate diversification. Strategic Management Journal, 20(11), 1063–1069. Anderson, R. C., & Reeb, D. (2003). Founding family ownership and firm performance: Evidence from the S&P 500. Journal of Finance, 58(3), 1301–1328. Anderson, R. C., & Reeb, D. (2004). Board composition: Balancing family influence in S&P 500 firms. Administrative Science Quarterly, 49(2), 209–237. Ang, J. S., Cole, R. A., & Lin, J. W. (2000). Agency costs and ownership structure. Journal of Finance, 55(1), 81–106. Bagsarian, T. (1997). Timken builds on bearings. Iron Age New Steel, 13(12), 50–54. Barnett, W. P., & Carroll, G. R. (1995). Modeling internal organizational change. Annual Review of Sociology, 21, 217–236. Barney, J. (1991). Firm resources and sustained competitive advantage. Journal of Management, 17(1), 99–120. Barney, J. B., & Hansen, M. H. (1994). Trustworthiness as a source of competitive advantage. Strategic Management Journal, 15(Winter Special Issue), 175–190. Baum, J. (1996). Organizational ecology. In: S. Clegg, C. Hardy, & W. Nord (Eds), Handbook of organization studies (pp. 77–114). Thousand Oaks, CA: Sage Baum, J., & Singh, J. (1994). Evolutionary dynamics of organizations. New York, NY: Oxford University Press. Beehr, T. A., Drexler, J. A., Jr., & Faulkner, S. L. (1997). Working in small family businesses: Empirical comparisons to non-family businesses. Journal of Organizational Behavior, 18(3), 297–312. Bornstein, P. (1983). All in the family. Forbes, 132(August 29), 152–153. Bourdieu, P. (1980). Le capital social: Notes provisoires. Actes de la recherche en sciences sociales, 31, 2–3. Braudel, F. (1984). Civilization and capitalism: 15th–18th century. Vol. III: The perspective of the world. Berkeley, CA: University of California Press. Bubolz, M. (2001). Family as a source, User and builder of social capital. Journal of SocioEconomics, 30(2), 129–131. Burns, T., & Stalker, G. (1961). The management of innovation. London, UK: Tavistock. Claessens, S., Djankov, S., Fan, J., & Lang, L. (2002). Disentangling the incentive and entrenchment effects of large shareholdings. Journal of Finance, 57(6), 2741–2771. Collins, J. C., & Porras, J. I. (1995). Built to last: Successful habits of visionary companies. New York, NY: HarperBusiness. Das, T. K., & Teng, B. S. (2001). Trust, control, and risk in strategic alliances: An integrated framework. Organization Studies, 22(2), 251–283. Davis, J., Schoorman, R., & Donaldson, L. (1997). Towards a stewardship theory of management. Academy of Management Review, 22(1), 20–47. De Geus, A. (1997). The living company. Boston, MA: Harvard Business School Press. Demsetz, H. (1988). Ownership, control, and the firm. The organization of economic activity series (Vol. 1). New York, NY: Blackwell. Demsetz, H., & Lehn, K. (1985). The structure of corporate ownership. Journal of Political Economy, 93(6), 1155–1177. Dieryckx, I., & Cool, K. (1989). Asset stock accumulation and sustainability of competitive advantage. Management Science, 35(12), 1504–1511.
238
DANNY MILLER AND ISABELLE LE BRETON-MILLER
Dominguez, A. (1998). Bossless culture working. Denver Post, 12(July), J5. Donaldson, L., & Davis, J. (1991). Stewardship theory or agency theory. Australian Journal of Management, 16, 49–64. Donckels, R., & Frohlich, E. (1991). Are family businesses really different. Family Business Review, 4(2), 149–160. Dreux, D. R. (1990). Financing family business. Family Business Review, 3(3), 225–235. Finkelstein, S., & Hambrick, D. (1996). Strategic leadership: Top executives and their effects on organizations. Minneapolis/St. Paul, MN: West Publishing Co. Fukuyama, F. (1995). Trust: The social virtues and the creation of prosperity. New York, NY: Free Press. Galbraith, J. (2000). Designing the global corporation. San Francisco, CA: Jossey Bass. Gallo, M. A., & Vilaseca, A. (1996). Finance in family business. Family Business Review, 9(4), 387–401. Gersick, K., Davis, J., Hampton, M., & Lansberg, I. (1997). Generation to generation: Life cycles of the family business. Boston, MA: Harvard Business School Press. Goffee, R., & Scase, R. (1985). Proprietor control in family firms. Journal of Management Studies, 22(1), 53–68. Gomez-Mejia, L. R., Nunez-Nickel, M., & Gutierrez, I. (2001). The role of family ties in agency contracts. Academy of Management Journal, 44(1), 81–95. Graham, M., & Shuldiner, A. (2001). Corning and the creation of innovation. New York, NY: Oxford. Guzzo, R., & Abbott, S. (1990). Family firms as Utopian organizations. Family Business Review, 3(1), 23–33. Hannan, M. T., & Freeman, J. (1984). Structural inertia and organizational change. American Sociological Review, 49(2), 149–164. Harrington, A. (2003). Who’s afraid of a new product? Fortune, 148(November 10), 189–192. Hoopes, D. G., & Miller, D. (in press). Owner preferences, competitive heterogeneity and strategic capabilities. Family Business Review, 19(3). Jacobs, M. T. (1991). Short-term America: The causes and cures of our business myopia. Boston: Harvard Business School Press. James, H. S. (1999). Owner as manager, extended horizons and the family firm. International Journal of the Economics of Business, 6(1), 41–55. Jensen, M., & Meckling, W. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4), 305–360. Kang, D. (1998). Family ownership and performance in public corporations: A study of the U.S. fortune 500, 1982–1994. Working Paper 00-0051, Boston, MA: Harvard Business School. Khanna, T., & Palepu, K. G. (2000). Is group affiliation profitable in emerging markets? An analysis of diversified Indian business groups. Journal of Finance, 55(2), 867–8891. Khurana, R. (2002). Searching for a corporate savior: The irrational quest for charismatic CEO’s. Princeton, NJ: Princeton University Press. Knott, A. M., Bryce, D., & Posen, H. (2003). On the strategic accumulation of intangible assets. Organization Science, 14(2), 192–207. Lansberg, I. (1999). Succeeding generations: Realizing the dream of families in business. Boston, MA: Harvard Business School Press. Lawrence, P., & Lorsch, J. (1967). Organization and environment. Cambridge, MA: Harvard University Press.
The Best of Both Worlds: Exploitation and Exploration
239
Lubatkin, M. H., Ling, Y., & Schulze, W. S. (2002). Fairness in family firms. Working Paper, University of Connecticut, Storrs, CT. Lund, R. T., Bishop, A. B., Newman, A. E., & Salzman, H. (1993). Designed to work: Production systems and people. Englewood Cliffs, NJ: Prentice-Hall. Mackie, R. (2001). Family ownership and business survival. Business History, 43(3), 1–32. Makadok, R. (2003). Doing the right thing and knowing the right thing to do. Strategic Management Journal, 24(10), 1043–1055. March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2(1), 71–87. McCartney, L. (1988). Friends in high places. New York, NY: Macmillan. McConaughy, D. L., Matthews, C., & Fialco, A. (2001). Founding family controlled firms: Performance, risk and value. Journal of Small Business Management, 39(1), 31–49. McConaughy, D. L., Walker, M. C., Henderson, G. V., & Mishra, C. S. (1998). Founding family controlled firms: Efficiency and value. Review of Financial Economics, 7(1), 1–19. Miller, D. (1990). The Icarus paradox: How exceptional companies bring about their own downfall. New York, NY: HarperBusiness. Miller, D. (2003). An asymmetry-based view of competitive advantage. Strategic Management Journal, 24(10), 961–976. Miller, D., & Friesen, P. (1984). Organizations: A quantum view. Englewood Cliffs, NJ: PrenticeHall. Miller, D., & Le Breton-Miller, I. (2005). Managing for the long run: Lessons in competitive advantage from great family businesses. Boston, MA: Harvard Business School Press. Miller, D., & Shamsie, J. (2001). Learning across the life cycle. Strategic Management Journal, 22(8), 725–745. Morck, R. K., Shleifer, A., & Vishny, R. W. (1990). Do managerial objectives drive bad acquisitions. Journal of Finance, 45(1), 31–48. Morck, R. K., & Yeung, B. (2003). Agency problems in large family business groups. Entrepreneurship Theory and Practice, 27(4), 367–382. Naphiet, J., & Ghoshal, S. (1998). Social capital, intellectual capital, and organizational advantage. Academy of Management Review, 23(2), 242–266. Palmer, D., & Barber, B. (2001). Challengers, elites and owning families: A social class theory of corporate acquisitions. Administrative Science Quarterly, 46(1), 87–120. Pfeffer, J. (1998). The human equation. Boston, MA: Harvard Business School Press. Porter, M. E. (1996). What is strategy? Harvard Business Review, 74(6), 61–78. Pruitt, B. (1999). Timken: From Missouri to Mars – A century of leadership in manufacturing. Boston, MA: Harvard Business School Press. Reid, R., & Harris, R. (2002). The determinants of training in SMEs in northern Ireland. Journal of Education and Training, 44(8/9), 443–450. Ruef, M. (2006). Boom and bust: The effect of entrepreneurial inertia on organizational populations. In: J. Baum, S. Dobrev & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 29–72). Oxford, UK: JAI/Elsevier. Schulze, W. S., Lubatkin, M. H., Dino, R. N., & Buchholtz, A. K. (2001). Agency relationships in family firms: Theory and evidence. Organization Science, 12(2), 99–116. Schulze, W. S., Lubatkin, M., & Dino, R. (2003). Dispersion of ownership and agency in family firms. Academy of Management Journal, 46(2), 179–194. Sheridan, J. H. (1992). America’s best plants: Timken. Industry Week, 241(20), 5354.
240
DANNY MILLER AND ISABELLE LE BRETON-MILLER
Singh, J. (2006). Ecology, strategy and organizational change. In: J. Baum, S. Dobrev & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management, (Vol. 23, pp. 177–214). Oxford, UK: JAI/Elsevier. Spector, R., & McCarthy, P. (1995). The nordstrom way: The inside story of America’s No 1 customer service company. New York, NY: Wiley. Teece, D., Pisano, G., & Shuen, A. (1997). Dynamic capabilities and strategic management. Strategic Management Journal, 18(7), 509–533. Tifft, S., & Jones, A. (1999). The trust: The private and powerful family behind the New York times. Boston, MA: Little Brown. Villalonga, B., & Amit, R. (in press). How do family ownership, management and control affect firm value. Journal of Financial Economics. Vinas, T. (2002). Manufacturing deserves investment. Industry Week, 251(1), 22. Ward, J. (2004). Perpetuating the family Business. Marietta, GA: Family Enterprise Publishers. Weber, J., et al. (2003). Family Inc. Business Week, 3857(November 10), 100–114. Wiersema, M. F., & Bantel, K. A. (1992). Top management team demography and corporate strategic change. Academy of Management Journal, 35(1), 91–121. Zahra, S. A. (2003). Ownership involvement and international expansion: An empirical test of the stewardship theory among family firms. Journal of Business Venturing, 18(4), 495–512.
PART IV: ORGANIZATIONAL LEARNING
241
242
IF IT DOESN’T KILL YOU: LEARNING FROM ECOLOGICAL COMPETITION Henrich R. Greve and Hayagreeva Rao ABSTRACT Learning theory explains how organizations change as a result of experience, and can be used to predict the competitive strength of individual organizations and competitive pressures in organizational populations. We review extant learning theoretical propositions on how competitive strength is affected by experienced competition, founding conditions, and observed failures of other organizations. In addition, we propose that niche changes are an important source of learning. We test these propositions on data from the Norwegian general insurance industry. We find that historical density increases failure rates, contrary to some earlier findings, and also that the effect of founding density supports the density delay rather than trial-by-fire hypothesis. We find that failures of others before and during the lifetime of the organization reduce failure rates, and niche changes reduce failure rates for joint-stock companies but not for mutual firms. Overall the findings suggest that organizations learn more cheaply from the failures of others than from their own experiences, and that the stresses of competition can overwhelm the learning effects of competition.
Ecology and Strategy Advances in Strategic Management, Volume 23, 243–271 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23008-0
243
244
HENRICH R. GREVE AND HAYAGREEVA RAO
INTRODUCTION In the early stages of ecological and learning research, selection and learning were depicted as rival processes and there was little attention bestowed on how they interacted to affect vital rates. Since then a number of studies explicitly recognize the interdependence between selection and learning; organizational ecologists have drawn on learning when seeking to deepen a theory of how organizations interact with each other; learning theorists draw on selection arguments when seeking to broaden a theory of the environment to encompass other organizations that learn (e.g. Levinthal, 1991a; Barnett, Greve, & Park, 1994; Swaminathan, 1996; Baum & Ingram, 1998). Our goal in this paper is to extend the interchange between organizational ecologists and learning researchers. To date, the extant literature emphasizes the connection between experience and competitive strength and has focused on how different types of experience influence competitive strength. However, reported research has devoted little attention to how organizations learn from organizational changes and thus derive competitive strength. The main exceptions have been studies investigating how the effects of changing niche position are contingent on the competitive conditions and balanced by costs of implementing changes (Amburgey, Kelly, & Barnett, 1993; Baum & Singh, 1996; Greve, 1999; Dobrev, Kim, & Hannan, 2001; Dobrev, Kim, & Carroll, 2003; Kim, Dobrev, & Solari, 2003). We wish to highlight the effect of organizational change on competitive strength for three interrelated reasons. First, if experience causes change in competitive strength, then it must be through some form of organizational change. Thus, organizational change mediates the effect of experience on competitive strength. Second, organizational change is a special form of experience – it is a unique and salient event that organizations learn intensely (March, Sproull, & Tamuz, 1991) and may culturally embed through stories and myths (Schein, 1990). Thus, change experience is different from inert experience. Finally, there are many types of organizational changes that may contribute to competitive strength, ranging from minor unrecorded adjustments in internal procedures to major events such as entry of new markets. Investigating whether major changes can explain organizational competitiveness helps researchers discover whether a gradualist or punctuationist perspective on organizational learning is more productive (Tushman & Romanelli, 1985). Major change events may be crucial for adaptation, or they may be red herrings that draw attention away from the more important business of everyday incremental changes.
If it Doesn’t Kill You: Learning from Ecological Competition
245
Below, we outline how organizations learn from competition and from the failures of other organizations, and then turn our attention to how they learn from their own changes. We also explore asymmetries in learning between the immediate impact of learning and the time-dependent impact of learning. We derive hypotheses and analyze data on Norwegian general insurers, and develop implications for subsequent research.
THEORY AND HYPOTHESES Learning from Competition Strategic management research has a long-standing interest in the effects of competition, and the claim that rivalry increases capabilities is a focal point of this research. The ‘‘Diamond model’’ (Porter, 1990) posits a set of interrelated mechanisms that jointly increase firm capabilities. First, a dynamic industry with high firm rivalry will push individual firms to develop better products and production technologies in order not to fall behind their competitors. Second, high customer demands trigger such rivalry, and are in turn escalated by the experience of receiving frequent improvements in product characteristics or costs. Third, the capabilities of firms in related and supporting industries are improved because the high pace of innovation in the focal innovation requires close cooperation and exchange of ideas with industries providing inputs or supporting services. Finally, even input factors such as employees benefit from the upgrade in firm capabilities triggered by rivalry. Strategy researchers have made progress in verifying parts of this model such as the predictions of local spread of knowledge (Jaffe, Trajtenberg, & Henderson, 1993) and of rivalry triggering counter-moves (Chen & Miller, 1994), but investigation of the central claim of rivalry creating stronger competitors has mainly taken place in organizational ecology. Organizational ecology contains theory positing that organizations which have experienced intense competition are stronger competitors (Barnett et al., 1994; Swaminathan, 1996). This can be reasoned both through a learning model and a selection model (Barnett et al., 1994). In a learning model, intense competition leads to lower performance, which triggers problemistic search and attempts to improve the organization (Cyert & March, 1963). When these attempts are successful, the organization becomes more resistant to competitive pressures. In a selection model, intense competition weeds out the weaker organizations, resulting in no change in the competitiveness of
246
HENRICH R. GREVE AND HAYAGREEVA RAO
each organization but in a population that is, on average, more resistant to competitive pressures (Barnett, Swanson, & Sorenson, 2003). On balance, exposure to fit organizations increases the competitive strength of the organization (Barnett et al., 1994, 2003; Barnett & Hansen, 1996). Thus, two lines of reasoning on organizational adaptation lead to the same prediction: H1. The average historical density faced by organization decreases its failure rate. A related argument emphasizes the competitive pressures at the time of organizational founding. According to ecological theory, young organizations change more easily than older ones, so the initial environmental conditions and managerial choices become imprinted into the organizational routines (Stinchcombe, 1965). An implication is the density-delay prediction that organizations facing insufficient resources at birth become hobbled by their adaptations to this condition and never reach full competitive strength (Carroll & Hannan, 1989). The density-delay prediction has seen supportive evidence (Carroll & Hannan, 1989; Dobrev, Kim, & Carroll, 2002), but has also faced an opposing hypothesis that intense competition in the early period is beneficial in the long run. This opposing argument is known as the trial-by-fire hypothesis, and posits that organizations founded under periods of intense competition learn intensely from their own experience and the experience of their cohort so that they not only reach the level of resistance to competition enjoyed by organizations founded under more benign conditions, but actually overshoot and become more resistant (Carroll & Hannan, 1988; Swaminathan, 1996). The trial-by-fire hypothesis resembles hypothesis 1 in its behavioral theory of how organizations learn to resist competition, but differs in the assumption on when such learning occurs. Both hypotheses can be argued from either a problemistic search or a selection perspective, but it is more realistic to assume that search and selection interact. Low performance resulting from intense competition not only produces problemistic search, but also risk taking (Bromiley, 1991; Fiegenbaum, 1990). When managers seek to solve problems and take risks, the result is a mixture of useful and harmful changes that increase the variance of organizational performance (Greve, 1999). Because competitive pressures trigger these changes, they occur in a highly selective environment and thus lead to elimination of the organizations making failed experiments, which means that the population is left richer in organizations that made successful experiments. Thus, the combination of learning and selection ratchets up the resistance to competitive pressures in the organizational population. This learning-selection effect
If it Doesn’t Kill You: Learning from Ecological Competition
247
operates more intensely in young organizations because they change more readily, and thus expose themselves to risk, and also because they have smaller economic buffers against failure (Levinthal, 1991b; Miner, Amburgey, & Stearns, 1990). These arguments lead to: H2. The density at organizational founding decreases the failure rate. However, the trial-by-fire argument directs attention to a subtle effect of competitive pressure (Swaminathan, 1996). According to this argument, it is important to distinguish the initial failure risk of recently founded organizations from the rate of improving the organizational resistance to failure. We may refer to the former as the initial failure rate, and the latter as the learning rate. The learning rate can be given much the same interpretation as learning rates in learning-curve research (Argote, 1999). For organizational learning from competition, the trial-by-fire argument is that high competition at founding gives a high risk of failure, but also constitutes a stimulating learning environment that speeds the learning rate. This gives an alternate form of Hypothesis 2: H2 (alt). The density at organizational founding increases the initial failure rate and the learning rate. Learning from Failures Organizations learn from the experience of others through a broad range of transmission mechanisms (Levitt & March, 1988). Organizational members observe events in the environment and draw inferences from these (Greve, 1996; Greve & Taylor, 2000). Organizational members learn about new technologies and practices and implement them (Cohen & Levinthal, 1990; Tsai, 2001). Organizations hire members with experience from other members of the population and use their knowledge (Drazin & Rao, 2002; Rao & Drazin, 2002). Some of these learning mechanisms affect a new organization in an experienced population, and thus are a form of congenital learning (Huber, 1991; Ingram & Baum, 1997). Others affect an organization during its operation, and thus constitute vicarious learning. Organizations can potentially learn from the experience of others based on many different events, but special attention has been given to learning from organizational failures (Ingram & Baum, 1997; Chuang & Baum, 2003). Organizational failures are highly salient because they represent a manager’s worst fears about what might happen to one’s own organization, but are also good news in the form of relief from competition. As a result,
248
HENRICH R. GREVE AND HAYAGREEVA RAO
organizational failures draw the attention of managers and become events that need to be interpreted and acted on in order to become less threatening (Daft & Weick, 1984), making them occasions for intense learning and adjustment behaviors by managers seeking to avoid the same fate (Miner, Kim, Holzinger, & Haunschild, 1999). Organizational failures are not only salient events that cause intense learning; they are also events that tend to reduce some of the risks of harmful learning. Learning by making one’s own experiments, as in organizational change, is a behavior with short-term risks because many of the behaviors that are explored will turn out to be maladaptive (Greve, 1999). Even behaviors that are adaptive in the long run can be costly in the short run and thus cause failure (March, 1981; Hannan & Freeman, 1984; Miner et al., 1990). Observing the failures of others is a much less costly form of learning because the focal organization is free riding on the experiments of others. A sufficiently high number of observed failure events improve managerial estimates of which actions are harmful because they reliably lead to failures. For managers who keep track of competitors that do not fail as well, observing many failures along with non-failing organizations also improves the estimates of which actions are risky because they lead to failures and successes, and which are safe because they rarely lead to failure. The extent to which managers can avoid taking risky or reliably harmful actions is thus roughly proportional to the number of failures they have observed. Learning from failure may create organizations that are less likely to fail but forego some profit opportunities. When managers avoid strategies that are associated with the failure of other organizations or strategies that other organizations have withdrawn from, they forfeit opportunities to tinker with those strategies. The avoided strategy might in fact be effective in a different environment, such as at a later time or in a different market segment, or a variation of the avoided strategy might be effective (Greve, 1995). Observed failures in an uncertain environment thus cause biased estimates of the value of a strategy (Denrell & March, 2001) and restrict the range of strategic exploration (March, 1991). These effects are reduced if managers are able to build accurate mental models of the industry that help them interpret the causes of failures, but simple learning from failure through avoidance can occur and lead to reduced exploration and risk taking in the absence of such interpretation. Because managers may have had experience in the industry before founding the organization, or may have access to historical knowledge, even failures that occur before founding affect their behavior (Ingram & Baum, 1997). Indeed, it is an interesting theoretical issue whether failures observed
If it Doesn’t Kill You: Learning from Ecological Competition
249
before the founding or after are most consequential. The lessons gained from failures before the organizational founding can be embedded in its founding structures and procedures, and thus becomes parts of its (inert) organizational core. The lessons gained from failures after the organizational founding are presumably more salient to management, and may cause stronger (but also more costly) adjustments. If the organization is committed to its current strategy, on the other hand, observation of failures by others does not cause changes (Chuang & Baum, 2003). Here, we simply propose a separate effect of learning before and after the organizational founding: H3. The accumulated failures in the population after the organizational founding reduce the organizational failure rate. H4. The accumulated failures in the population until the organizational founding reduce the organizational failure rate. Adapting through Change The theory and hypotheses so far have made causal links from specific forms of organization and population experience to organizational failure rates. Clearly, these causal links all rely on a process of experience causing organizational changes, which again affect the likelihood of failure. One way of testing the theory would thus be through measurement of the intermediate change processes, provided change events could be measured accurately. Such investigation is not necessary for testing the theory, however, because the main interest is in whether specific forms of experience intensify learning processes relative to other forms of experience. Difficulties in measuring all relevant forms of changes and weighting their effect on organizational failure risks may in fact caution against such tests. Despite this concern, investigation of the adaptive effects of organizational changes is important. First, testing the effect of specific strategic changes on failure rates helps partition the effects of experience into those mediated by large-scale changes and those mediated by smaller, hard-toobserve changes. Large-scale changes have special importance for the theory of organizational ecology when they alter the organizational core of goals, forms of authority, technology, and market strategy (Hannan & Freeman, 1984). The organizational core is subject to stronger requirements of reliability and accountability than the periphery, and thus disruptions of it imply particularly high risk of failure (Hannan & Freeman, 1984). On the other hand, the organizational core contains the characteristics that have the
250
HENRICH R. GREVE AND HAYAGREEVA RAO
greatest effect on the adaptation to the environment, so changing the core is the strongest lever for bringing the organization from a starting state of intense competition or poor adaptation to an improved end state (Barnett & Carroll, 1995). Because core changes imply complex reorganization of the internal organization (Barnett & Freeman, 2001) and uncertain assessment of alternative actions (Harrison & March, 1984), they imply a significant risk of reduced performance (Greve, 1999). A change to the core is often followed by a reassessment and further adjustments, which further weakens the organization in the short-term, but holds the promise of a better environmental adaptation. The net effect of the making a single core change is thus unclear, as it depends on the balance of the short-term increase in failure risk and the longer-term hope for better adaptation. When examining the accumulated changes by an organization over its life span, however, the net effect is easier to argue. Making multiple changes means, first of all, that the organization has had sufficient resources to avoid failure after its first change. Second, it means that the organization has made additional adjustments, and thus has been able to learn from its mistakes and find a better adaptation to its environment. Thus, the prediction is for a reduction in failure rate by organizations that have done multiple changes over their life span: H5. The accumulated changes by the organization reduce the organizational failure rate. The effect of organizational change on the failure rate likely depends on the organization form as well. First, organizations have different governing institutions, and these differences may affect several components of the process costs of change. The potential for internal strife is greater in organizations with relatively more decentralized power bases, the speed of changing is lower in organizations with protracted decision-making routines, and the ability to make adjustments after the initial change event is lower in more formalized organizations. Such distinctions are often difficult to capture in the longitudinal data sets that are most suitable for testing the consequences of change, however, so investigators are left to use cruder measures instead. Our data of general insurance firms have a mixture of joint-stock and mutual organizations, which are governed differently with a much greater potential for customer voice in the decision-making in mutuals, where the customers are owner/members. It seems likely that this will make mutuals more inert, but we do not offer a formal hypothesis because the distinction between these two ownership forms depends on institutional
If it Doesn’t Kill You: Learning from Ecological Competition
251
features, and its consequences for adaptation may therefore vary across contexts. A more general distinction is between generalists that serve a large portion of a market and specialists that serve only a small niche. The economic basis of a generalist strategy is economies of scale because it allows for larger organizations, sometimes mixed with economies of scope when coverage of one part of the market makes coverage of nearby parts cheaper. These static efficiency differences of generalists are counterweighted by greater difficulty of adaptation in generalist organizations (Ingram & Baum, 1997; Dobrev et al., 2003). Because generalists have multiple areas of activity, their managers face greater ambiguity in interpreting performance feedback and deciding on responses, which increases the difficulty of finding correct responses (Greve, 2003). Their multiple areas of activity also create constraints because changes to one area may alter the competitiveness in another in ways that are difficult to anticipate, and thus their adaptive problem becomes highly complex (Rivkin, 2001). It follows that the benefits of change should be lower for generalists (Ingram & Baum, 1997; Dobrev et al., 2003): H6. The accumulated changes by the organization reduce the organizational failure rate less for organizations covering a greater niche span.
DATA AND METHODOLOGY General Insurance We use data from the Norwegian general insurance industry to investigate these hypotheses.1 This industry has a lengthy and comparably well-documented history and good records on an important form of core change: entry into new market niches and exits from old ones. Entry into the industry or additional niches did not face regulatory barriers except an approval of a license, and firm failures were sufficiently frequent to allow analysis. The niche structure of general insurance is interesting because it differs from the industries used in earlier work on the consequences of niche entry and exit. Unlike radio stations (Greve, 1999), but like automobile manufacturers (Dobrev et al., 2001, 2003; Kim et al., 2003), daycares (Baum & Singh, 1996), and newspapers (Amburgey et al., 1993), insurance firms can have varying niche width so that entry into one niche does not automatically mean exit from another. Unlike the radio, newspaper, daycare, and
252
HENRICH R. GREVE AND HAYAGREEVA RAO
automobile industries, the niches in the insurance industry do not have a clear spatial structure with variables spanning user characteristics (demographics of radio listeners and political affiliation of newspaper readers) or product characteristics (engine sizes of automobiles). General insurance niches are defined by the intersection of the object covered (e.g., a building) and the risk covered (e.g., fire), and specific insurance niches usually insure a single object types against one or more risks or, more rarely, insure multiple object types against a single risk. It is possible to make discontinuous subdivisions of insurance niches based on the type of customer, object, and risk, but these subdivisions do not form a space with meaningful distances. Similarly, the expertise needed for effective operations in insurance is based on a general body of knowledge on risk evaluation, contract writing, and protection, but each niche also requires substantial specific expertise. For example, correct pricing of machine insurance (which is a niche in our data) requires personnel with the ability to judge the machine interruption or damage risk for different types of machinery installed in factories. Firms use niche-specific expertise for assessment and pricing of risk when contracts are written and evaluation of cause and damage when claims are filed. Niche entry thus constitutes a change in the marketing strategy of the firm and entails changing the technical core, so it modifies two parts of the organizational core (Hannan & Freeman, 1984). Formal organizations offering insurance in Norway first appeared in the mid-18th century. Prior to this period, merchant shippers purchased maritime insurance from insurance firms abroad and individuals in Norway, so the product was known. The early evolution of the insurance industry was a period of difficult economic and social conditions. Norway separated from Denmark in 1814 and was forced to join a union with Sweden, which it left in 1905. The early time of the union was a period of economic rebuilding after years of disastrous naval blockade and invasion threats during the Napoleonic wars, and saw capital scarcity in business and poverty among farmers, fishermen, and the urban underclass. The Norwegian state wanted to develop the insurance industry (as well as a number of other industries), but lacked resources for active promotion and instead adopted a posture of benign nonintervention (Færden, 1967). The general insurance industry grew through foundings of new firms rather than through entries by firms active in other industries. In particular, banking and general insurance firms did not attempt entry into the other industry during the study period, though there were some cross-entry by bank and life insurance firms in the later part of it. Most general insurance firms operate in niches where the customers are firms or individual entrepreneurs and the risks
If it Doesn’t Kill You: Learning from Ecological Competition
253
are associated with damage to production equipment. For example, in Norway, maritime insurance is important and split into niches for protection and indemnity (P&I; damage done by the ship to others), ship loss/damage, cargo loss/damage, and fishing (loss/damage of vessel, gear, and crew). Though generalists are also active in these niches, most maritime insurers specialize in just one niche. One example is Gard, which is a P&I mutual formed in Arendal in 1907 by local ship owners. It has since expanded internationally and is now the world’s largest P&I mutual with 12.5 percent of the market in deep-sea shipping. Land-based industry also buys a number of insurance products either from specialists or as one of multiple niches served by generalist firms. Insurance for individuals has a greater proportion of generalist firms offering multiple forms of insurance. Many of these were originally founded for fire insurance of buildings or goods, and later expanded to offer insurance against other individual risks such as automobile, personal injury, burglary, and water damage to buildings. An example is Det Norske Brandassuranceselskap (The Norwegian Fire Insurance Company), which was founded in Bergen in 1923 and was active in 11 niches (including maritime insurance) when it failed in 1965. Data Sources Life histories of all general insurance firms in Norway were constructed using the annual reports of the Norwegian Insurer’s association from 1912 (volume 1) through 1999, annual reports from the Insurance Council (the regulatory authority of the insurance industry in that period) from 1912 (volume 1) through 1986, three monographs on the history of the insurance industry (Færden, 1967; Lorange, 1935; Wigum, 1993), and firm histories for some of the firms. All insurance companies recorded in these sources are in the data, and the founding dates are known to the day for most, and to the nearest year for the rest except a few of the earliest firms. Data on general economic conditions were drawn from the Norwegian Central Bureau of Statistics except for the GDP numbers, which are from two standard textbooks on the economic history of Norway (Hodne & Grytten, 2000, 2002). The data analyzed here include all general insurance firms (including mutuals) except village fire-insurance mutuals, which were weakly integrated with the rest of the industry. Instead, village fire mutuals inherited some features of pre-modern forms of mutual help arrangements and were spread by a social movement of (largely rural) self-reliance that also produced the savings banks. The village fire mutuals operated under a separate law and without supervision from the Insurance Council.
254
HENRICH R. GREVE AND HAYAGREEVA RAO
Hypothesis Testing and Control Variables The following variables test the hypotheses. As usual, all density measures are counts of organizations in the focal market (Norway) with the same organizational form (general insurance company). Average density is the weighted average of the density faced by the organization over its history, with each year weighted by a discount factor of 0.9tt0, where tt0 is the difference between the focal year and the year whose density is weighted. This specification differs from earlier work using an un-weighted average of the density in the last 10 years (Barnett et al., 1994; Barnett & Hansen, 1996), but was chosen because it matches the weighting done in the change variable. The weighted measure correlates highly with the 10-year measure (0.96), and the choice does not affect the findings. This variable varies over time and across organizations from different founding cohorts. It tests Hypothesis 1. Founding density is the density at the time of founding (Carroll & Hannan, 1989). It differs across organizational cohorts and is constant over time in a given organization, and is entered to test Hypothesis 2 versus 2(alt). Failures at founding is the number of failures in the organizational population before the founding of the focal organization (Ingram & Baum, 1997). It varies across organizations and is constant over time in each organization, and it tests Hypothesis 3. Failures since founding is the number of failures in the organizational population after the founding of the focal organization (Ingram & Baum, 1997). It varies across organizations and over time within each organization, and tests Hypothesis 4. These variables include mergers and acquisitions because a firm dissolving through merger or acquisition can provide information useful for organizational learning, just as a firm dissolving through other processes. Organizational change is the accumulated number of niche changes done by the focal organization. This variable was made by coding the descriptions of activities that each insurer gave in the yearbook of the Insurer Association, and comparing them year-by-year to find entries and exits. The niche delineation used to generate this variable recognizes 34 different niches, and the highest number of niches simultaneously occupied by a firm in the data is 16. The variable is calculated as a weighted average with each year weighted by a discount factor of 0.9tt0, where tt0 is the difference between the focal year and the year whose density is weighted in order to let more recent change experience be more influential. The factor 0.9 was chosen as the best fitting after a searching over weights over a 0.1 interval grid. It tests Hypothesis 5, and is used in interactions with organizational form (no prediction) and the count of niches (Hypothesis 6).
If it Doesn’t Kill You: Learning from Ecological Competition
255
The models also contain control variables that capture focal firm characteristics, economic conditions and density dependent competition. One variable is an indicator of whether the focal firm is a mutual insurer (owned by the customers). The alternative is a joint-stock firm, as other organizational forms are not seen in this population. Although there was some variation in mutual insurer size and market coverage, the modal mutual insurer had a geographically restricted area of operation and operated through a mixture of economic function and local social mobilization. In this they resemble other mutual firms that have been studied (Barnett & Carroll, 1987; Staber, 1989), and are likely to display the high resistance to failure and high degree of inertia typical of mutual firms. This variable is constant over time within each firm, as transitions between the mutual and joint-stock forms were not observed in this time period.2 Specialist is an indicator variable for a firm that serves only one market niche (based on the niche classification used to categorize niche change above). Niche count is the number of niches that the firm served. The indicator variable and the count of niches are both entered in case the relation from count of niches to failure risk is not linear. Both of these variables vary across firms and over time within each firm. Few economic time series are available for as long back as these data cover, but those that are available reflect important economic sectors in 19th century Norway. Some of them are also important currently, and consideration of data availability and ongoing economic importance led to the retention of the following economic variables: Total ships is the number of merchant ships registered in Norway, and was included because of the key role of shipping in the Norwegian economy in general and the insurance industry specifically. In recent years, this variable understates the ships controlled by Norwegian shippers, but no adjustment for this fact was done because the effect of ships registered in low-cost ship registries (such as Panama and Liberia) on the Norwegian economy is small. The merchant fleet does not include fishing vessels. GDP growth is the growth in inflationadjusted GDP. Finally, population density is entered as a linear and a square term to capture density dependent competition. This variable is needed in order to distinguish the effect of contemporaneous competition from the effects of historical competition hypothesized in H1 and H2. It is also needed to distinguish the learning effect of failures hypothesized in H3 and H4 from the competition-reducing effect of failures. With current density entered, all effects of experience are net of the current density-dependent competition. All variables are entered with a one-year lag to avoid simultaneity effects.
256
HENRICH R. GREVE AND HAYAGREEVA RAO
GDP growth, however, is contemporaneous (it is the growth from last year to the focal year) because the effect of GDP growth on business can be very fast, and a simultaneity bias is not plausible. Fig. 1 plots the density of all casualty insurers and the sub-densities of insurers with the mutual and stock form. The most obvious feature of the graph is the peak in density around World War I, which was a boom period for maritime insurance, but it is also interesting to note how well the mutual form is doing. It had higher density than the joint-stock form throughout the 19th century, and it took an era of high founding rates of stock companies for them to catch up. This era was in large part a result of new economic opportunities in the early 20th century such as the rapid expansion of merchant shipping and industry. Tables 1 and 2 display the descriptive statistics and correlations of the variables in the analysis. The high frequency of mutuals and specialists is notable, as is the positive but moderate correlation of these variables. The high density of insurers may seem surprising, but should not be. Most general insurance firms do business with firms rather than individuals, so the number of insurers in the economy far exceeds the number that the reader will be familiar with. The high proportion of specialists also contributes to the density. For example, the data contain several mutuals that only insure fishing vessels, and each of them mainly or only draws customers from a limited geographical area. While the market niche and regional specialization of these insurers cause them to avoid direct competition with each 200 180 160
Total Stock Mutual
Density
140 120 100 80 60 40 20 0 1800
1820
1840
1860
1880
1900
1920
1940
Year
Fig. 1.
Density of Insurers.
1960
1980
2000
257
If it Doesn’t Kill You: Learning from Ecological Competition
Descriptive Statistics.
Table 1. Variable Mutual Specialist Nichecount Total ships GDP growth Density Density 2/100 Founding density Average density Failures at founding Failures experienced Changes Changes stock Changes niche count
Mean
Std. Dev.
Min
Max
0.571 0.573 2.041 5426 0.031 116.334 147.992 94.518 115.301 73.297 80.336 0.208 0.188 1.114
0.494 0.494 2.354 1513 0.042 35.576 82.964 48.436 34.893 68.613 72.558 0.505 0.494 3.929
0 0 1 2272 0.096 2 0.04 2 2 5 0 0 0 0
1 1 16 8552 0.171 196 384.16 196 196 332 329 4.250 4.250 56.246
Table 2. 1 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.
Mutual Specialist Niche count Total ships GDP growth Density Density 2/100 Founding density Average density Failures at founding Failures experienced Changes Changes Stock Changes Niche count
1.00 0.30 0.38 0.11 0.02 0.13 0.13 0.20 0.17 0.14 0.05 0.39 0.44 0.30
2
Correlation Coefficients. 3
4
5
6
7
8
9
10
11
13
14
1.00 0.51 1.00 0.09 0.09 1.00 0.02 0.05 0.03 1.00 0.13 0.18 0.46 0.07 1.00 0.11 0.15 0.50 0.07 0.97 1.00 0.16 0.16 0.32 0.10 0.45 0.38 1.00 0.17 0.23 0.48 0.12 0.91 0.84 0.62 1.00 0.10 0.05 0.23 0.10 0.20 0.14 0.70 0.35 1.00 0.10 0.27 0.22 0.12 0.21 0.13 0.10 0.39 0.12 1.00 0.29 0.58 0.17 0.03 0.26 0.25 0.18 0.29 0.03 0.22 1.00 0.28 0.58 0.15 0.03 0.25 0.25 0.18 0.28 0.03 0.19 0.96 1.00 0.30 0.76 0.09 0.03 0.17 0.16 0.11 0.20 0.00 0.19 0.82 0.82
other, they constitute diffuse competition against each other and against generalists, which might otherwise have been able to insure their customers. The correlation table shows that firms with more niches will tend to have made more changes to the niches. This is a result of the firm histories, which mainly show insurers starting as specialists or with a few niches, and then becoming generalists through niche entry. Although the market change variable is the sum of entries and exits, it is dominated by entries. A set of correlations of some concern is that of average density with current density
258
HENRICH R. GREVE AND HAYAGREEVA RAO
and founding density. Despite these positive correlations, the coefficients of these variables did not show instability in the modeling. The findings thus seem secure, but it is worthwhile to keep in mind that computation of the partial effects of each of these variables gives a somewhat hypothetical depiction of their effects, as they covary in the data. Dependent Variable and Model The dependent variable is an event of failure, defined as firm dissolution except through mergers and acquisitions. This leaves voluntary dissolutions and bankruptcies, which are usually results of poor economic performance. There are 285 such failures in the data, and 49 mergers and acquisitions. Insurers enter the data the year they are founded, and remain at risk of failure until they fail, leave the data through a merger or acquisition, or stay until the end of the study period (1998). The analysis is thus an event history model with failure as the focal event, and merger, acquisition, or the end of the study as censoring events. There are 399 insurers in the data, and 370 of these enter the analysis. The rest are omitted because of missing data on the firm or the economic time series, which start in 1835. The first insurer covered by the analysis was established in 1828, and there were two insurers active in 1835. Five had entered and failed before 1835, and a sixth was operating but is omitted from the analysis because it was government operated (this was a fire insurance mutual that did business nationwide, but was strongest in cities and towns because of competition from the village mutuals in the rural districts). Preliminary analysis was used to assess whether the age dependence of the failure rate could be modeled by any of the parametric event history models. The findings suggested that the hazard rate most resembled a monotonously declining function such as the Gompertz model, which implies the usual liability of newness (Freeman, Carroll, & Hannan, 1983), but this model showed some indications of insufficient fit. In response, the Cox model was picked instead. This model has a flexible specification of the time dependence of the hazard rate, and has a proportional hazards specification of the effect of the covariates (like the log-linear vector in the Gompertz model). Because we also hypothesize an age-dependent effect, we add a time-dependent vector to the usual Cox model, resulting in the following specification: rit ¼ rt expðbX it Þ expðgY it lnðtÞÞ Here, the first two terms are the usual Cox specification with a flexible time-dependent specification of the rate (rt) and a log-linear link function for the covariate vector X (of the firm i and time (age) t) and its associated
If it Doesn’t Kill You: Learning from Ecological Competition
259
coefficients b. The last term contains a covariate vector Y with coefficients g, which is multiplied with the natural logarithm of time (age) to produce a non-proportional effect. In the models below, founding density is entered in Y. This extension of the Cox model is implemented in Stata 8.0, which was used to estimate the model. Three time-dependent specifications were tried (linear, natural logarithm, and square root), and the natural logarithm was chosen for its superior fit. The age of firms is on a scale of months since founding.
FINDINGS The results of the analysis are reported in Table 3. The model is built up gradually from model 1, with only control variables, to the saturated model 6. Although few changes to the coefficient estimates are found as the model is built up, the ones that occur are instructive, and so the models will be discussed one at a time. Model 1 has the control variables only. It shows that mutuals are much less likely to fail, and that specialists are more likely to fail. The greater failure rate of specialists is probably not explained by environmental variability (Hannan & Freeman, 1989), as the insurance industry operated in a stable environment except in some periods such as the wars. On the other hand, the insurance industry may have had scale advantages that benefited generalists, and it did not achieve levels of concentration that would have triggered niche partitioning effects (Carroll, 1985). Thus, small scale and competition from generalists is a likely cause of the high failure rates of specialists. Consistent with the finding on specialists, greater generalism (more niches) gives lower failure rates. Organizational size is omitted from the specification because of missing data from 1835 to 1917, so the usual caution to the findings apply: generalism correlates with organizational size, and thus the liability of specialism found in this analysis includes a spurious effect of the omitted variable organizational size (Barron, West, & Hannan, 1995). The omission may also affect the estimates of other variables, though we do not have high correlations of other variables and size in the period for which size data are available. Total ships have a negative effect on the failure rate, reflecting a direct beneficial effect of shipping on the maritime insurers and possibly also a multiplier effect of the economically important shipping industry on the economy as a whole. GDP growth has a negative and insignificant effect on failure rates. The density variables have the expected signs, but the coefficient magnitudes are small and only the squared term is significant.
260
Table 3.
Proportional Mutual Specialist Niche count Total ships GDP growth Density Density 2/100 Founding density Average density
1 0.632 (0.135) 0.472 (0.149) 0.163 (0.045) 0.00022 (0.00006) 0.906 (1.072) 0.007 (0.006) 0.006 (0.003)
2 0.487 (0.147) 0.542 (0.152) 0.141 (0.047) 0.00028 (0.00007) 1.012 (1.107) 0.049 (0.012) 0.013 (0.004) 0.010 (0.003) 0.029 (0.006)
3 0.484 (0.147) 0.551 (0.153) 0.146 (0.047) 0.00032 (0.00007) 1.004 (1.115) 0.049 (0.012) 0.012 (0.004) 0.031 (0.018) 0.039 (0.008)
4 0.620 (0.153) 0.558 (0.152) 0.084 (0.049) 0.00028 (0.00007) 1.050 (1.093) 0.050 (0.012) 0.014 (0.004) 0.010 (0.003) 0.030 (0.006)
5 0.629 (0.152) 0.572 (0.154) 0.089 (0.049) 0.00031 (0.00007) 1.020 (1.099) 0.050 (0.011) 0.012 (0.004) 0.038 (0.019) 0.041 (0.008)
6 0.687 (0.151) 0.586 (0.152) 0.083 (0.049) 0.00031 (0.00007) 1.037 (1.098) 0.051 (0.012) 0.012 (0.004) 0.039 (0.019) 0.042 (0.008)
7 0.604 (0.152) 0.611 (0.157) 0.023 (0.062) 0.00031 (0.00007) 1.012 (1.102) 0.050 (0.011) 0.012 (0.004) 0.036 (0.019) 0.041 (0.008)
8 0.664 (0.153) 0.616 (0.155) 0.030 (0.062) 0.00031 (0.00007) 1.029 (1.101) 0.050 (0.011) 0.012 (0.004) 0.038 (0.019) 0.042 (0.008)
HENRICH R. GREVE AND HAYAGREEVA RAO
Model
Cox Models of Insurance Firm Failures.
0.008 (0.002) 0.016 (0.003) 0.593 (0.167)
0.010 (0.003) 0.022 (0.004) 0.632 (0.167)
0.010 (0.003) 0.022 (0.004) 0.311 (0.325) 1.063 (0.373)
0.010 (0.003) 0.022 (0.004) 0.347 (0.262)
0.115 (0.093)
Age dependent Founding density Wald w2 Pseudo-log likelihood
0.009 (0.003) 0.022 (0.004)
0.005 (0.002) 126.37 1313.778
142.32 1291.868
142.05 1289.138
0.005 (0.002) 160.75 1285.463
162.21 1281.816
The regressions have age dependence set to the logarithm of age. Observations: 14,729. Significant at 5%; Significant at 1%; two-tailed significance tests; robust standard errors within parentheses.
0.005 (0.002) 178.11 1279.611
0.005 (0.002) 153.72 1280.721
0.010 (0.003) 0.022 (0.004) 0.452 (0.332) 0.960 (0.414) 0.095 (0.095) 0.005 (0.002) 170.55 1278.905
If it Doesn’t Kill You: Learning from Ecological Competition
0.007 (0.002) 0.017 (0.003)
Failures at founding Failures since founding Changes since founding Changes s. f. Stock Changes s. f. Niche count
261
262
HENRICH R. GREVE AND HAYAGREEVA RAO
Model 2 adds the variables that test Hypotheses 1–4. In this analysis, the control variables for density dependence become larger and significant (and stay significant for the rest of the analysis), suggesting that the omission of the learning effects of these variables may have masked the density dependence effect. The coefficient estimate of average density is also positive and significant, which contradicts the learning-from-competition argument of Hypothesis 1. The coefficient estimate of founding density is positive and significant, which supports a density delay effect (Carroll & Hannan, 1989) instead of the learning model implied by the trial-by-fire effect (Swaminathan, 1996). Hypothesis 2 is thus contradicted for now, but later models will examine the more complete version of the learning argument given in H2(alt). Failures at founding and failures since founding are both negative and significant, which supports the learning from failure reasoning behind Hypotheses 3 and 4 (Ingram & Baum, 1997). Model 3 adds the age-dependent specification of founding density to the model. The proportional effect of founding density is kept in order to test both effects simultaneously, as the hypothesis requires. Now the main effect is insignificant, but the age-dependent effect is positive and significant. This is opposite to the expectation of a negative effect given by hypothesis 2a. Instead of the hypothesized learning effect, the findings suggest that the original argument of tight niche packing drives the density delay effect (Carroll & Hannan, 1989). Learning should reduce the failure rate over time and resource scarcity should have most pronounced effects soon after founding, which both suggest a negative effect in the time-dependent vector. Tight niche packing, on the other hand, is a liability that is not reduced over time, and becomes more pronounced if the economies of scale increase or competitors increasingly learn to exploit economies of scale. Hence, positive age dependence favors tight niche packing over the two other effects of founding density. Model 4 returns to a specification without founding density in the timedependent vector and adds changes since founding. The coefficient is negative and significant, so hypothesis 5 of a reduced failure rate as a result of making changes is supported. Hence, changing the organizational niche position is one mechanism by which organizational experience is turned into increased competitiveness. This finding is maintained in model 5, which adds founding density back into the time dependent vector, and has findings that are mainly consistent with those in model 3. The exception is that model 5 actually has a negative effect and significant effect of founding density. It seems from this pattern of findings that organizations benefit from the accelerated learning that occurs as a result of high founding density, but that the mal-adaptation from tight niche packing is more important in the long run.
If it Doesn’t Kill You: Learning from Ecological Competition
263
Model 6 tests Hypothesis 5 again, but takes into account one institutional detail of these data. Most of the organizations are mutuals, which are founded on a combination of economic and community service goals, and thus have a market niche that in part is determined by noneconomic considerations. Arguably such an organizational form is more inert than the standard joint-stock company, and thus it may react differently to changes. It seems natural to propose that changes in market niche will cause mutuals to become more likely to fail because such changes tear at the seams of the social mobilization that gives them high survival chances. Joint-stock companies, on the other hand, can learn from market niche changes in the neutral way that the hypothesis proposes. Following this reasoning, model 6 enters an interaction of joint-stock form and accumulated changes to the model. As expected, the benefit of changing is limited to joint-stock companies, since the interaction is negative and significantly larger than the main effect (po0.01). Mutuals do not show a liability of changing, as the positive main effect is not significant, but clearly do not benefit either. This set of coefficients is consistent with mutuals being more inert than joint-stock firm as a result of their greater commitment to the focal niche. The same findings can, however, occur if the mutuals instead are less adept at finding new niches with less competition (Baum & Singh, 1996). More specifically, mutuals may have exited niches where competition was low and entered niches where competition was high due to the presence of stocks; although stocks tended to fail more, they also may have generated more competition. Most of the niche changes in the data are expansions, so in order to improve the life chances an organization needs to find additional niches that are sufficiently better than the existing ones to justify the liability of changing. Model 7 removes the interaction of change and niche count and enters the interaction of change and niche count that tests Hypothesis 6. The coefficient estimate of the interaction effect is insignificant, a finding that is repeated in model 8 when the interaction of change and joint-stock firm is entered back. Thus, the data do not give support to a greater liability of change for multi-niche firms. A possible reason is that these terms are quite collinear because most firms started with one or a few niches, and thus the interaction effect is difficult to distinguish.
DISCUSSION In order to advance the integration of ecological research and learning research, we sought to investigate how learning and selection interact in a
264
HENRICH R. GREVE AND HAYAGREEVA RAO
population of general insurance firms. Our evidence suggests that learning affects selection processes, and underscores the complex relationship between both processes. The best evidence of beneficial learning came from investigating effects of failures in the population on the failure of the focal firm. Clearly, learning from the misfortunes of others is largely beneficial; as it helps decision makers develop knowledge of ‘‘fatal actions’’ that should be avoided (Miner et al., 1999). It is also an inexpensive form of vicarious learning that only requires observation of competitors, unlike learning from the organization’s own changes, which entails costs and risks of change. Although there is a negative effect of failures in the population on the focal firm’s failure, this does not necessarily imply that managerial learning from failure is highly accurate. A well-known problem in learning from failure is the ‘‘hot stove effect’’ of over-generalization (Denrell & March, 2001). Managers who observe a failure and seek to draw implications from it can over-generalize along two dimensions. The first is to reduce the interpretation of an uncertain outcome to that of a certain one, so that an action that was actually risky (high variance in outcome) comes to be seen as harmful (low mean of outcome). The second is to increase the interpretation of a cause to be broader than the actual cause, as when a firm entering a niche that it had low capabilities for competing in judges that all niche entry is harmful (this is the ‘‘hot stove effect’’ of avoiding any kind of stove after touching a hot one). Both forms of over-generalization are likely when managers learn from a small set of experiences (March et al., 1991). While they suggest that a given manager may draw incorrect implications from a given failure event, our findings imply that these errors seem to aggregate out somewhat: a population of firms will be more likely to avoid failure after observing a sample of failures. On the other hand, recent competition had a harmful effect in these findings, contrary to prior work finding that it benefits the firm (Barnett et al., 1994). An important difference in context may be that our findings are from a single population of firms (though with some niche differences), whereas the earlier work was in geographically delineated populations. It might well be that a geographically delimited subpopulation loses fitness when facing insufficient competition (as a cross sectional effect relative to other subpopulations), but that a single population of firms will tend to be less fit following a period of high density (as a longitudinal effect). If organizations that experience a peak in density followed by a drop have lower failure rates during the drop, as these findings suggest, then the following learning-selection effect is suggested: The experience of intense competition accelerates learning when it occurs, but not sufficiently to overcome the increased competition. After the
If it Doesn’t Kill You: Learning from Ecological Competition
265
density drops, this learning is retained and the competitive pressure is reduced, leading to lower failure rates. This retention of fitness is an important finding that is contrary to the ‘‘forgetting’’ effects known from learning curve research (Argote, 1999). Organizational inertia may play a positive role in fitness retention by creating barriers against modifying consequential organizational characteristics such as technology and market strategy. The effect of density delay suggests some limits on learning. We found that high founding density led to an increasingly higher rate of failure as the organization aged. Clearly, this contradicts a theory of organizations learning to overcome initial maladaptation, and instead suggests that tight niche packing has increasingly detrimental effects. Here, inertia in market strategy may be functioning to the disadvantage of organizations. Clearly, the prediction of tight niche packing was correct for this industry, as firms founded under conditions of high density tended to have narrow niches – often they were specialists in one form of insurance, and sometimes they were also geographically delineated. It is not clear exactly how large a casualty insurance firm needs to be in order to fully exploit advantages of scale, but it seems likely that the narrow-niche specialists were too small, and suffered the consequences of their scale disadvantage. Here, a somewhat less inert market position might have helped. The findings on the effects of change suggests another limit of learning, namely, ideology, and leads to speculation on whether the mutual form has any option but to have the right niche from the start. We found that the mutuals generally suffered from entering and exiting market niches, while stock companies benefited from niche change. The key difference between the forms is the ideology of mutuals, which seems to amplify the adverse effects of change. Mutuals have highly committed customers, and need to reciprocate by staying committed to their original market niche. Stock companies were less constrained and accordingly showed a net improvement of life chances as a result of changing. The fewer constraints on stock companies may also have facilitated search for good niches to target for expansion, whereas mutuals that sought to expand into additional niches may have had a need to justify the expansion as a natural extension of existing services. Our paper studies learning as a ‘population-level’ process precipitated by failures and realized through niche changes. Yet, it also has implications for organizational-level micro-processes concerning attention, experience, and improvisation that overcomes the weight of inertia. Work such as ours shows systematic differences in the effects of different types of experience, which is especially useful knowledge when juxtaposed with evidence on intraorganizational learning (Argote & Ophir, 2002). Our evidence on the
266
HENRICH R. GREVE AND HAYAGREEVA RAO
constraints imposed by a mutual form and by the competition at founding is supportive of micro-level research showing that organizational identities are consequential for behaviors and not easily changed (Dutton & Dukerich, 1991). Our findings on the effect of failures in the organizational population ought to inspire micro-level research on avoidance learning from the failures of others, which is a form of learning that clearly has received less attention than mimetic learning from the successes of others. Our evidence on how the accumulated number of changes benefit organizations raises the question of whether organizations become more flexible through a strengthening of regular change processes (March, 1981) or whether instead an ability to improvise is cultivated and realized as Crossan and Hurst (2006) point out in their companion paper. Although the research traditions on the intraorganizational processes of change and the adaptive consequences of change differ greatly in approach and level of analysis, they complement each other as tools for understanding organizational learning, and more work is needed to explore such complementarity. A number of research questions are suggested by the findings. We have shown that organizations benefit from the failures of others, but it is also of interest to examine whether they learn from the changes of others. Work on competitive dynamics have studied the performance of organizations as a function of their initiation of competitive moves and response to the competitive moves of others, which is a promising direction for developing a learning theory of competition (Chen & Hambrick, 1995; Smith, Ferrier, & Ndofor, 2001). On the other hand, this work has focused on the immediate effect of competitive moves rather than on the long-term learning effects that learning theorists would have emphasized. Our measures of niche entry and exit may be too rough to permit detailed examination of competitive dynamics, but we regard this as an important question for future research. We are also interested in exploring the effects of average density further. Clearly, work from additional contexts is needed to test our conjecture that the learning advantages of high density are a cross sectional effect rather than a temporal one. If this conjecture is correct, it would be different from ordinary density dependence effects, which can also be shown cross in cross sections (Greve, 2002). On the other hand, it could be that the average density is less important than the average age of competitors, or the average duration of their competitive interaction (Barnett & Hansen, 1996). Fine-grained and large datasets are needed to test these propositions against each other. We think that integrated development of the ecological and learning theory of niches is an important issue for future theorizing and empirical investigation. We have treated niche changes as a consequential change that
If it Doesn’t Kill You: Learning from Ecological Competition
267
organizations learn from and change their fitness with. Niche changes clearly have that role, but also play other roles in the organizational ecology. Elaboration of niches creates a web of market contacts in which some competitors meet directly, sometimes in multiple markets, and selectively learn from some competitors but not from others (Greve & Baum, 2001). This microstructure of competition has learning effects that have not been fully developed theoretically. We suspect, for example, that the effects of founding density and observed failures vary by niche because events in the focal niche are more consequential and salient. The effects may also vary by form, and we are not sure whether form or niche is the more relevant condition establishing which organizations learn from each other. The interchange of ideas between organizational ecology and organizational learning has opened a broad canvas of research questions that call for further exploration.
NOTES 1. General insurance includes all kinds of insurance except life insurance. It is sometimes referred to as nonlife insurance or property-liability insurance. 2. The data did show two dissolutions of mutuals that occurred nearly simultaneously with foundings of joint-stock firms in the same town and with similar name. Possibly these are the result of members dissolving a mutual with the intent of founding a joint-stock firm instead.
ACKNOWLEDGMENTS We are grateful to Thorolf Berg of Gjensidige NOR for help with source materials and advice on the insurance industry and to the Norwegian Research Council for research support (project 161318/V10). The paper was greatly improved by comments from Joel A. C. Baum, Stanislav D. Dobrev, and Arjen van Witteloostuijn. Hans-Christian Hustad and Sindre Bornstein provided valuable research assistance.
REFERENCES Amburgey, T. L., Kelly, D., & Barnett, W. P. (1993). Resetting the clock: The dynamics of organizational change and failure. Administrative Science Quarterly, 38, 51–73. Argote, L. (1999). Organizational learning: Creating, retaining, and transferring knowledge. Boston: Kluwer Academic Publishers.
268
HENRICH R. GREVE AND HAYAGREEVA RAO
Argote, L., & Ophir, R. (2002). Intraorganizational learning. In: J. A. C. Baum (Ed.), Companion to organizations (pp. 181–207). Oxford: Blackwell. Barnett, W. P., & Carroll, G. R. (1987). Competition and mutualism among early telephone companies. Administrative Science Quarterly, 32, 400–421. Barnett, W. P., & Carroll, G. R. (1995). Modeling internal organizational change. In: J. Hagan & K. S. Cook (Eds), Annual Review of Sociology (pp. 217–236). Greenwich, CT: JAI Press. Barnett, W. P., & Freeman, J. (2001). Too much of a good thing? Product proliferation and organizational failure. Organization Science, 12, 539–558. Barnett, W. P., Greve, H. R., & Park, D. Y. (1994). An evolutionary model of organizational performance. Strategic Management Journal, 15, 11–28. Barnett, W. P., & Hansen, M. T. (1996). The Red Queen in organizational evolution. Strategic Management Journal, 17, 139–157. Barnett, W. P., Swanson, A.-N., & Sorenson, O. (2003). Asymmetric selection among organizations. Industrial and Corporate Change, 12, 673–695. Barron, D. N., West, E., & Hannan, M. T. (1995). A time to grow and a time to die: Growth and mortality of credit unions in New York City, 1914–1990. American Journal of Sociology, 100, 381–421. Baum, J. A. C., & Ingram, P. (1998). Survival-enhancing learning in the Manhattan hotel industry, 1898–1980. Management Science, 44, 996–1016. Baum, J. A. C., & Singh, J. V. (1996). Dynamics of organizational responses to competition. Social Forces, 74, 1261–1297. Bromiley, P. (1991). Testing a causal model of corporate risk taking and performance. Academy of Management Journal, 34, 37–59. Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in populations of organizations. American Journal of Sociology, 90, 1262–1283. Carroll, G. R., & Hannan, M. T. (1989). Density delay in the evolution of organizational populations: A model and five empirical tests. Administrative Science Quarterly, 34, 411– 430. Chen, M.-J., & Hambrick, D. C. (1995). Speed, stealth, and selective attack: How small firms differ from large firms in competitive behavior. Academy of Management Journal, 38, 453–482. Chen, M.-J., & Miller, D. (1994). Competitive attack, retaliation and performance: An expectancy-valence framework. Strategic Management Journal, 15, 85–102. Chuang, Y.-T., & Baum, J. A. C. (2003). It’s all in the name: Failure-induced learning by multiunit chains. Administrative Science Quarterly, 48, 33–59. Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35, 128–152. Crossan, M. M., & Hurst, D. (2006). Strategic renewal as improvisation: Reconciling the tension between exploration and exploitation. In: J. A. C. Baum, S. D. Dobrev & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 275–303). Oxford UK: JAI/Elsevier. Cyert, R. M., & March, J. G. (1963). A behavioral theory of the firm. Englewood Cliffs, NJ: Prentice-Hall. Daft, R. L., & Weick, K. E. (1984). Toward a model of organizations as interpretation systems. Academy of Management Review, 9, 284–295.
If it Doesn’t Kill You: Learning from Ecological Competition
269
Denrell, J., & March, J. G. (2001). Adaptation as information restriction: The hot stove effect. Organization Science, 12, 523–538. Dobrev, S. D., Kim, T.-Y., & Carroll, G. R. (2002). The evolution of organizational niches: U.S. automobile manufacturers, 1885–1981. Administrative Science Quarterly, 47, 233–264. Dobrev, S. D., Kim, T.-Y., & Carroll, G. R. (2003). Shifting gears, shifting niches: Organizational inertia and change in the evolution of the U.S. automobile industry, 1885–1981. Organization Science, 14, 264–282. Dobrev, S. D., Kim, T.-Y., & Hannan, M. T. (2001). Dynamics of niche width and resource partitioning. American Journal of Sociology, 106, 1299–1337. Drazin, R., & Rao, H. (2002). Harnessing managerial knowledge to implement product-line extensions: How do mutual fund families allocate portfolio managers to old and new funds? Academy of Management Journal, 45, 609–619. Dutton, J. E., & Dukerich, J. M. (1991). Keeping an eye in the mirror: Image and identity in organizational adaptation. Academy of Management Journal, 34, 517–554. Fiegenbaum, A. (1990). Prospect theory and the risk-return association. Journal of Economic Behavior and Organization, 14, 184–203. Freeman, J., Carroll, G. R., & Hannan, M. T. (1983). The liability of newness: Age dependence in organizational death rates. American Sociological Review, 48, 692–710. Færden, K. (1967). Forsikringsvesenets Historie i Norge 1814–1914. Oslo: Wesmanns Skandinaviske Forsikringsfond. Greve, H. R. (1995). Jumping ship: The diffusion of strategy abandonment. Administrative Science Quarterly, 40, 444–473. Greve, H. R. (1996). Patterns of competition: The diffusion of a market position in radio broadcasting. Administrative Science Quarterly, 41, 29–60. Greve, H. R. (1999). The effect of change on performance: Inertia and regression toward the mean. Administrative Science Quarterly, 44, 590–614. Greve, H. R. (2002). An ecological theory of spatial evolution: Local density dependence in Tokyo banking, 1894–1936. Social Forces, 80, 847–879. Greve, H. R. (2003). Organizational learning from performance feedback: A behavioral perspective on innovation and change. Cambridge, UK: Cambridge University Press. Greve, H. R., & Baum, J. A. C. (2001). A multiunit, multimarket world. In: J. A. C. Baum & H. R. Greve (Eds), Multiunit organization and multimarket strategy: Advances in strategic management (Vol. 18, pp. 1–28). Oxford, UK: Elsevier. Greve, H. R., & Taylor, A. (2000). Innovations as catalysts for organizational change: Shifts in organizational cognition and search. Administrative Science Quarterly, 45, 54–80. Hannan, M. T., & Freeman, J. (1984). Structural inertia and organizational change. American Sociological Review, 49, 149–164. Hannan, M. T., & Freeman, J. (1989). Organizational Ecology. Cambridge, MA: Harvard University Press. Harrison, J. R., & March, J. G. (1984). Decision making and postdecision surprises. Administrative Science Quarterly, 29, 26–42. Hodne, F., & Grytten, O. H. (2000). Norsk Økonomi i Det 19. A˚rhundre (Norwegian Economy in the 19. Century). Bergen: Fagbokforlaget. Hodne, F., & Grytten, O. H. (2002). Norsk Økonomi i Det 20. A˚rhundre (Norwegian Economy in the 20. Century). Bergen: Fagbokforlaget.
270
HENRICH R. GREVE AND HAYAGREEVA RAO
Huber, G. P. (1991). Organizational learning: The contributing processes and the literatures. Organization Science, 2, 88–115. Ingram, P., & Baum, J. A. C. (1997). Opportunity and constraint: Organizations’ learning from the operating and competitive experience of industries. Strategic Management Journal, 18, 75–98. Jaffe, A. B., Trajtenberg, M., & Henderson, R. M. (1993). Geographic localization of knowledge spillovers as evidenced by patent citations. Quarterly Journal of Economics, 108, 577–598. Kim, T.-Y., Dobrev, S. D., & Solari, L. (2003). Festina lente: Learning and inertia among Italian automobile producers, 1896-1981. Industrial and Corporate Change, 12, 1279–1301. Levinthal, D. A. (1991a). Organizational adaptation and environmental selection: Interrelated processes of change. Organization Science, 2, 140–145. Levinthal, D. A. (1991b). Random walks and organizational mortality. Administrative Science Quarterly, 36, 397–420. Levitt, B., & March, J. G. (1988). Organizational learning. In: W. R. Scott, & J. Blake (Eds), Annual review of sociology (Vol. 14, pp. 319–340). Palo Alto, CA: Annual Reviews. Lorange, K. (1935). Forsikringsvesenets Historie i Norge Inntil 1814. Oslo: Wesmanns Skandinaviske Forsikringsfond. March, J. G. (1981). Footnotes to organizational change. Administrative Science Quarterly, 26, 563–577. March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2, 71–87. March, J. G., Sproull, L. S., & Tamuz, M. (1991). Learning from samples of one or fewer. Organization Science, 2, 1–13. Miner, A. S., Amburgey, T. L., & Stearns, T. M. (1990). Interorganizational linkages and population dynamics: Buffering and transformational shields. Administrative Science Quarterly, 35, 689–713. Miner, A. S., Kim, J.-Y. J., Holzinger, I. W., & Haunschild, P. R. (1999). Fruits of failure: Organizational failure and population-level learning. In: A. Miner & P. Anderson (Eds), Advances in strategic management (Vol. 16, pp. 187–220). Greenwich, CT: JAI Press. Porter, M. E. (1990). The competitive advantage of nations. New York: Free Press. Rao, H., & Drazin, R. (2002). Overcoming resource constraints on product innovation by recruiting talent from rivals: A study of the mutual fund industry, 1986–1994. Academy of Management Journal, 45, 491–507. Rivkin, J. W. (2001). Reproducing knowledge: Replication without imitation at moderate complexity. Organization Science, 12, 274–293. Schein, E. H. (1990). Organizational culture. American Psychologist, 45, 109–119. Smith, K. G., Ferrier, W. J., & Ndofor, H. (2001). Competitive dynamics research: Critique and future directions. In: M. A. Hitt, R. E. Freeman & J. S. Harrison (Eds), The Blackwell handbook of strategic management (pp. 315–361). Oxford: Blackwell. Staber, U. H. (1989). Organizational foundings in the cooperative sector of Atlantic Canada: An ecological perspective. Organization Studies, 10, 381–403. Stinchcombe, A. L. (1965). Social structure and organizations. In: J. G. March (Ed.), Handbook of organizations (pp. 142–193). Chicago: Rand McNally. Swaminathan, A. (1996). Environmental conditions at founding and organizational mortality: A trial-by-fire model. Academy of Management Journal, 39, 1350–1377.
If it Doesn’t Kill You: Learning from Ecological Competition
271
Tsai, W. (2001). Knowledge transfer in intraorganizational networks: Effects of network position and absorptive capacity on business unit innovation and performance. Academy of Management Journal, 44, 996–1004. Tushman, M. L., & Romanelli, E. (1985). Organizational evolution: A metamorphosis model of convergence and reorientation. In: B. M. Staw & L. L. Cummings (Eds), Research in organizational behavior (pp. 171–222). Greenwich, CT: JAI Press. Wigum, S. (1993). ‘‘Lykken Er Da˚rens Formynder’’: Historie Om Gjensidig Fiskeriforsikring i Norge (‘‘Luck is the Fool’s Guardian’’: A History of Mutual Fisheries Insurance in Norway). Bergen, Norway: Hegland Trykkeri.
This page intentionally left blank
272
STRATEGIC RENEWAL AS IMPROVISATION: RECONCILING THE TENSION BETWEEN EXPLORATION AND EXPLOITATION Mary M. Crossan and David K. Hurst ABSTRACT Management theory has paid scant attention to the nature and reconciliation of the tension between exploration and exploitation, in spite of its central importance to strategic renewal. This paper uses Hurst’s (1995) ecocycle to frame the tension and employs complexity theory to examine how the tension manifests itself across levels and time. Improvisation is advanced as a process to reconcile and manage the tension between exploration and exploitation.
A decade ago Baum (1996, p. 106) issued a clarion call: ‘‘Now is the time to expand the boundaries of ecological and adaptationist perspectives to create a combined approach that sees the processes of adaptation and selection as complementary and interacting.’’ The apparent contradictions between
Ecology and Strategy Advances in Strategic Management, Volume 23, 273–298 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23009-2
273
274
MARY M. CROSSAN AND DAVID K. HURST
adaptation at the level of the firm through organization learning and adaptation at the level of population through environmental selection can be best understood and resolved by examining the individual and organizational forces that can encourage learning and change in a context where ecological forces constrain choice and foster organizational inertia. This conflict mirrors the tension within firms identified by March (1991) between exploration (learning and change) and exploitation (routine and repetition). We employ an ecological perspective to understand the underlying tensions and then use research from the field of improvisation to examine how individuals and organizations can counter the ecological forces that manifest themselves in organization inertia. The chapter begins with a discussion of the tension between exploration and exploitation. It then presents Hurst’s (1995) ecocycle as a means of framing the tension. We use the ecocycle to describe a natural or evolutionary process of adaptation. Complexity theory is employed to discipline the analogies between natural systems and human systems, to build on the ecocycle, and to explain how firms can become trapped in either exploration or exploitation modes at the expense of the other. Finally, improvisation is offered as an approach to suggest ways in which managers can break away from the natural tendencies of complex systems. The ecocycle and complexity theory serve to describe and explain how firms tend to operate, while improvisation takes a normative perspective to suggest how firms might operate. Our intent is to move from description, in the discussion of the ecocycle and complexity theory, to provocative prescription, in the discussion of improvisation. Implications for research and management are then presented.
EXPLORATION AND EXPLOITATION March (1991, p. 71) succinctly described exploration, exploitation and the balance between the two. Exploration includes things captured by terms such as search, variation, risk taking, experimentation, play, flexibility, discovery, innovation. Exploitation includes such things as refinement, choice, production, efficiency, selection, implementation, execution. Adaptive systems that engage in exploration to the exclusion of exploitation are likely to find that they suffer the costs of experimentation without gaining many of its benefits. They exhibit too many undeveloped new ideas and too little distinctive competence. Conversely, systems that engage in exploitation to the exclusion of exploration are likely to find themselves trapped in suboptimal stable equilibria.
Strategic Renewal as Improvisation
275
Recognizing and managing the tension between exploration and exploitation is a ‘‘primary factor in system survival and prosperity’’ (March, 1991, p. 71) and one of the critical challenges of strategic renewal (Crossan, Lane, & White, 1999). This tension has been variously described as a tension between creation and maturity, flexibility and efficiency (Lant & Mezias, 1992), variation and selection (Ashby, 1960; Hannan & Freeman, 1987), feed-forward and feed-back (Crossan et al., 1999), inertia and stress (Huff, Huff, & Thomas, 1992) and mindful and less-mindful behavior (Levinthal & Rerup, 2005). Although the exploration/exploitation tension and related concepts have long been identified, theorists often find themselves on different sides of a theoretical gulf or even in a debate, and have tended to polarize rather than synthesize the discussion (Levinthal & Rerup, 2005). As a result, there has been little in the way of research that addresses the nature of the tension and how it might be reconciled. As Weick (1998, p. 551) explains: There is currently an abundance of conceptual dichotomies that tempt analysts to choose between things like control and innovation, exploitation and exploration, routine and non-routine, and automatic and controlled, when the issue in most organizations is one of proportion and simultaneity rather than choice.
Pfeffer (1982) identified three distinct perspectives of action that underpin different theoretical approaches, and tend to resist attempts at reconciliation (rational, constrained and emergent perspectives). The rational action perspective views action as purposive, intentionally or boundedly rational and prospective or goal-directed. The rational action perspective portrays managers as ‘‘knowing what to do, and free to act.’’ This is the central tenet of contingency theory (Lawrence & Lorsch, 1967): firms’ challenges are analyzable and managers can act to meet them. In contrast, the constrained action perspective views action as both internally and externally constrained or situationally determined. The best known expressions of this view are resource dependence theory (Pfeffer & Salancik, 1978) and population ecology (Hannan & Freeman, 1977). The constrained action perspective portrays managers as ‘‘knowing what to do, but not free to act.’’ Finally, the emergent action perspective portrays action as an unfolding process (March & Olsen, 1976) in an undecipherable environment. Actors are cognitively constrained and rationality is constructed after the fact in a process of retrospective sense making (Weick, 1979) or post hoc rationalization. The emergent action perspective portrays managers as ‘‘free to act, but not knowing what to do.’’
276
MARY M. CROSSAN AND DAVID K. HURST
When applied to the field of strategy the different perspectives of action can be seen to manifest themselves in the 10 schools of strategy identified by Mintzberg (1990), of which three have dominated the strategy literature (design school, planning school, positioning school). Mintzberg (1994, p. 2) describes the three dominant schools as follows: y the ‘‘design school’’ considers strategy making as an informal process of conception, typically in a leader’s conscious mind. The design school model, sometimes called SWOT also underlies the second, which I call the ‘‘planning school’’ and which accepts the premises of the former, save two – that the process be informal and the chief executive be the key actor y The third, which I call the ‘‘positioning school,’’ focuses on the content of strategies more than on the processes by which they are prescribed to be made.
Underpinning these three dominant schools is a rational perspective of strategy: a goal-oriented, instrumental rationality that reflects the origins of the concept in theories of industrial organization and neoclassical microeconomics. Six schools (cognitive, entrepreneurial, learning, political, cultural, and environmental) rely primarily on the constrained and emergent perspectives. Although they have received some attention, these schools remain largely disconnected from the mainstream discussion of strategy. Mintzberg, Ahlstrand, and Lampel (1998, p. 302) suggest that a tenth school, the configuration school, ‘‘differs from all the others in one fundamental respect: it offers the possibility of reconciliation, one way to integrate the messages of the other schools.’’ They highlight Hurst’s organizational ecocycle as a framework in the configuration school with potential to integrate the other schools. We selected Hurst’s framework not only because of this integrative capacity, but also because it can incorporate Pfeffer’s three perspectives on management action, while framing the tension between exploration and exploitation. It is not our intent to compare and contrast Hurst’s framework with Mintzberg’s 10 schools. Rather, since the framework draws heavily on all of the schools, we will briefly present the framework and show how it serves to integrate and extend the discussion into the arena of improvisation.
ORGANIZATIONS AS ECOSYSTEMS – CONNECTING THE PERSPECTIVES Hurst (1995) suggests that Pfeffer’s three perspectives are analogous to the phases of ecological succession to be found in the development of natural ecosystems (Holling, 1986; Gunderson & Holling, 2002) and that they can
277
Strategic Renewal as Improvisation
be connected via a well-defined process: the organizational ecocycle (Hurst & Zimmerman, 1994). Forests and other natural ecosystems are dynamically stable entities. While their elements live and die, forests survive through continual cycles of creation, growth, destruction, and renewal. This pattern is called an ecocycle to distinguish it from the more familiar life cycle (Greiner, 1972; Kimberly & Miles, 1980) that applies to individual organisms. A life cycle is generally depicted as an S-shaped, or logistics curve; ecocycle consists of two such curves arranged to form an infinite loop, as shown in Fig. 1. The ecocycle is consistent with the weak selection hypothesis (Singh, 2006) in which both adaptive learning within the organization and selection mechanisms at higher levels (population, community) interact to shape the evolution of organizations. Indeed, we can expect to find the ecocycle dynamic present at every level where whole systems are being studied. It has been suggested, for example, that whereas individual organizations excel at the exploitation of their environment, much exploration takes place at the population level (Levinthal & March, 1993; March, 1999) and there is empirical support for this view (Baum & Ingram, 1998; Baum, Li, & Usher,
The Organizational Ecocycle Emergent Actor
Weak
Rational Actor
Constrained Actor
CONNECTEDNESS Fig. 1.
The Organizational Ecocycle.
Strong
278
MARY M. CROSSAN AND DAVID K. HURST
2000; Greve & Rao, 2006). Thus, it seems likely that the roles of exploration and exploitation will never be reconciled without looking at multiple levels of analysis simultaneously (Singh, 2006). In this discussion of the ecocycle we begin at the systems level before going on to discuss its application to single organizations. It should be clear throughout this section, however, that individual organizations cannot be located unequivocally at any single level of analysis. Some business firms – diversified, multi-product companies for example – have characteristics of both a single organization and a population. Seen as a stylized, two-dimensional representation of a system’s trajectory in a multi-dimensional phase space, the ecocycle’s dimensions and the amplitude of its trajectories are the subject of debate (Ulanowicz, 1997). Nonetheless, the ecocycle concept is helpful in that its descriptive power makes it simple enough to be grasped while being complex enough to capture and use some of the key paradoxical elements (Poole & Van de Ven, 1989) encountered in the study of complex systems. The front loop (solid line) is the familiar, conventional life cycle. In human organizations it tracks the growth of a technical system (Emery & Trist, 1973) from birth through maturation to decline and crisis. The back loop (dotted line) of the ecocycle is the less familiar, renewal cycle of creative destruction and reconception that only higher-level systems can follow. It begins in the confused aftermath of a crisis. The constraints (both internal and external) that bind the system are shattered and the large hierarchical structures (trees in the case of the forest) that monopolize resources are fragmented. It ends with the creation of new contexts (fertile soil) with accessible resources (nutrients, water, sun) in which new, small-scale organisms (weeds, seedlings) can flourish, setting the stage for another cycle of birth and growth. In human organizations the back loop can be seen as the development trajectory of a social system as it evolves from a group of scattered individuals into a community. The spirals on either side of the ecocycle will be discussed in the following sections. There are several analogies between the development paths of natural and human ecosystems. (a) New growth, for example, emerges on the edges of each system and in open patches within it. In these places there is equal access to resources (which are relatively plentiful) and, initially, little competition. A wide variety of young, small-scale organisms – entrepreneurs in human ecosystems – can co-exist. In the open patch, anything grows – ecologists call these organisms ‘‘r-strategists’’ (MacArthur & Wilson, 1967) after r, the growth
Strategic Renewal as Improvisation
279
factor in the logistics equation. This entrepreneurial phase is characterized by trial-and-error learning with much experimentation and vicarious learning in an environment that is impossible to analyze. The dot-com boom of the late 1990s can be seen as an outbreak of ‘‘weeds’’ in the Western corporate ecology, with hundreds of small-scale organisms being showered with resources for which there was little competition. All that entrepreneurs had to do was to produce a business plan; although very few of their fanciful business models proved to be viable as the unknowable future unfolded. Such entrepreneurial action can be classified as pre-rational or emergent. Burgelman (1983, 2002) describes its corporate-level counterpart as autonomous strategy. (b) Over time, as the patches become crowded, classical Darwinian competition for resources breaks out. Organisms have either to specialize or to dominate as generalists. The more efficient users of these resources survive while many disappear. In environments that do not favor specialists, the overall variety of organisms in the ecosystem declines. In natural systems the mid-point of this stage is signalled by the appearance of a so-called ‘‘climax species’’ that outcompetes all others. In technical systems the process is catalyzed by the emergence of a ‘‘dominant design’’ at the inflexion point of the curve, which signals a change from product innovation to process innovation (Utterback, 1994). As structural inertia theory (Hannan & Freeman, 1984) suggests, for human organizations to survive into this phase, high reliability of performance and high levels of accountability will be required. This is achieved by institutionalization of organizational purposes (which by now are well understood) and the standardization of routines: strategy that was previously autonomous is now deliberately induced (Burgelman, 1983, 2002). Survivors of the dot-com bust of the early 2000s were companies like Amazon, E-Bay, Google and Yahoo, firms that had developed clear, tested strategies and had invested in the infrastructure and networks necessary to garner resources in a competitive environment. This survival requirement, however, as Hannan and Freeman (1984) point out, is a double-edged sword, with the inertia that accompanies it creating internal constraints on the organization’s capacity to change in the future. (c) Eventually, growth slows and the patch becomes mature. Organisms become constrained internally as more and more resources are bound up and integrated within the large-scale organizations that now dominate the ecosystem. The carrying capacity (K) of the ecospace has been reached, hence the description by ecologists of organisms in this phase of the cycle as K-strategists (MacArthur & Wilson, 1967). Now few resources may be
280
MARY M. CROSSAN AND DAVID K. HURST
available for new growth: in natural systems the stage is set for a destructive release of resources from the hierarchical structures that bind them. In human systems this is a period of slower expansion as economies of scale (if any) are achieved, markets become saturated, and the structure of the industry becomes settled. Products that were once considered unique may by now have become generic. Firms and their managers are now constrained by a large scale, tightly connected network of structures, systems, and technology, all of which they have introduced over time to embed the original value-creating process (Foster & Kaplan, 2001). In the new contexts created by changing environments, these structures, processes, and embedded technologies, once considered strengths, may become weaknesses, hampering the efforts of the firm to change (Christensen, 1997). With opportunities for internal organizational adaptation constrained, selection by the environment takes over: organizations in this constrained phase may be vulnerable to sudden changes in the social, political, and technical contexts in which they operate. The well-documented woes of the ‘‘Big Three’’ automobile manufacturers and their suppliers, the so-called ‘‘legacy’’ airlines and many integrated steel mills are illustrative of the daunting challenges that companies and their managers can face in this phase of the ecocycle. The resources they control are being released back into the environment as they abandon products and markets, shutter plants, downsize their workforces and cycle through bankruptcy proceedings. In these examples, descriptions (a) and (c) are examples of emergent and constrained behaviors encountered in both natural and human ecosystems. Natural systems display only these two phases. In human ecosystems, however, a transitional phase exists between emergence and constraint. It is characterized by increasing rationality – the ability to associate cause with effect and to measure and calculate – while actors retain their ability to act. It is precisely this capacity for participants to think, act and learn that allows intelligent systems to suspend the natural tendency of all complex systems to run to ruin. From an evolutionary perspective Lamarckian inheritance processes trump Darwinian selection mechanisms in this phase of an intelligent system’s development. Although exploration and exploitation have been presented as separate loops in the ecocycle, both activities occur simultaneously as described below. Indeed a healthy ecosystem will consist of patches containing varying mixes of both activities. Depending upon the relative scales of the system and its observers, this mix of activities may be difficult to spot. Most forests, for example, look darkish green (indicative of the late exploitation phase)
Strategic Renewal as Improvisation
281
because the destructive precursors to the exploration phase (fire in the case of many forests) work at much finer scales in space and time than the growth processes on the front loop of the ecocycle. This is another example of how exploitation and exploration may function at different levels of analysis. In the section that follows we use complexity theory to tighten our comparative ‘‘mapping’’ between natural and human systems and to preserve the principle of systematicity, whereby higher order relations (i.e. relations between relations) are preserved while lower order relations are dropped (Tsoukas, 1991).
THE MANAGERIAL CHALLENGE – SEARCHING FOR THE ‘‘SWEET SPOT’’ Ecosystems can be viewed as complex adaptive systems (CAS) the generic properties of which may be studied under the rubric of complexity theory (Waldrop, 1992). Composed of interacting agents that exhibit emergence and self-organization, CAS are open, dynamic, so-called dissipative systems (Prigogine & Stengers, 1984) that function most adaptively on the boundary between order and disorder, known as the edge of chaos (Kauffman, 1995). The use of complexity theory in management (e.g. Stacey, 1991, 1992; Organization Science, 1999) represents researchers’ most recent attempt to introduce an explicit systems approach to the study of organizations (Academy of Management Journal, 1972; Ashmos & Huber, 1987). Complexity theory stresses the turbulent and unpredictable nature of environments that require organizational flexibility, resilience, and the capacity to both explore and exploit (Crossan, Lane, White, & Klus, 1996). Brown and Eisenhardt (1997, 1998) used it to describe the concept of competing on the edge of chaos, and as a theoretical starting point to suggest that organizational change can arise when ‘‘order springs from chaos’’ (1998, p. 14). The ecocycle, as a pattern that represents the trajectory of an ecosystem in dynamic equilibrium, may represent what is known in complexity theory as a chaotic or strange attractor. In dynamic systems, attractors are the patterns of a system’s trajectories that seem to act as basins to which the system continually returns, although never in a predictable way. The Lorenz attractor (Lorenz, 1963), one of the best known, has a distinctive ‘‘butterfly’’ shape not dissimilar to that of the ecocycle. As organizations traverse the ecocycle’s double loop, they can become trapped in spirals at either the exploration or exploitation ends of the loop (see Fig. 1).
282
MARY M. CROSSAN AND DAVID K. HURST
The management literature supports our claim that organizations are attracted to either exploration or exploitation. Ghemawat and Costa (1993) analyzed the tension between static and dynamic efficiency and found a tendency toward the extremes. Miller and Friesen (1980, 1982) found the same polarization, describing it in terms of the momentum of entrepreneurial and conservative firms. They observed within entrepreneurial (exploration-oriented) firms and conservative (exploitation-oriented) firms that past practices, trends and strategies tend to keep evolving in the same direction, perhaps eventually reaching dysfunctional extremes. They built on this notion to propose innovation strategies for each type of firm, suggesting that the challenges each faces are quite different. This is consistent with the ecocycle model; moving from exploration to exploitation is different than moving from exploitation to exploration. Levinthal and March (1993, p. 105) describe the repercussions of getting locked into either loop of the ecocycle: An organization that engages exclusively in exploration will ordinarily suffer from the fact that it never gains the returns of its knowledge. An organization that engages exclusively in exploitation will ordinarily suffer from obsolescence. The basic problem confronting an organization is to engage in sufficient exploitation to ensure its current viability and, at the same time, to devote enough energy to exploration to ensure its future viability. Survival requires a balance, and the precise mix of exploitation and exploration that is optimal is hard to specify.
If managers are to avoid being trapped in either of the two spirals in the ecocycle they have to find a way to have the organization ‘‘dwell’’ in the central ‘‘Rational Actor’’ phase of the cycle. When they reach the stage where strategy is induced rather than autonomous and the requirements for organizational reproducibility have been met, they will naturally want to exploit their ecological niche using their newly developed core competencies. If this is all they do, however, sooner or later (every industry and technology has its own scale and tempo) (Brown & Eisenhardt, 1998), they will be swept into the right-hand spiral, from which escape is extremely difficult. They will need to preempt this process by ‘‘tacking’’ against the prevailing exploitation ‘‘wind’’ to explore. Such an ‘‘ambidextrous’’ organization (Tushman & O’Reilly, 1997) would be able to loiter profitably in what might be thought of as the organization’s ‘‘sweet spot,’’ that elusive epiphany in space and time where minimal effort produces maximal result. Here the single organization can take on some of the protean survival attributes of a polymorphic system rather than remain a specialized entity within such a system: it can become more like a ‘‘forest’’ and rather less like a ‘‘tree.’’ Such an organization would have architectures that uses tight and loose coupling
283
Strategic Renewal as Improvisation
simultaneously – units are tight-coupled internally but loose-coupled with other sub-units within the same organization (Benner & Tushman, 2003). This suggests that organizations can be capable of taking on a fractal dimension (Mandelbrot, 1983) between specialism and generalism. The trajectory of such an organization is illustrated in Fig. 2. The oscillation between different organizational forms on the part of managers has been observed empirically but, absent changes in environmental conditions to which they are adjusting, it has usually been seen as indecisive vacillation rather than intelligent adaptation. Nickerson and Zenger (2002) suggest, however, that the challenge for managers is that their choices of formal structure do not define an organization’s actual functionality, which is reflected by the informal structure. Using Tichy and Fombrum’s (1979) metaphor they describe formal structure as comprising the ‘‘pegs’’ on which the emergent network of the informal structure hangs. The work of Siggelkow and Levinthal (2005) shows how under certain conditions different ‘‘pegs’’ (organizational structures) may result in different competency traps or sticking points. Under such circumstances
The Organizational “SweetSpot” Emergent Actor
Rational Actor
Constrained Actor
The“Sweet Spot”
Weak
CONNECTEDNESS Fig. 2.
The Organizational ‘‘Sweet Spot.’’
Strong
284
MARY M. CROSSAN AND DAVID K. HURST
a sequence of different organizational structures may be beneficial to the organization’s performance. Strictly speaking the ‘‘tacking’’ trajectory in Fig. 2 implies a sequential approach to exploration and exploitation, rather than the simultaneous approach usually associated with improvisation. Like many apparent contradictions, however, this one is easily resolved by specifying the scale at which the dynamic is being observed – a rapid, fine-grained oscillation between exploration and exploitation, for example, would appear to many observers to be simultaneous rather than sequential. Changes in the organization’s formal structure, the ‘‘pegs’’, on the other hand, are discrete moves and may have significant costs associated with them. One of the advantages of improvisation is that it can be used as a fine-grained tool to operate on the informal organization directly, pre-figuring any more formal change activities and enhancing their effectiveness once such changes have been made.
IMPROVISATION The foregoing discussion serves to describe the tension between the processes of exploration and exploitation using Hurst’s ecocycle. Some theorists have responded to this tension by advocating a contingent approach that argues for alignment between the organization, strategy and environment along the lines of mechanistic organizations focusing on exploitation in mature and stable environments or organic organizations (Burns & Stalker, 1961) favoring exploration in more dynamic environments (Rowley, Behrens, & Krackhardt, 2000). Others have suggested ‘‘ambidextrous’’ organizations (Tushman & O’Reilly, 1997) to manage the tension. In contrast, we introduce improvisation as a more fine-grained approach that captures the simultaneous application of exploration and exploitation. While there is a contingent nature to the effectiveness of improvisation it needs to be understood at a finer-grained level than its alignment with stable or dynamic environments, as discussed in more detail below. With the managerial challenge framed by the ecocycle, we can delve more deeply into the co-existence of exploitation and exploration. As Crossan (1998) points out, improvisation is more than a metaphor; it is both a perspective and a technique that has direct applications in the field of management. In this section we will develop the theoretical links and create a bridge from theory to practice. Improvisation theory has drawn heavily from the study of jazz and theater improvisation to examine a process that has been variously described as
Strategic Renewal as Improvisation
285
imagination guiding action in an unplanned way (Chase, 1988), the ability to make do with available resources (Weick, 1993a), intuition that incorporates creation and execution at the same time (Solomon, 1986), and the convergence of composition and execution (Moorman & Miner, 1998a). According to Weick (1998, p. 551) improvisation acknowledges the simultaneous presence of a variety of concepts, including exploration and exploitation that have been treated as dichotomies. Improvisation is a mixture of the pre-composed and the spontaneous, just as organizational action mixes together some proportion of control with innovation, exploitation with exploration, routine with non-routine, automatic with controlled. The normally useful concepts of routine y and innovation y have become less powerful as they have been stretched informally to include improvisation. Thus a routine becomes something both repetitious and novel, and the same is true for innovation. A similar loss of precision y has occurred in the case of decision making where presumptions of classical rationality are increasingly altered to incorporate tendencies toward spontaneous revision. Neither decisions nor rationality can be recognized in the resulting hodge podge. What is common among all of these instances of lost precision is that they attempt to acknowledge the existence of improvisation, but do so without giving up the prior commitment to stability and order in the form of habit, repetition, automatic thinking, rational constraints, formalization, culture, and standardization.
It is this inherent quality of the simultaneous treatment of exploration and exploitation that is of particular interest as it provides a unique perspective on how these processes can be managed. In pursuit of applying improvisation to understanding organization structure (Hatch, 1998a, 1998b), organization memory (Moorman & Miner, 1997), new product development (Moorman & Miner, 1998b), change management (Orlikowski & Hofman, 1997), organizational redesign (Weick, 1993b), organizational analysis (1998), strategy (Perry, 1991, 1994; Crossan et al., 1996), and organizational learning (Crossan & Sorrenti, 1997) improvisation theory has become much more refined. One critical aspect of the developing theory has been the parsing of its descriptive and prescriptive elements. Drawing heavily from jazz and theater improvisation where performance has essentially been built into the phenomenon, researchers have largely observed ‘‘effective’’ improvisation. As a result, improvisation theory has tended to blend descriptive and prescriptive elements. In response, Vera and Crossan (2005, p. 205) parsed the descriptive and prescriptive elements of improvisation defining improvisation as the ‘‘creative and spontaneous process of trying to achieve an objective in a new way.’’1 They propose five factors that impact the effectiveness of improvisation: expertise, teamwork skills, experimental culture, real-time information and communication and memory.
286
MARY M. CROSSAN AND DAVID K. HURST
Improvisation is an activity that takes place on the edge of chaos, on the constraint curves that bound every CAS. Improvisation never takes place in a vacuum: it takes place in a context that comprises both discipline (exploitation) and freedom (exploration). But it is a constrained freedom. Improvisation can be more appropriately characterized as a process through which dichotomies or paradoxes are vetted in action. Through improvisation the apparent paradox between exploration and exploitation is resolved. Routines and prior knowledge (exploitation) are ingredients or inputs to improvisation (Weick, 1998) that are blended in a creative and spontaneous process to produce novel outcomes (exploration). It is a process, which by its nature is able to accommodate the ambiguity, complexity and diversity of inputs. It is not our intent to cover the domain of improvisation theory, but rather to focus on three areas of particular interest: (1) the meaning of spontaneity; (2) the manifestation of improvisation across levels of the organization; and (3) the challenges associated with improvisation.
Spontaneity Theorists have suggested that what sets improvisation apart from processes of creativity and innovation is the dimension of spontaneity. Although this dimension has attracted interest for its application to situations where there is no time to plan (e.g. crisis situations) or where the future is so uncertain that planning may be of little value, the implications of spontaneity have wider application. While spontaneity is not a well-understood theoretical construct, it is anchored in one that has deep roots and along with it, much complexity – the study of time. Crossan, Cunha, Vera, and Cunha (2005) delve into the relationship between time and improvisation and demonstrate how improvisation can be used to resolve two major time dichotomies associated with organizational phenomena: clock versus event time and linear versus cyclical time. It is these potentially conflicting time perspectives that underpin much of the discussion about exploration and exploitation. For example, routines and the notion of path dependency in which ‘‘history matters’’ is anchored in a linear concept of time in which the past weighs heavily on the future. Indeed, Garud and Karnoe (2001) offer ‘‘path creation’’ as a contrasting perspective to path dependency in which ‘‘entrepreneurs meaningfully navigate a flow of events even as they constitute them’’ (p. 2). Many of the concepts they touch on – time, bricolage, and mindfulness – are important
Strategic Renewal as Improvisation
287
elements of improvisation and Garud and Karnoe certainly acknowledge improvisation as ‘‘a way of navigating and shaping emerging processes’’ (p. 24). Essentially, improvisation is the theoretical frame that can be employed to both understand and manage the tension between exploration and exploitation. As Crossan et al. (2005, p. 129) suggest: ‘‘through the effective use of improvisation processes, individuals and groups in organizations cope with, and coordinate the conflicting demands of the co-existing time perspectives.’’ The spontaneous aspect of improvisation emphasizes the ‘‘in the present’’ orientation of the process. It is a process that blends the past, present and future simultaneously. As Crossan et al. (2005, p. 140) note: ‘‘the improvisational process enables individuals and groups to draw on their potential capacity to access the past (without intending to repeat or replicate it) and to enhance and enrich action through a future-oriented vision and a unique collective experience of the present. This is seen in jazz improvisation: group members bring a rich repertoire of musical skill and memory and seek to enhance it through the collective experience of composing and playing in the moment.’’ Perhaps even more provocative is Tulku’s (1980, p. 37) hint at the possibility of an increased level of ‘‘knowingness’’ when we can ‘‘contract more of the space and time that are available in each apparently fixed and limited interval of ordinary time.’’
Improvisation Across Levels A powerful aspect of improvisation theory is that it spans levels of the organization. This is perhaps seen most heavily in the training that underpins theater improvisation. Training at the individual level involves several key areas: being able to simultaneously rely on and break out of existing cognitive structures; being in the present, and being better prepared to risk the 4Cs – the desire to be consistent, comfortable, competent and confident (Claxton, 1984). It is important to note that effective improvisation requires a solid base of technical skills and experience in the domain in which it is employed (Crossan, 1998; Hatch, 1998b; Weick, 1998), often referred to as expertise. At the group level, there is a focus on listening and communication with a particular emphasis on what improvisers call ‘‘yes anding,’’ which means building on the ideas of others rather than blocking those ideas. Although it was the apparent lack of organization level constructs or artifacts that initially attracted researchers to improvisation, researchers quickly discovered
288
MARY M. CROSSAN AND DAVID K. HURST
this was not the case. In particular, there are important organization culture elements that support the risk-taking at the individual level, and collaboration at the group level. Effective improvisation appears to operate with a minimal set of constraints that center on process including being in the present, yes anding, and never leaving a fellow actor hanging. Table 1 provides a summary of the characteristics of improvisation across levels as identified in prior research. It is this spanning of levels that provides the connective tissue between the individual and organization, providing a fine-grained view of how the organization impacts individuals and groups (exploitation) and how individuals and groups can affect what becomes institutionalized at the organization level (exploration). To recap: Hurst’s ecocycle serves to frame the tension between exploration and exploitation, and improvisation serves to reconcile or resolve it. We view the management of constraints as being a critical aspect of managing the exploration/exploitation tension. Through improvisation, constraints are discovered and vetted. Improvisation itself has a minimal set of constraints, yet there are challenges in managing improvisation. By examining the challenges associated with improvisation we are able to see more clearly the challenges associated with reconciling exploration and exploitation and hence strategic renewal and the conditions under which improvisation is more or less effective.
Challenges to Improvisation and Renewal We propose four significant challenges associated with improvisation: (1) employing improvisation in high velocity and turbulent environments; (2) identifying and managing the tolerance for error; (3) developing individual and group improvisational skill; and (4) managing memory, since memory both aids and impedes improvisation. We discuss each of these challenges below. High Velocity Environments Brown and Eisenhardt (1997, 1998) see improvisation as ideally suited to high velocity environments. Crossan and Sorrenti (1997) suggest that improvisation is characterized by a high degree of spontaneity – it is by definition a flexible and responsive approach to deal with a rapidly changing environment. However, Weick (1998) notes that under pressure of time, individuals tend to revert to familiar and comfortable patterns of action.
Table 1.
Group
Organization
OrganizationEnvironment
Willingness to forego planning and rehearsing in favour of acting in real timeb Well-developed understanding of internal resources and the materials at handa Proficient without blueprints and diagnosisa Ability to identify minimal structures for embellishinga Predisposed to recognize partial relevance of previous experience to present noveltya High confidence to deal with nonroutine eventsa Skilful at paying attention to performance of othersa,b Preference for and comfort with process rather than structure, which makes it easier to work with ongoing development, restructuring, and realization of outcomes, and easier to postpone the question, what will it amount to?a Willingness to risk the ‘‘4 Cs’’ – the desire to be consistent, comfortable, confident and competentb Solid base of technical skilla,b,c,d Individuals take the lead at different timesb,c Ability to agree on minimal structures for embellishinga Presents of associates similarly committed to and competent at impromptu making doa,b Skilful at building on performance of others to keep the interaction going and to set up interesting possibilities for one anothera,b,a Ability to maintain the pace and tempo at which others are extemporizinga Focused on coordination here and now and not distracted by memories or anticipationa,b Rich and meaningful set of themes, fragments, or phrases on which to draw for ongoing lines of actiona,d Common goalb Tolerance for error within organization – especially reward systemsb Culture of friendship vs. professionalismb Emotional tension and releasec Real-time information flowsd Communion among players and audience membersc Customer has a tolerance for errorb Environmental turbulence and unpredictability requires improvisational capabilityb,d Need to be open to cues from the environmentb
Strategic Renewal as Improvisation
Individual
Characteristics of Improvisation.
a
Weick, 1998. Crossan et al., 1996. c Crossan and Sorrenti, 1997; Hatch, 1998b. d Moorman and Miner, 1997, 1998. b
289
290
MARY M. CROSSAN AND DAVID K. HURST
Therefore, while high velocity and turbulent environments may be ideally suited to improvisation, improvisational action may be constrained by psychological tendencies to rely on familiar routines. To understand why individuals may choose (or choose not) to improvise, it is necessary to examine the other key challenges of improvisation. Tolerance for Error To remain on the edge of chaos is to manage the tension between exploration and exploitation. We suggest that one of the primary management tasks is to manage the tolerance for error. Improvisation requires a tolerance for error because error is inherent in the process it comprises: experimentation, innovation and exploration. The challenge for organizations is to assess when and where they can tolerate error (explore) and when and where they must produce and perform (exploit), perhaps flawlessly. For example, Crossan et al. (1996) point out that there is no tolerance for error when a jet takes off, but there is ample tolerance for error in flight simulators. Crossan and Sorrenti (1997) point out that the environment poses constraints on improvisation. These constraints relate directly to tolerance for error. While these constraints may appear fixed, we argue that they are often negotiable. Creating room for improvisation requires negotiating the constraints and therefore the tolerance for error. For example, organizational constraints, such as compensation systems may create a low tolerance for error if individuals are punished for making mistakes. In contrast, as Tretheway and McDougal (1998, p. 65) describe ‘‘when a new product at Owens Corning fails, no one gets fired. The company celebrates the failure by throwing a party and giving each team member a savings bond.’’ The well-known story of the ‘‘failed’’ glue that transformed into the ‘‘Post-it note’’ shows how 3M manages its tolerance for error by minimizing the constraints that restrict the ability of individuals to improvise. While there is much to be managed internally, one of the major constraints on tolerance for error lies externally with the customer. In the arenas of jazz or theater, where improvisation was first studied, customers or audiences expect that, while there will be moments of genius in composition, there will also be many less-than-flawless performances. This kind of error tolerance is not as prevalent when the same customers attend more traditional productions and concerts, which are designed to perform as flawlessly as possible, that is, to exploit. In the management arena, there is a need to negotiate with the customer to ensure that exploitation requirements do not drive out exploration requirements. A tangible example of this is in the area of ‘‘quality.’’ Where
Strategic Renewal as Improvisation
291
quality was once defined more narrowly around product specifications, the definition has been adapted to incorporate areas such as innovation, thereby providing greater flexibility in action (Sobek, Ward, & Liker, 1999). Learning from failure not only occurs within an organization, but also across organizations. However, Greve and Rao caution that ‘‘when managers avoid strategies that are associated with the failure of other organizations or strategies that other organizations have withdrawn from, they forfeit opportunities to tinker with those strategies’’ (Rao, p. 23, 2006). Developing Individual and Group Improvisational Skills Tolerance for error manifests itself at both the individual and group levels. Improvisation requires risking what Claxton (1984) referred to as the four Cs: the desire to be consistent, comfortable, competent and confident. In rapidly changing and unpredictable environments, individuals are drawn out of their comfort and competence zones. Given the strong psychological tendency to try and preserve these comfort zones, one of the key challenges of improvisation is to develop the individual and team skills to risk the four Cs. Research on improvisation in music and theater suggests individual and group techniques to develop improvisational skill. At the individual level, improvisation training teaches individuals to access creative thinking and silence overly analytical orientations (emphasize exploration over exploitation) by carrying out contradictory actions (Crossan et al., 1996). The tendency to rely on the familiar is actively managed by purposefully avoiding familiar patterns. Individuals will only be willing to escape familiar patterns if they are prepared to take some psychological risk. While the organizational tolerance for error will impact this willingness, individuals must also possess a comfort level with making mistakes. Improvisers call this ‘‘developing the psychological risk muscle.’’ In addition, this willingness also depends on the group’s willingness to engage improvisation. Key to improvisation is building on the offers of others through a process called ‘‘yes-anding.’’ Actors are trained to be open to the ‘‘offers’’ or opportunities presented by their fellow actors, and by members of the audience. The importance of the individual and group process to improvise, and in particular to break out of familiar patterns, is heightened when the role of memory in improvisation is examined. Memory Mezias and Glynn (1993), drawing on the work of Cohen and Levinthal (1990) and Damanpour (1991), suggest that ‘‘organizations with more
292
MARY M. CROSSAN AND DAVID K. HURST
expansive knowledge bases are better able to take advantage of opportunistic search and the serendipitous discoveries it may yield’’ (p. 94). While this may be true, improvisation reveals a paradox or tension associated with the role of memory. Just as improvisation builds on memory or structure, it also works to escape it in order to innovate and renew. In their study of new product development Moorman and Miner (1998b) investigated the incidence and effectiveness of improvisation. They found improvisation to be prevalent and to occur when organizational memory is low but environmental turbulence high. They also found that when organizational memory is low, improvisation has a negative effect on design effectiveness, cost efficiency and time efficiency. When organizational memory is high, however, improvisation has a positive effect on the same outcomes. They suggest that memory may be an impediment to improvisational activity, but may facilitate its effectiveness. Low memory may be associated with young organizations that have little institutional history and with organizations that fail to institutionalize learning (Crossan et al., 1999). Sorensen and Stuart (2000) contend that growth and age lead to increasing amounts of innovation of decreasing relevance to external issues. Young firms produce fewer innovations, but are far more likely to be working on relevant external issues than their older counterparts. In the case of young firms, memory does not impede improvisation and its associated innovation. However, as discussed previously, young firms are more likely to be trapped in the innovative exploration loop. It appears that memory may create a selective perception and reliance on familiar ways that impedes seeing things in new and different ways. Barr, Stimpert, and Huff (1992) nevertheless propose that ‘‘organizational renewal requires managers to change their mental models in response to environmental changes and that delays in this process will be associated with decline’’ (p. 16). Improvisation reconciles the tension between reliance on memory and escape from it through a process that acknowledges expertise but, as discussed previously, attempts to ensure that the expertise does not drive out innovation. Nevertheless, improvisation varies in degree based on the extent to which it relies on current thinking. In this regard Weick (1998) cites Konitz to suggest a continuum that ranges from interpretation, through embellishment and variation ending in improvisation. In summary, improvisation is a process structured in a way that enables individuals and organizations to discover and manage constraints that, we claim, are critical to managing the exploration/exploitation tension. The improvisation process has its own restrictive and enabling constraints,
Strategic Renewal as Improvisation
293
apparent in the efficacy of memory, the demands of high-velocity environments, the level of tolerance for error, and the development of individual and group improvisational skills. While several management processes encompass aspects of improvisation, we are aware of no process as comprehensive or essential to our understanding of how the tensions between exploration and exploitation are managed.
IMPLICATIONS FOR RESEARCH AND MANAGEMENT The perspective offered in this paper suggests several key points. Exploration and exploitation do not represent a dichotomy. A more relevant analysis emphasizes their concurrent existence and their relative proportions (Weick, 1998). The exploration and exploitation loops are attractors and to avoid getting locked into a dysfunctional spiral, organizations need to actively manage both processes. Strategic management over-emphasizes the process of exploitation and fails to address how firms manage the tension between exploration and exploitation. The concept of the ecocycle aptly frames this tension, and the concept of improvisation provides an approach to manage it. While researchers have not looked at mechanisms to manage the exploration/exploitation tension, managers have tended to resolve the tension using an approach similar to that of researchers: the separation of exploration and exploitation in time and space. On a functional basis, exploration has often been reserved for research and development, with exploitation being the domain of manufacturing. However, innovative Japanese manufacturers demonstrated the pitfalls of this false dichotomy (Womack & Jones, 1996). Firms have also tended to set up different organizational structures and systems to deal with the two as separate entities, spinning off many new businesses in the process. Few businesses, however, can focus solely on either exploration or exploitation; separating the two may reinforce or feed the negative aspects of the two processes that lead to dysfunctional attractors. For 90 years the U.S. Parks Service suppressed fires in national parks, intending to preserve the forests in their pristine state. The policy seemed to work in the early years, but fires grew more difficult to suppress as the forests aged and the ecosystems became brittle and dry. Despite abandoning the policy in the early 1970s the Parks Service was unable to avert massive
294
MARY M. CROSSAN AND DAVID K. HURST
fires such as those that devastated Yellowstone National Park in 1988 (Jeffery, 1989). Forestry managers now understand that fire cannot be kept out of forests indefinitely. Mature forests need fire to break the constraints on new growth; fire suppression only makes the inevitable blazes catastrophic. Though it seemed obvious and perfectly rational to put out fires, foresters were actually doing more harm than good in their attempts to manage the forest. Similarly, in organizations, management may be doing more harm than good if they fail to understand the underlying system and the constraints that need to be managed. Although strategic management researchers recognize the importance of managing on the edge of chaos (and thereby managing the tension between exploration and exploitation) there has been little in the way of theory to guide them. This paper has attempted to pursue one of the promising areas advocated by Brown and Eisenhardt: improvisation. Distilling the literature on improvisation, we presented four key challenges associated with implementing improvisation. We know there is a need for a strong technical base to improvise effectively, yet it is difficult to escape it, particularly under time pressure. We discussed how individuals (through developing the risk muscle), groups (through ‘‘yes-anding’’), and organizations (through managing the tolerance for error) could develop and manage improvisation. Foresters now prescribe fire as a way of creating open patches for renewal on their scale and timetable. In mature organizations improvisation can be seen as a prescribed burn, a process that managers can use at several levels to create open patches in space and time for exploration and renewal. Their consequent ability to control their own destinies is the central vision of evolutionary engineering (March, 1994).
NOTES 1. For a comprehensive list of definitions see Moorman and Miner (1997).
ACKNOWLEDGMENTS We want to thank the editors for their extremely helpful comments on the many drafts of this paper and for their patient support while the paper was in preparation.
Strategic Renewal as Improvisation
295
REFERENCES Academy of Management Journal. (1972). Theme issue: General systems theory. 15, 403–440 Ashby, W. R. (1960). Design for a brain (2nd ed.). New York: Wiley. Ashmos, D. P., & Huber, G. P. (1987). The systems paradigm in organization theory: Correcting the record and suggesting the future. Academy of Management Review, 12(4), 607–621. Barr, P. S., Stimpert, J. L., & Huff, A. S. (1992). Cognitive change, strategic action, and organizational renewal. Strategic Management Journal, 13, 15–36. Baum, J. A. C. (1996). Organizational ecology. In: S. R. Clegg, C. Hardy, & W. R. Nord (Eds). Handbook of organization studies (pp. 77–114) Thousand Oaks, CA: Sage. Baum, J. A. C., & Ingram, P. (1998). Survival-enhancing learning in the Manhattan hotel industry, 1898–1980. Management Science, 44(7), 996–1016. Baum, J. A. C., Li, S. X., & Usher, J. M. (2000). Making the next move: How experiential and vicarious learning shape the locations of chains acquisition. Administrative Science Quarterly, 45(4), 766–802. Benner, M. J., & Tushman, M. L. (2003). Exploitation, exploration and process management: The productivity dilemma revisited. Academy of Management Review, 28(2), 238–256. Brown, S. L., & Eisenhardt, K. M. (1997). The art of continuous change: Linking complexity theory and time-passed evolution in relentlessly shifting organizations. Administrative Science Quarterly, 42(1), 1–34. Brown, S. L., & Eisenhardt, K. M. (1998). Competing on the edge: Strategy as structured chaos. Boston, MA: Harvard Business School Press. Burgelman, R. A. (1983). A model of the interaction of strategic behaviour, corporate context, and the concept of strategy. Academy of Management Review, 8(1), 61–70. Burgelman, R. A. (2002). Strategy is destiny. New York, NY: The Free Press. Burns, T., & Stalker, G. M. (1961). The management of innovation. London: Tavistock. Chase, M. P. (1988). Improvisation: Music from the inside out. Berkeley, CA: Creative Arts. Christensen, C. M. (1997). The innovator’s dilemma. Boston, MA: Harvard Business School Press. Claxton, G. (1984). Live and learn: An introduction to the psychology of growth and change in everyday life. London: Hampton and Row Publishers. Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35, 128–152. Crossan, M. M. (1998). Improvisation in action. Organization Science, 9(5), 593–599. Crossan, M. M., Cunha, M., Vera, D., & Cunha, J. (2005). Time and organizational improvisation. Academy of Management Review, 30(1), 129–145. Crossan, M. M., Lane, H. W., & White, R. E. (1999). An organizational learning framework: From intuition to institution. Academy of Management Review, 24(3), 522–537. Crossan, M. M., Lane, H. W., White, R. E., & Klus, L. (1996). The improvising organization: Where planning meets opportunity. Organizational Dynamics, 24(4), 20–35. Crossan, M. M., & Sorrenti, M. (1997). Making sense of improvisation. Advances in Strategic Management, 14, 155–180. Damanpour, F. (1991). Organizational innovation: A meta-analysis of effects of determinants and moderators. Academy of Management Journal, 34, 555–590. Emery, F. E., & Trist, E. L. (1973). Towards a social ecology. New York: Plenum Publishing. Foster, R. N., & Kaplan, S. (2001). Creative destruction. New York: Currency.
296
MARY M. CROSSAN AND DAVID K. HURST
Garud, R., & Karnoe, P. (2001). Path dependence and creation. Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Ghemawat, P., & Costa, J. E. R. I. (1993). The organizational tension between static and dynamic efficiency. Strategic Management Journal, 14, 59–72. Greiner, L. E. (1972). Evolution and revolution as organizations grow. Harvard Business Review, 50(4), 37–46. Greve, H. R., & Rao, H. (2006). If it doesn’t kill you: Learning from ecological competition. In: J. A. C. Baum, S. D. Dobrev, & A. van Witteloostuijn (Eds). Ecology and strategy: Advances in strategic management (p. 23). Oxford, UK: Elsevier. Gunderson, L. H., & Holling, C. S. (Eds). (2002). Panarchy: Understanding transformations in human and natural systems London: Island Press. Hannan, M. T., & Freeman, J. (1977). The population ecology of organizations. American Journal of Sociology, 82(5), 929–964. Hannan, M. T., & Freeman, J. (1984). Structural inertia and organizational change. American Sociological Review, 49(2), 149–164. Hannan, M. T., & Freeman, J. (1987). The ecology of organizational foundings: American labour unions, 1836–1985. American Journal of Sociology, 92, 910–953. Hatch, M. J. (1998a). Exploring the empty spaces of organizing: How improvisational jazz helps redescribe organizational structure. Organization Studies, 20(1), 75–100. Hatch, M. J. (1998b). Jazz as a metaphor for organizing in the 21st century. Organization Science, 9(5), 556–557. Holling, C. S. (1986). The resilience of terrestrial ecosystems: Local surprise and global change. In: W. C. Clark, & R. E. Munn (Eds), Sustainable development of the biosphere (pp. 292– 317). Cambridge, MA: Cambridge University Press. Huff, J. O., Huff, A. S., & Thomas, H. (1992). Strategic renewal and the interaction of cumulative stress and inertia. Strategic Management Journal, 13, 55–75. Hurst, D. K. (1995). Crisis and renewal: Meeting the challenge of organizational change. Boston, MA: Harvard Business School Press. Hurst, D. K., & Zimmerman, B. J. (1994). From life cycle to ecocycle. Journal of Management Inquiry, 3(4), 339–354. Jeffery, D. (1989). Yellowstone: The fires of 1988. National Geographic, 175(2), 255–273. Kauffman, S. (1995). At home in the universe. New York: Oxford University Press. Kimberly, J. R., & Miles, R. H. (1980). The organizational life cycle. San Francisco: Jossey-Bass. Lant, T. K., & Mezias, S. J. (1992). An organizational learning model of convergence and reorientation. Organization Science, 3, 47–71. Lawrence, P., & Lorsch, J. (1967). Organization and environment. Boston, MA: Division of Research, Harvard Business School. Levinthal, D. A., & March, J. G. (1993). The myopia of learning. Strategic Management Journal, 14, 95–112. Levinthal, D. A., & Rerup, C. (2005). Crossing an apparent chasm: Bridging mindful and lessmindful perspectives on organizational learning. Unpublished paper. Lorenz, E. (1963). Deterministic nonperiodic flow. Journal of Atmospheric Sciences, 20, 130–141. MacArthur, R. H., & Wilson, E. O. (1967). The theory of Island biogeography. Princeton, NJ: Princeton University Press. Mandelbrot, B. (1983). The fractal geometry of nature. New York: W. H. Freeman.
Strategic Renewal as Improvisation
297
March, J. G. (1991). Exploration and exploitation in organization learning. Organization Science, 2, 71–87. March, J. G. (1994). The evolution of evolution. In: J. A. C. Baum, & J. V. Singh (Eds), Evolutionary dynamics of organizations (pp. 39–49). New York, NY: Oxford University Press. March, J. G. (1999). The future, disposable organizations, and the rigidities of imagination. In: J. March’s (Ed.), The pursuit of organizational intelligence (pp. 179–192). Malden, MA: Blackwell Publisher’s Inc. March, J. G., & Olsen, J. P. (Eds). (1976). Ambiguity and choice in organizations. Bergen, Norway: Universitetsforlaget. Mezias, S. J., & Glynn, M. A. (1993). The three faces of corporate renewal: Institution, revolution, and evolution. Strategic Management Journal, 14, 77–101. Miller, D., & Friesen, P. H. (1980). Momentum and revolution in organizational adaptation. Academy of Management Journal, 23, 591–614. Miller, D., & Friesen, P. H. (1982). Successful and unsuccessful phases of the corporate life cycle. Organization Studies, 4(4), 339–356. Mintzberg, H. (1990). The design school: Reconsidering the basic premises of strategic management. Strategic Management Journal, 11, 171–195. Mintzberg, H. (1994). The rise and fall of strategic planning. New York: The Free Press. Mintzberg, H., Ahlstrand, B., & Lampel, J. (1998). Strategy safari: A guided tour through the wilds of strategic management. New York: The Free Press. Moorman, C., & Miner, A. (1997). The impact of organizational memory on new product performance and creativity. Journal of Marketing Research, 34(1), 91–106. Moorman, C., & Miner, A. (1998a). Organizational improvisation and organizational memory. Academy of Management Review, 23(4), 698–723. Moorman, C., & Miner, A. (1998b). The convergence of planning and execution: Improvisation in new product development. Journal of Marketing, 62(3), 1–20. Nickerson, J. A., & Zenger, T. R. (2002). Being efficiently fickle: A dynamic theory of organizational choice. Organization Science, 13(5), 547–566. Organization Science. (1999). Special issue: Applications of complexity theory to organization science, 10(3), p. 215. Orlikowski, W. J., & Hofman, J. D. (1997). An improvisational model for change management: The case of groupware technologies. Sloan Management Review, 38(2), 11–21. Perry, L. T. (1991). Strategic improvising: How to formulate and implement competitive strategies in concert. Organizational Dynamics, 19(4), 51–64. Perry, L. T. (1994). Real time strategy: Improvising team based planning for a fast changing world. Organizational Dynamics, 22(3), 76–77. Pfeffer, J. (1982). Organizations and organization theory. Boston: Pitman Publishing Inc. Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations: A resource dependence perspective. New York: Harper & Row. Poole, M. S., & Van De Ven, A. H. (1989). Using paradox to build management and organization theories. Academy of Management Review, 14(4), 562–578. Prigogine, I., & Stengers, I. (1984). Order out of chaos. Toronto: Bantam Books. Rowley, T., Behrens, D., & Krackhardt, D. (2000). Redundant governance structures: An analysis of structural and relational embeddedness in the steel and semiconductor industries. Strategic Management Journal, 21(3), 369.
298
MARY M. CROSSAN AND DAVID K. HURST
Siggelkow, N., & Levinthal, D. A. (2005). Escaping real (non-benign) competency traps: Linking the dynamics of organizational structure to the dynamics of search. Strategic Organization, 3(1), 85–115. Singh, J. V. (2006). Ecology, strategy and organizational change. In: J. A. C. Baum, S. D. Dobrev & A. van Witteloostuijn (Eds), Ecology and strategy: Advances in strategic management (Vol. 23, pp. 177–214). Oxford, UK: Elsevier. Sobek II, D. K., Ward, A. C., & Liker, J. K. (1999). Toyota’s principles of set-based concurrent engineering. Sloan Management Review, 40(2), 67–82. Solomon, L. (1986). Improvisation II. Perspectives in new music, 24, 224–235. Sorensen, J. B., & Stuart, T. E. (2000). Aging, obsolescence and organizational innovation. Administrative Science Quarterly, 45(1), 81–113. Stacey, R. (1991). The chaos frontier: Creative strategic control for business. Oxford, UK: Butterworth Heinemann. Stacey, R. (1992). Managing chaos: Dynamic business strategies in an unpredictable world. London, UK: Kogan Page. Tichy, N. M., & Fombrum, C. (1979). Network analysis in organizational settings. Human Relations, 32, 923–965. Tretheway, B., & McDougal, S. (1998). Innovate or evaporate. Association Management, 50(12), 65–69. Tsoukas, H. (1991). The missing link: A transformational view of metaphors in organizational science. Academy of Management Review, 16(3), 566–585. Tulku, T. (1980). Dimensions of thought – current explorations in time, space, and knowledge, 2. Berkeley, CA: Dharma. Tushman, M. L., & O’Reilly, C. A. (1997). Winning through innovation: A practical guide to leading organizational change and renewal. Boston, MA: Harvard Business School Press. Ulanowicz, R. E. (1997). Ecology, the ascendent perspective. New York, NY: Columbia University Press. Utterback, J. M. (1994). Mastering the dynamics of innovation. Cambridge, MA: Harvard Business School Press. Vera, D., & Crossan, M. M. (2005). Improvisation and innovative performance in teams. Organization Science, 16(3), 203–224. Waldrop, M. M. (1992). Complexity. New York: Simon & Schuster. Weick, K. E. (1979). The social psychology of organizing (2nd ed.). Reading, MA: AddisonWesley. Weick, K. E. (1993a). The collapse of sensemaking in organizations: The Mann Gulch disaster. Administrative Science Quarterly, 38(4), 628–652. Weick, K. E. (1993b). Organizational redesign as improvisation. In: G. P. Huber, & W. H. Glick (Eds), Organizational change and redesign (pp. 346–379). Oxford, UK: Oxford University Press. Weick, K. E. (1998). Improvisation as a mindset for organizational analysis. Organization Science, 9(5), 543–555. Womack, J., & Jones, D. (1996). Lean thinking. New York: Simon & Schuster.
PART V: TECHNOLOGY STRATEGY
299
300
TECHNOLOGY CHOICE, TRANSACTION ALIGNMENT, AND SURVIVAL: THE IMPACT OF SUB-POPULATION ORGANIZATIONAL STRUCTURE Lyda S. Bigelow ABSTRACT A recent stream of research in strategy has demonstrated the effect of boundary of the firm decisions on firm performance by integrating concepts and methods from organizational ecology with predictions from transaction cost economics (e.g. Silverman et al., 1997; Bigelow, 1999; Nickerson & Silverman, 2003; Argyres & Bigelow, 2005). This work has confirmed that managing organizational boundary choices (or governance structures) efficiently has ramifications for firms’ survival chances. But further questions delineating the conditions under which governance structure alignment has a greater or lesser effect on firm survival remain. In this paper, we consider how selection pressures may differ according to a firm’s adoption of either a mature or an evolving technology. Using ecological insights regarding competitive intensity and sub-population density, we test for the evidence of the role of sub-population organizational (governance) structure within a technology class. We present Ecology and Strategy Advances in Strategic Management, Volume 23, 301–333 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23010-9
301
302
LYDA S. BIGELOW
preliminary results using an 18-year panel of the population of U.S. automobile manufacturers from 1916 to 1934. The primary preliminary findings: Within a population, individual misalignment diminishes survival. However, the aggregate governance structure of firms within a technology sub-population has a greater effect on the survival of a focal firm than the governance choice of the individual firm. These findings suggest that governance choices in aggregate within technologically localized sub-populations may influence firm survival. Further, this paper adds to a body of work that utilizes ecological concepts to extend organizational theory.
1. INTRODUCTION Understanding the trade-offs involved in relying on internal or external sourcing of components and/or services has been a central focus of the transaction cost economics research program. And indeed, many of the hazards associated with governing highly asset-specific transactions by relying on external markets rather than vertical integration have been theoretically elaborated (e.g. Williamson, 1991, 1996). Recent reviews of the empirical transaction cost literature list hundreds of studies that support the central prediction that highly asset-specific transactions are more likely to be vertically integrated than outsourced (Klein & Shelanski, 1995; Boerner & Macher, 2004). However, both these reviews conclude that there is a shortage of studies that connect this governance decision with firm performance. Of the relatively small number of studies which investigate the relationship between transaction alignment and firm performance, most rely on survey measures of outcomes and a few look at comparative costs (e.g. Masten, Meehan, & Snyder, 1991; Walker & Poppo, 1991). Very few have analyzed the link between alignment and financial performance; though early studies tested whether the M-form of organization was more profitable for diversified firms (e.g. Armour & Teece, 1978). In an effort to expand the research on governance effects and firm performance, a few studies (e.g. Silverman, Nickerson, & Freeman, 1997; Nickerson & Silverman, 2003; Argyres & Bigelow, 2005) have successfully combined organizational ecology and transaction cost theories, utilizing firm survival as a proxy for performance. As Barnett and Carroll (1995) and Carroll and Hannan (2000) have suggested, combining theoretically derived measures of organizational features in hazard rate models that estimate mortality risks for an entire population
Technology Choice, Transaction Alignment, and Survival
303
of firms, is one way of demonstrating a theory’s usefulness at making these kind of predictions. This paper incorporates that logic by testing the effect of misalignment on mortality. We extend this growing area of research by adding the effect of aggregate governance structure decisions within a technological segment of the total population. The idea here is to demonstrate that competitive intensity varies across time and market segment. Thus, the impact of theoretically optimal organizational structure from a transaction cost perspective may vary, depending on the evolution of the industry and the competitive pressure exerted by local sub-population density. This is a concept that is familiar to organizational ecology researchers, but less theoretically and empirically developed in the transaction cost literature. By incorporating these basic ecological insights, we may extend the alignment – performance research in transaction cost economics. We may also contribute to ecological research by adding to the research investigating sub-population density effects and local competitive intensity. The study will also be one of the first to empirically test ideas in Barnett and Carroll (1995) and Carroll and Hannan (2000) regarding the benefit of combining theoretically derived predictions of optimal structure in ecologically informed models of survival. Also, the results here reflect the opportunity to generate additional insights in competitive processes by combining economic efficiency concerns with ecological assertions that there are other selection mechanisms besides efficiency that may lead to the failure of firms (e.g. Van Witteloostuijn, 1998; Carroll & Hannan, 2000). The paper is organized as follows: in the following section, we describe the imperative of demonstrating the performance implications of efficient governance, Section 3 describes the theoretical and empirical constraints, Section 4 describes the role of industry evolution and technologically local competitive intensity, Section 5 describes data and methods, and we close with a discussion of results.
2. PERFORMANCE IMPLICATIONS OF TRANSACTION COST ECONOMIZING Transaction cost researchers have established a solid empirical and theoretical base for the link between transaction features and governance structure (for a review see Shelanski & Klein, 1995). Numerous studies have confirmed the central predictions of the theory, namely that key theoretically derived variables describing transactional attributes do correlate with postulated governance choices in a transaction cost economizing way. Demonstrating the link between efficient mode of governance and
304
LYDA S. BIGELOW
transaction features is only half the battle. A complete test of the theory is to document a link between this alignment and organizational performance. We use organizational survival as a performance metric and then seek to test for a link between transaction alignment and the likelihood of failure. Unlike research on transaction alignment, the work done on the performance implications of transaction cost economizing is relatively underdeveloped. Both the elaboration of the theoretical logic and the conduct of empirical tests are required. Without being able to make the theoretical and empirical link to performance implications, the relevance of transaction cost economizing to the broader strategy field remains limited. Indeed, there are only two published papers to date (Silverman et al., 1997; Nickerson & Silverman, 2003) that model the relationship between transaction cost economizing and mortality. There is recent interest, however, in moving beyond first-order organizational structure and focusing on the performance implications of transaction alignment. In a study of R&D alliances, Sampson (2004) found that the alignment of transactions according to TCE predictions conferred collaborative benefits not found in transactions organized otherwise and increased patenting activity generated through these alliances. Poppo and Zenger (1998) found overall satisfaction with costs, quality, and responsiveness increased with the alignment of governance. Nickerson, Hamilton, and Wada (2001) combined concepts from strategic positioning with transaction cost considerations and found that the efficient alignment of governance structure with resources and capabilities increased delivery time performance in a study of firms in the international courier and small package services industry in Japan. Most recently Mayer and Nickerson (2005), in a first of its kind study, found that if a large information technology services firm outsourced or internalized a transaction in accordance with transaction cost predictions, then the profitability of that exchange is increased. But the problem remains that even this recent work on performance is cross-sectional and focuses on transaction rather than firm-level performance. Little work examines the long-term impact on the firm of organizing the structure of the firm in accordance with transaction cost economizing in mind. Drawing on concepts and methods from organizational ecology, a theory that is devoted to understanding the dynamics of organizational vital rates and the organizational, population, and sub-population factors that can affect these vital rates could provide useful insights into longitudinal assessments of transaction cost (or other organizational theories) predictions of performance. Clearly, Williamson has discussed these issues (Williamson, 1999). Williamson makes reference to the fact that weak-form selection environments
Technology Choice, Transaction Alignment, and Survival
305
are a reasonable assumption as a test for transaction cost economizing (Williamson, 1985, 1996). He also claims that, eventually, inefficiency invites its own demise (Williamson, 1996, p. 234). This assumption stems from the idea, articulated most prominently by Alchian (1950) and Friedman (1953), that market pressures act to ‘‘select out’’ inefficient firms in a manner similar to that of natural selection. Williamson (1985) writes: Natural selection forces do not always operate quickly y Firms that are buffered against product market rivalry y and against capital market discipline y can postpone the reckoning. But these would appear to be the exception rather than the rule. Where incumbent managements are not pressed to adopt the new procedures by economic events, successor managements, often in conjunction with the appointment of a new chief executive, commonly will (pp. 129–130).
Transaction cost theory thus presumes that in general, a firm cannot rely on market power to make up for its inefficiencies. Hence, economizing, rather than strategizing, becomes transaction cost theory’s prescription for superior financial performance. This presumption in transaction cost theory that market forces tend to select firms that economize on transaction costs and achieve efficient alignment is made without regard to the industry conditions. The fundamental issue of whether or not inefficiency does or does not actually invite its own demise is not a question often raised in the economics literature, but has been raised by organizational researchers (e.g. Van Witteloostuijn, 1998). Organizational ecology theory has in fact been explicit in its assumption that there may be many reasons for firms to exit an industry, with inefficiency being just one of many reasons for firm failure. One solution to this potential inconsistency is to rely on the specifics of the empirical setting. We devote greater attention to this in the next section. The call to incorporate theoretically derived measures of organizational structure in ecological estimations of mortality is first raised by Barnett and Carroll (1995). Although the focus of their paper is reviewing and extending research on organizational change, the basic premise is easily applicable to tests of alignment such as that presented here. In fact, they emphasize that methodologically nothing precludes the incorporation of variables coding for alignment based on other organizational theories, e.g. the resource-based view of the firm or contingency theory. Again, although phrasing this in the context of parsing out content vs. process effects in the study of organizational change, the authors describe the lack of empirical tests which follow this model as a ‘‘missed opportunity’’ (Barnett & Carroll, 1995, pp. 228–229) and this conjecture on the role of ecology theory in strategy research is echoed five years later (Carroll & Hannan, 2000).
306
LYDA S. BIGELOW
Why have there been so few tests to date? There are both empirical and theoretical issues to contend with. We discuss some of these challenges in the following section.
3. THEORETICAL CONCERNS: LEVEL OF ANALYSIS, TRANSACTION INTERDEPENDENCE, TRANSACTION HISTORY Several theoretical concerns raise challenges to researchers pursuing projects that explore the link between performance and transaction cost economizing. These include level of analysis concerns, transaction interdependence, and transaction history. The level of analysis issue is perhaps the most significant (e.g., Masten, 1993), though it is not a theoretical concern limited to transaction cost economics. Indeed, it is a problem that plagues other similar types of organizational research (e.g. Carroll, 1997). 3.1. Level of Analysis At the heart of the level of analysis concern is the issue of whether or not the organization of a single transaction affects overall firm performance and, if it does, is the effect large enough to be measured econometrically. If we are interested in pursuing how the effects of misalignment affect such performance measures, then we need to construct a test in which we can build a plausible argument that inefficiencies and/or contracting hazards derived from that one transaction will have a firm-level impact. Since firm-level outcomes often are the result of complex interactions among transactions, it may be difficult to delineate how the arrangement of a given transaction in isolation affects total firm performance. Finding an appropriate test (e.g. determining transaction centrality) as well as collecting data has been a challenge to date. One solution to this problem is to choose a test of a transaction that can be considered a core or central activity. We treat this as an empirical issue and resolve it through the choice of industry and transaction, in our case the sourcing of engines in the early U.S. automobile manufacturing industry. We argue that the sourcing of a critical component such as an engine constitutes a core transaction. Engines can be defined as a key component based on several reasons: the magnitude of production costs they represent, their interdependence with other components, and their effect on the market
Technology Choice, Transaction Alignment, and Survival
307
positioning of the final product.1 According to economic historians who have studied the automobile industry (e.g. Hounshell, 1984; Raff, 1991; Abernathy, 1978), engine production could account for anywhere from 10% to 30% of total production costs during the period of observation used herein. This figure also depends on whether or not the firm in turn produces its own related drivetrain components and the nature of the engine technology. We simply argue here that any component that contributes to a significant share of total production costs may be considered a core transaction. In other words, if we can identify an effect given the problem then we should at least identify it here. Moreover, the type of engine used, how it is configured, the type of material used, and the technical specifications directly affect the product positioning of the firm, which further indicates the importance or criticality to firm-level performance of this transaction. Related to this issue of product positioning, in the early years of the automobile manufacturing industry, firms often advertised the sourcing of components – particularly engines – in an effort to pursue differentiated strategies based on quality, reliability, or craftsmanship. The relationship of product positioning and sourcing depends on the individual firm. Packard, for example, advertised the self-manufacture of components in an effort to bolster their differentiation strategy based on quality and craftsmanship. Ford, the most famous case of nearly complete self-manufacture of all components used this internal production, in part, to underscore a reputation for product reliability. Conversely, in an effort to assure quality and reliability, particularly in the predominant design period, many firms advertised the identity of their engine manufacturer (e.g. Lycoming was a prominent supplier often mentioned in advertisements). Thus, engine choice also is a core strategic choice further amplifying this transaction as critical to firm survival. And if we are going to observe an effect for the governance choice, then a core activity such as engine procurement is a good empirical context. 3.2. Transaction Interdependence An additional research challenge in this area is that there may be a transaction bundling or interdependence problem (Nickerson, 1997; Argyres & Liebeskind, 1999). Here the concern is that if transactions are interdependent, then there may be constraints placed upon the feasible choice set for governance alternatives. Thus, a transaction viewed without regard to its dependence on other transactions may appear to be misaligned, yet if the bundle of transactions is assessed its degree of alignment may be efficient.
308
LYDA S. BIGELOW
Related to this, Argyres and Liebeskind (1999) elaborate the path-dependent nature of governance. Similar to the bundling problem, the concern of examining transactions without regard to context may be problematic. They elaborate Williamson’s (1985, 1996) argument that history does matter, in that, what constitutes a feasible, implementable governance alternative may be constrained by the arrangement of previous transactions. However, in contrast to Williamson’s position, they argue that history plays a more substantive role in establishing the feasible set of governance alternatives, what Williamson (1994, 1996) describes as remediableness.2 As with the level of analysis concern, we rely on the particular features of our empirical setting to mitigate these potential theoretical problems. For the issue of transaction interdependence, we rely on information collected on the make-or-buy status of related transactions, namely the procurement of clutch and transmission components, which are the components that are most technically interdependent with the engine. We can use these additional covariates to test empirically whether or not the organizational form of the focal transaction is influenced by the form used to organize related transactions.3 Thus, the misalignment measure we employ is constructed by controlling for the organization of related transactions. 3.3. Transaction History Another concern stems from the effect of experience on transaction organization. There is a recent stream of research that suggests that firms may develop competencies in a particular mode of organization through experience. Research on alliances vs. acquisitions indicates that such competencies may ultimately drive down the cost of organization, favoring the mode with which the firm has the most experience (e.g. Gulati, 1998; Anand & Khanna, 2000; Villalonga & McGahan, 2005). Thus, the same experientially derived competencies may be at work for firms confronting make-or-buy decisions. Further, a firms’ transaction history may constrain its ability to arrange future contracts (Argyres & Liebeskind, 1999) once more altering the comparable costs of organization in a firm-specific manner based on past decisions. The analysis presented here employs information on previous make-orbuy decisions. This data can be used in conjunction with age, to control for experiential effects with a given form of organization. We also have information on the firm’s status as either a lateral entrant from another industry (de alio) or a start-up within the U.S. automobile manufacturing industry (de novo). This variable may affect the likelihood of adopting one form of
Technology Choice, Transaction Alignment, and Survival
309
organization over another if, for example, a de alio firm is endowed with a history of adopting a particular organizational form in the previous industry setting and this experience effect creates a bias that a similar de novo firm does not possess. Thus, de alio status may also affect a firm’s success with using a given organizational form in a new industry setting. The empirical demands in assessing the performance implications of transaction cost economics are onerous given that microanalytic transaction-level data as well as longitudinal firm-level data are needed. Particularly, if we are testing hazard rate models from an organizational ecology perspective, in order to avoid potential selection or survivor bias, we need comprehensive data on the entire population at risk of failure. This is a formidable empirical challenge. Fortunately, the minimum data requirements have been met in this particular analysis. With meeting the empirical demands and addressing major theoretical challenges, the primary hypothesis is fairly straightforward. An underlying premise of transaction cost reasoning is that, except in cases of societies with extraordinary institutional constraints, the existence of a selection mechanism is based on efficiency. As long as weak-form efficiency selection mechanisms operate, inefficient alignments cannot be sustained. Either they will be selected out, or they will be replaced. While the strength of the selection environment may vary with individual populations of organizations, or may vary within populations over time, ultimately firms that fail at getting the alignment right will suffer some penalty vis-a-vis firms that are better aligned. As Van Witteloostuijn (1998) and others have argued, however, there may be instances in which inefficiently organized firms may remain in the industry while more efficient firms will exit, turning our main prediction on its head. It is with this concern in mind, that we take the position in this paper that a better understanding of both the state of industry evolution and the degree of rival firms’ organizational alignment will help clarify predictions. In the empirical analysis that follows, alignment is achieved if a firm produces a highly asset-specific component or conversely, contracts out for a component of low assetspecificity. The penalties may vary depending on the nature of the misalignment. For example, a firm which produces inhouse a generalized component will likely face an efficiency disadvantage compared to rivals who ‘‘correctly’’ rely on the market for this transaction. Alternately, a firm which contracts out for a specialized component is likely to face various contracting hazards that increase its risk of failure compared to firms which take this highly specific transaction in-house. Although lags may delay the impact, transaction cost economics predicts that eventually
310
LYDA S. BIGELOW
misaligned firms will either be selected out of the population or they will undergo some sort of adaptive re-alignment.4 Thus, we hypothesize H1. Misalignment at any given point in time will have a negative effect on a firm’s subsequent survival chances.
4. INDUSTRY EVOLUTION AND COMPETITIVE INTENSITY An important question that remains, however, regards the role of competitive pressure and what conditions might intensify or buffer a firm from misalignment-driven failure. Beyond stipulating baseline assumptions of competitive conditions, the theory, as formulated by Williamson (1985, 1991, 1996), is relatively silent regarding the impact of market forces that promote efficiency. Ecological and evolutionary theories of economic organization, on the other hand, have paid explicit attention to selection processes and their implications for survival and efficiency, respectively. Evolutionary theories of the industry life cycle are of particular interest in this context, because they postulate general patterns in the waxing and waning of efficiency-driven selection forces that can be, and have been, made subject to empirical confirmation. Organizational ecology has refined the concept of density-driven selection pressures through work, which addresses the impact of sub-populations within industries. Studies of processes ranging from resource partitioning (e.g. Carroll, 1985), size-localized competition (e.g. Baum & Mezias, 1992), geographic proximity (e.g. Carroll & Wade, 1991; Bigelow, Carroll, Seidel, & Tsai, 1997), and market partitioning (e.g. Peli & Nooteboom, 1999) – to name just a few – demonstrate the potential of finer-grained distinctions among populations for explaining firm and population dynamics. Thus, by combining insights from industry life cycle theories and organizational ecology we can test for the effects of both firm and technologically related sup-population governance structures on firm survival. The theory of the ‘‘product life cycle’’ originally presented by Abernathy and Utterback (1978) stipulates that in the early stage of an industry’s evolution firms compete based on alternative ‘‘product designs’’ or overall ‘‘product conceptions’’ that differ in significant ways, such as in their performance along various dimensions. Entry is rapid and there is widespread experimentation with these alternate designs. Over time the rate of product innovation declines, while the rate of process innovation increases. This transition in the nature of innovation is often associated with the
Technology Choice, Transaction Alignment, and Survival
311
emergence of a ‘‘dominant design’’ (e.g. Christensen, Suarez, & Utterback, 1998). A design is said to become dominant when these concepts become widely shared. The emergence of such a design ushers in a new stage in the industry life cycle. Rather than competing primarily on the basis of alternative product designs, firms converge on the dominant design, and compete primarily on cost. With the emergence of a dominant design, organizational mortality increases, as larger more efficient firms drive out smaller less efficient competitors, causing a net decline in the number of firms in the industry. Over the past two decades, Steven Klepper and colleagues (e.g. Klepper, 1996) have developed a related but distinctive set of models and empirical studies of the industry life cycle. The stages of industry evolution in these models are similar to those in the dominant design theories; however, the key mechanisms driving the transitions between the stages are different. Generally, a ‘‘shake out’’ occurs in which firms with greater innovation expertise in the ultimate standard configuration of a product survive. This literature on the industry life cycle – both that associated with Utterback and colleagues, and that associated with Klepper and colleagues – thus demonstrates that selection forces become much stronger as technological innovation declines. Organizational ecology also emphasizes this phenomenon of industry maturation and concomitant increases in mortality hazards (e.g. Carroll & Hannan, 1995; Winter, Kaniovski, & Dosi, 2003). Whether the impetus for this change in the selection environment comes from product standardization that is driven purely by technological imperatives, or in part by customer tastes, market pressures eventually cause a shake out. These strong selection forces will cause firms that have higher total costs to exit the industry or adapt. We use this logic to examine two distinct technology trajectories that exist within an industry and consider the likely differences in competitive pressures for efficiency that are generated within that industry. A dynamic embedded within the industry life cycle perspectives discussed above is that those firms that utilize a technology that is mature or post-dominant design will now face greater efficiency pressure, while those that utilize a newer technology will enjoy diminished efficiency pressure, but greater pressure to build on or leverage innovation. Ecological work on competitive intensity (e.g. Barnett, 1997) distinguishes between organizations that may have good life chances based on theoretically derived advantages (e.g. size), yet not exert much competitive pressure on other organizations. This suggests that different organizations generate different levels of competitive pressure and that these organizations can be categorized according to some similar
312
LYDA S. BIGELOW
characteristic and their effect on other organizations’ mortality hazard is a measure of their competitive intensity. Using these arguments it is not nconsistent to expect that firms which economize on transaction cost (e.g. are more efficient), will exert competitive intensity under conditions in which efficiency-driven pressures are paramount. Additionally, firms that are inefficiently aligned from a transaction cost perspective will face greater risk of mortality if they utilize a mature technology. The competitive intensity of other organizations that compete within the same technology class sub-population can be estimated by compiling a measure of aggregate efficient alignment. Firms that operate in a mature technology, then, will face greater risk of failure if they face more efficiently organized competitors. On the other hand, firms that operate in a still evolving technology will not be subject to this same sub-population competitive dynamic. Thus, we offer the following hypotheses: H2. Aggregate misalignment within a technology class sub-population will have a negative effect on organizational mortality if the focal technology is mature. H3. Aggregate misalignment within a technology class sub-population will have no effect on organizational mortality if the focal technology is still evolving. In order to test hypotheses H2 and H3, we must identify technologies that differ in their maturity within the U.S. auto industry. We select a comparison of four-cylinder engine technology (mature) and six-cylinder engine technology (still-evolving). The four-cylinder engine is, of course, the engine technology adopted by Ford’s Model T introduced in 1908. Technological innovations declined as process innovations increased, catalyzed by Ford’s introduction of mass production in 1913 (e.g. Rae, 1959, 1984). While six-cylinder engines were utilized prior to 1916, most of the firms in this category competed on quality and technology (e.g. Packard, Pierce-Arrow). Experimentation with cylinder configurations continued well into the 1920s. Table 1 lists the density and proportion of new models for each engine technology by year. Note that six-cylinder models dominate, highlighting the waning of the four-cylinder technology, with six-cylinder models peaking in 1922, while four-cylinder models peak in the prior year. By 1926, the number of four-cylinder models declines 84% from this peak, while six-cylinder models decline by only 58% from their peak. Note as well, despite the overall trend in declining density and entry in the industry, new six-cylinder models continue to enter for much of the observation period, while entry of models
313
Technology Choice, Transaction Alignment, and Survival
Table 1.
Counts and Proportion of New Models by Technology Segment.
Year
1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933
4-Cylinder
6-Cylinder
Density
New Model
Density
New Model
43 31 31 51 44 32 15 12 8 5 5 4 3 3 2 5
0.93 0.16 0.13 0.25 0.20 0.12 0.13 0 0 0 0 0 0 0 0 0.20
57 51 55 69 73 61 51 32 30 28 25 19 17 14 14 13
0.96 0.10 0.09 0.13 0.16 0.08 0.02 0 0 0 0.04 0 0.06 0 0.14 0
using the mature four-cylinder technology virtually halts in 1924, perhaps not uncoincidentally the year identified by automotive historians as the emergence of a dominant design for the automobile (e.g. Abernathy, 1978). Table 2 provides data on the proportion of misaligned firms broken out by technology. Note that though the proportion of misaligned firms in the mature four-cylinder technology starts higher than the six-cylinder technology, this misalignment gradually declines until 1926, at which point virtually no firms remain misaligned. Again, this is consistent with theorizing about industry evolution; we simply adopt this theory to explain individual technology trajectories within the industry. In contrast, the proportion of misalignment among six-cylinder producers increases over the first six years of observation and then varies over time. This seems consistent with a technology that is continuing to evolve.
5. DATA AND METHODS We use a comprehensive, longitudinal, detailed database on automobile manufacturers and automobile parts suppliers.5 As explained in the
314
LYDA S. BIGELOW
Table 2.
Proportion Misaligned Firms by Technology Segment.
Year
4-Cylinder
6-Cylinder
1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933
0.256 0.226 0.226 0.216 0.205 0.219 0.133 0.083 0 0 0 0 0 0.333 0 0
0.123 0.118 0.145 0.130 0.178 0.279 0.196 0.219 0.233 0.143 0.36 0.368 0.176 0.214 0 0
preceding section, undertaking event-history analysis to test questions empirically having to do with boundary of the firm decisions can present onerous, even impossible data requirements. Fortunately, in this case, many of these data challenges were overcome because of complete demographic information (e.g. start and end dates, size, location, etc.) on the population at risk as well as component-level information for most of the firms. However, this data covers a limited period of time. Data on the automobile manufacturers was collected primarily from the following sources: The New Encyclopedia of Motorcars (Georgano, 1982), The World Guide to Automobile Manufacturers (Baldwin, Georgano, Sedgwick, & Laban, 1987), and the three volumes of The Standard Catalog of American Cars (Kimes & Clark, 1989; Flammang, 1989; Gunnell, 1987). Data on the make-or-buy decision as well as on the suppliers was collected primarily from the following sources: Motor Age, The Automotive Trade Journal, and Chilton’s Automobile Suppliers Directory (various years), and the volume: Automobile Specifications 1915–1945 (Lester-Steele, 1960). The period of observation is the years between 1916 and 1934 because this represents a period of time in which the industry experienced high levels of both demand and technological uncertainty; high levels of entry, exit, and density; and, for empirical purposes, a high degree of heterogeneity in terms of how firms were organized. Although firms exit an industry for many reasons, only firms that truly fail, i.e. are liquidated or go through a
315
Technology Choice, Transaction Alignment, and Survival 250
density
entry
exit
Number of Firms
200
150
100
50
0 1916 1918 1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 1940 Year
Fig. 1.
Density, Entry, and Exit Automobile Manufacturers, 1916–1940.
bankruptcy, are included in the analysis. In the construction of the dataset, 11 ending events were coded, including events such as mergers and acquisitions. Only ending events that represent true failure, i.e. liquidation or bankruptcy, are included here. Figures 1 and 2 present an overview of industry and governance structure data. Fig. 1 plots density, entry, and exit of firms over the observation period. Fig. 2 shows the comparison of the firms which produced engines inhouse with the numbers of firms who relied on suppliers for that same component. The unit of analysis for these figures is either the model or the firm, depending on the figure. Figure 1 shows the number of firms in the auto industry over the period 1916–1940. The figure shows that the number of firms slowly declined from a peak established in 1907, remaining near that level until 1914. The decline in density reverses in 1919 and density continues to rise again until 1922, and then begins a long decline that continued until the post-WWII period. The rise in density and surge in entry between 1919 and 1922 was likely caused in part by the end of the WWI. This surge of entry was significant, leading to a 30% increase in the number of firms from 1919 to 1922. Figure 2 reveals the trend for procurement of the first engine.6 This figure combines the data for firms that are single model producers with
316
LYDA S. BIGELOW
120 buy
make
100
Number of Firms
80
60
40
20
0 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 Year
Fig. 2.
Engine Procurement, U.S. Automobile Model 1, 1917–1934.
multi-model producers, though only the first model of the multi-model firm is included. The first model for multi-model firms is simply a coding convention. It does not signify importance in terms of number of units sold.7 Note that until about 1920, most model engines are bought. After that time, the gap between the numbers of models that use self-manufactured engines with those that use engines from suppliers is relatively small, though it begins to widen after 1924, the time of the emergence of the dominant design. Although there is some variability in percentages, by the end of the observation period roughly 75% of these models have integrated engine production, the rest rely on suppliers. These trends of course need to be controlled for in the analysis, which we do by controlling for year effects in the Probit model used to construct the misalignment variable. Although a trend emerges in engine procurement by the end of the observation period, the same is not true for other major related components. For example, though not shown here, transmission procurement was evenly split (Bigelow & Argyres, 2006). Further, these graphs indicate that indeed a high degree of heterogeneity existed in the population regarding governance at least in the early period of observation, providing the opportunity to empirically test these questions of changes in organizational form.
Technology Choice, Transaction Alignment, and Survival
317
Hazard rate models have been widely used for analyzing firm exit (e.g. Tuma & Hannan, 1984; Allison, 1984; Yamaguchi, 1991; Petersen, 1995). It is widely agreed that such models are appropriate when time-varying covariates are used. And, more importantly, these models are appropriate for undertaking longitudinal analysis with left or right censoring (of which the latter is more easily remedied) and for handling information on subjects that do not experience the event in question. These models estimate the instantaneous mortality rate for a firm (i) at age (t) that is usually defined as follows: ri ðtxi ðtÞÞ ¼ lim Pðt Tot þ DtT t; xi ðtÞÞ Dt Dt!0
where r(t) is the hazard rate of organizational mortality; T a random variable for the amount of time the organization (i) is at risk; xi a set of timevarying covariates for firm (i) at time (t); and P(.) is the probability that the event occurs between time t and t+Dt, given that the event did not occur prior to time t. Consistent with recent approaches to control for the effects of age, these hazard rate models are run in Stata (v 8.2) using the stpiece subroutine developed by Sorensen (1999). 5.1. Dependent Variable The dependent variable we use, the instantaneous rate of failure, is a measure of firm survival. We consider a firm’s ability to survive a proxy for firm performance. Although not as fine grained as the accounting measures often used in assessing performance in the strategy literature, it is not subject to the same potential for noisiness (Carroll & Hannan, 2000) and deterioration over time (e.g. Meyer & Gupta, 1994). Again, because we limit the events that signal exit from the population to true failures, we consider this likelihood of failure a useful measure of firm performance.8 5.2. Covariates The misalignment variable is of central concern. In order to test empirically for the effects of misalignment on survival, a measure that can be updated over time is constructed for the entire population of firms. We construct the measure in two steps (see the appendix for details). First, we use a measure of asset specificity based on the cubic displacement and cylinder configurations of the engine. The greater the value, the more unique or asset specific the transaction. Second, we incorporate the results from a Probit analysis
318
LYDA S. BIGELOW
that estimates the likelihood of integration using this measure of asset specificity. Misalignment (Misalign) is operationalized as a comparison of the actual make-or-buy decision with the estimate of the likelihood of integrating (p) from this Probit analysis. In other words, since make ¼ 1 and buy ¼ 0, the higher (p), the more misaligned is a firm that purchases engines. Similarly, the lower (p), the more misaligned is a firm that produced its own engines. We include this variable in the hazard rate model presented below. This variable is used to test the primary hypothesis, H1, which predicts that the coefficient of this variable will be positive. We then construct a measure of the degree of misalignment within technology class sub-populations. First, we identify firms that produce four- or six-cylinder models. We then generate a count by model of each engine type. Finally, we calculate a simple percentage of those models that are misaligned within each technology. These variables, percent misaligned 4cyl and percent misaligned 6cyl are used to test H2 and H3. 5.3. Control Variables Consistent with recent research estimating mortality, we use a piece-wise constant rate model using categorizations of age following a 3-, 7-, and 10year specification. This specification is based, in part, on a brief discussion in Williamson (1985, p. 23) suggesting that under weak-form selection environments, exit due to inefficiency may take 5–10 years. Since we have two technology trajectories and a limited observation window, we feel that it is an appropriate age specification. We also include control variables that might be expected to impact firm survival other than transaction cost economizing. Carroll et al. (1996), using the same survival data we use but for a much longer period (1895–1981), found that entrants into the auto industry that had had previous experience in a related industry (de alio entrants) tended to enjoy better survival chances than entrants that lacked such experience (de novo entrants). Accordingly, we include a dummy variable, De alio, that takes on the value of 1 if the firm was a de alio entrant, and zero otherwise. Klepper (2002) also emphasizes the advantages of prior experience in related industries in creating greater innovation capability. Thus, this literature suggests that de alio status should be negatively associated with the hazard rate in our model, though perhaps for different reasons. We included also firm size (firm size) measured by sales (entered as actual size/1000, following Carroll et al., 1996) as an explanatory variable. On the one hand, larger size might be expected to increase survival chances if economies of scale are significant. Size may also be associated with superior
Technology Choice, Transaction Alignment, and Survival
319
capability, as in Klepper’s (1996) model. On the other, sociologists have argued that larger firms suffer from excessive bureaucracy and politicization, and should therefore display lower survival chances during periods of technological change (e.g. Hannan & Freeman, 1977). Carroll et al. (1996) found, however, that the effect of size on the hazard was consistently negative and significant in their auto industry study. Finally, we also include a dummy variable, left censored, for those firms that were alive prior to our observation period. Since we have complete histories of these firms and control for age, we do not expect the coefficient to be significant. All timevarying co-variates are updated on an annual basis. We do not include other typical ecological control variables such as density, density-squared, or density-at-founding, for two reasons. First, the objective of this study is to examine the impact of subpopulation density effects in the context of an assumption of efficiency-driven selection pressures. Clearly, ecological theory is agnostic regarding the determinants of increased competitive intensity in a population. Thus, if other drivers are behind the impact of density effects, we risk adding confounding selection processes by including these terms. Second, in separate models not reported here, adding these conventional density controls reduced the overall fit of the model. Interestingly, however, industry structure alone – as measured by density, density-squared, density at founding, and sub-population density had consistently significant effects consistent with conventional ecological reasoning.
6. RESULTS Tables 3–5 display correlations among key variables, including those variables that are used in the underlying calculation of the misalign variable (Table 3). Tables 4 and 5 provide correlations by engine technology. Though age and size are correlated more highly for the four-cylinder sub-population than the six-cylinder sub-population, none of these correlations are significant. Further, the difference in the correlations is consistent with the argument that the four-cylinder engine technology is a mature technology and as firms age and continue to use this technology they are likely adapting larger production runs in an effort to reap economies of scale. Table 6 provides results from the piece-wise constant rate mortality models. Given the importance of correctly specifying age dependence, this model allows mortality rates to vary freely across age groups. The rate is assumed to hold constant within whatever age group is defined. Recall that we use the age split of (0,3,7,10).9 Model 1 provides a baseline model using
320
LYDA S. BIGELOW
Correlations, All Firms.
Table 3. 1 1. 2. 3. 4. 5. 6. 7. 8.
Misalign Firm age Firm size De alio Number of models Engine make–buy Left censored GNP
2
0.1805 0.1993 0.0268 0.1291 0.1848 0.1578 0.0194
Table 4.
0.2183 0.1892 0.3224 0.3619 0.5605 0.4676
Misalign Firm age Firm size De alio Number of models
Table 5.
0.3770 0.2914 0.1273 0.1138
Misalign Firm age Firm size De alio Number of models
0.1812 0.4105 0.1838 0.1293 0.2102
0.1096 0.0662 0.1758 0.0296
5
6
7
0.1735 0.1423 0.3787
0.1387 0.1651
0.0937
2
0.4865 0.1663 0.4145
3
0.2391 0.3733
4
0.1682
Correlations, 6-Cylinder Segment. 1
1. 2. 3. 4. 5.
4
Correlations, 4-Cylinder Segment. 1
1. 2. 3. 4. 5.
3
0.1392 0.2152 0.0409 0.0461
2
0.1611 0.1715 0.1529
3
0.1712 0.1039
4
0.0197
data on all firms over the observation period. Models 2 and 4 replicate this baseline specification, but we now separate out by sub-population. Model 2 utilizes only firms that produce four-cylinder models and Model 4 utilizes only firms that produce six-cylinder models. Models 3 and 5 add in the respective percent misaligned variables. Our primary concern is to test to see if the effect of misalignment is evident, and if so, is it in the hypothesized positive direction and does it hold after adding the most important controls. As seen in the separate age-split variables in Model 1, there is a clear pattern of negative age dependence, with the oldest firms in our sample
321
Technology Choice, Transaction Alignment, and Survival
Table 6.
Piece-Wise Constant Rate Mortality Models of U.S. Automobile Manufacturers 1917–1934.
Ageo3 years Age 3–7 years Age 7–10 years Age410 years Misalign Firm size Left censored De alio
Model 1
Model 2
Model 3
Model 4
Model 5
1.471
1.051
1.343
1.489
2.556 (.509) 2.835 (.547) 2.321 (.713) 3.222 (.764) 0.606 (.658) 0.0001 (4.00E-5) 0.275 (.444) 0.158 (.284)
(.275) 1.581 (.296) 0.942 (.403) 1.691 (.409) 0.932 (.466) 0.0001 (4.00E-5) 0.381 (.284) 0.355 (.201)
(.484) 1.093 (.534) 0.546 (.704) 1.599 (.804) 0.586 (.832) 0.0002 (0.0001) 0.231 (.467) 0.676 (.335)
Percent misaligned 4cyl
(.825) 1.191 (.838) 1.706 (.952) 0.167 (.941) 0.406 (.903) 0.0003 (0.0001) 0.106 (.476) 0.923 (.351) 10.634 (2.752)
(.386) 1.623 (.397) 1.027 (.585) 1.644 (.564) 0.666 (.661) 0.0001 (4.00E-5) 0.633 (.425) 0.177 (.279)
6.231 (1.903)
Percent misaligned 6cyl Number of observations 959 Log likelihood 86.91 0.000 Prob4w2 Number of firms 178
322 34.37 0.000 86
322 28.53 0.000 86
629 50.14 0.000 116
629 45.04 0.000 116
o0.01. po0.05. po0.10.
reaping the greatest protection from mortality hazards. These results are consistent with the many studies that find negative effects of firm age on the hazard rate. The coefficient on firm size is also negative and highly significant, suggesting the presence of economies of scale or superior capability that contribute to survival over the entire period. Because recent ecological research has focused on better estimations of size effects (e.g. Dobrev, Kim, & Carroll, 2002; Dobrev & Carroll, 2003) and because this research has utilized data on the population of U.S. automobile manufacturers, we also analyze several models with a revised size specification (see Table 7). The premise of these papers is that while arguments regarding the effect of firm size on firm performance are ubiquitous in the
322
LYDA S. BIGELOW
Table 7. Piece-Wise Constant Rate Mortality Models of U.S. Automobile Manufacturers 1917–1934 with Revised Size Specification. Model 1 Ageo3 years Age 3–7 years Age 7–10 years Age410 years Misalign Firm size Sizeo50 Relative size 102 Relative size2 102 Left censored De alio
1.702 (.306) 1.736 (.311) 1.052 (.401) 1.690 (.404) 0.841 (.413) 0.00003 (0.0006) 0.524 (.236) 1.023 (.729) 1.052 (2.707) 0.380 (.282) 0.406 (.201)
Percent misaligned 4cyl
Model 3
Model 5
1.123 (.854) 1.040 (.856) 1.577 (.970) 0.235 (.963) 0.421 (.892) 0.0004 (0.0003) 0.438 (.390) 1.630 (2.734) 48.445 (159.647) 0.0.15 (.492) 0.971 (.349) 10.866 (2.780)
2.901 (.599) 3.051 (.614) 2.449 (.742) 3.280 (.791) 0.485 (.671) 0.00006 (0.00008) 0.741 (.362) 0.400 (1.024) 0.441 (33.017) 0.183 (.437) 0.207 (.283)
6.133 (2.068)
Percent misaligned 6cyl Number of observations Log likelihood Prob4w2 Number of firms
959 82.99 0.000 178
322 27.77 0.000 86
629 42.72 0.000 116
o0.01. po0.05. po0.10.
organizational and strategy literatures, there is actually a paucity of research that goes beyond examining the first-order effects of size. Size is generally considered as an absolute rather than relative measure. Both Dobrev, Kim, and Carroll (2002) and Dobrev and Carroll (2003) contend that a firm’s size relative to its rivals is what determines whether and how much protection from mortality is conferred via absolute size. Given these latest insights we
Technology Choice, Transaction Alignment, and Survival
323
tried specifying the model outlined in Dobrev and Carroll (2003). However, owing to the fact that this paper utilizes an observation period that is a subset of the nearly century-long observation period in the other papers, we have different results. Using the exact specification revealed no useful findings, due to the comparatively smaller number of observations and the presence of outliers. However, a slightly modified version of their specification revealed some interesting results though these results are consistent with the primary findings in Table 6. Table 7 presents Models 1, 3, and 5 with the revised specification. In addition to the size variable, these new models include a dummy variable for those firms that are quite small, producing less than 50 cars a year, sizeo50 as well as a relative size variable and a squared relative size term. A comparison of the log-likelihood values for the models shows that these new size specifications do not improve over the simpler size specification used in Table 6. However, the use of a dummy variable reveals that for those firms competing in the newer six-cylinder technology segment, small size does significantly increase the risk of failure. This result in combination with the aggregate sub-population misalignment measure suggests that when it comes to efficiency considerations, production cost concerns realized through economies of scale do matter, but that transaction costs concerns regarding boundary of the firm decisions may be less significant in light of the need to access new technologies and/or new markets. The coefficient on de alio in Model 1 is negative and statistically significant, though not as significant as size and age. These results suggest that de alio entrants displayed lower hazard rates than de novo firms over the period as a whole. Recall that this was one of the main findings in Carroll et al. (1996), who analyzed the same survival data but over a longer historical period. Finally, given that we are looking at only a limited number of years in the industry, we need to ascertain that there is no left-censoring bias. The control for this is a dummy variable, which is never significant in any of the models, indicating that left censoring is not biasing the results. Model 1 also provides support for H1. Misalignment has a strong significant positive effect on the likelihood of failure. Even after other variables are added to the model, the main effect of misalignment holds. (The results of the misalignment main effect are similar no matter which age split is used.) Comparing Models 2 and 4 in Table 6 we see that the coefficient of the misalign variable, though positive, is no longer significant. However, turning to Models 3 and 5, we see that when we add the percent misaligned variable for each technology we find that the misalignment of rivals has a strong and significant effect on firm survival. In the case of the mature technology, we
324
LYDA S. BIGELOW
find support for H2. As the percentage of inefficiently organized firms increases, the risk of failure for the focal firm is greatly diminished. This suggests that governance structure does indeed matter, and governance structure in aggregate is a potentially useful measure of the competitive intensity of the environment faced by firms competing within a technology class. In contrast, however, Model 5 suggests that efficiency in and of itself may not be the only driver of competitive intensity. We observe a positive and strongly significant effect of the degree of misalignment of rivals within the still-evolving six-cylinder technology sub-population. These results do not support H3, though they are not inconsistent with the nature of the technology trajectory and the impact of aggregate governance choices on firm survival. Obviously, there are several plausible explanations for this pattern of results. One possible explanation is that firms feel the pressure to get to market and to innovate and getting the alignment right from a transaction cost perspective is a secondary concern. Throughout most of the observation period (i.e. prior to 1930), getting to market and thereby establishing viability among customers, suppliers, and investors was a substantial concern. Pursuing a technology that involves the need for continued innovation may mean that there are potential risks to those firms that emphasize efficiency over investing in innovation. Perhaps there are benefits, despite the degree of asset specificity, for contracting out for the latest technology or alternatively, by investing in integration capabilities in order to continue to reap the benefits of learning and developing proprietary technological skills. What we do observe are dramatic differences in the aggregate governance choices of technological sub-populations. Further analysis of the data, particularly controlling for differences in the degree of aggregate and individual misalignment, will help clarify these results.
7. CONCLUSION The results presented above are among the first to test the performance implications of transaction cost economics by combining a measure of transaction cost alignment with a hazard rate model of firm mortality. As explained in the introduction, there is a staggering amount of empirical research that demonstrates that organizing along transaction cost economizing principles does occur, but there is a relative dearth of studies that have made the link between such economizing and firm-level performance. Of this recent research stream, a handful of studies have demonstrated the
Technology Choice, Transaction Alignment, and Survival
325
benefit of combining ecological insights and methods with transaction cost predictions of alignment (e.g. Silverman et al., 1997; Bigelow, 1999; Nickerson & Silverman, 2003; Argyres & Bigelow, 2005). Although the usefulness of testing theoretical predictions of organization in organizational ecology models of exit has been elaborated elsewhere (Barnett & Carroll, 1995; Carroll & Hannan, 2000), theoretical and empirical hurdles have stymied such research efforts. The paper presented here extends this line of research by including ecologically informed proxies for sub-population dynamics generated by differences in the maturity of competing technological trajectories. Getting the alignment right does indeed matter, but only if efficiency-based selection pressures are strong. Transaction cost economics and, indeed, other organizational theories do not explicitly specify how competitive pressures can be evaluated and incorporated in their theoretical predictions of firm structure. As Van Witteloostuijn (1998) and others have noted, inefficiency may or may not be a primary driver of firm mortality. Thus, as we introduce conceptions of differences in selection pressures via technological and industry evolution, it becomes apparent that efficient alignment is not always a first-order priority. Whether it is or not depends on the nature of the underlying technology utilized. This is of interest not only to transaction cost researchers, but to organizational and strategy scholars as well, who have long debated the importance of transaction cost governance choices. We now can add that consideration of the nature of technological evolution may mediate the relationship between governance structure and firm survival. We also note that there may be benefits in linking what we have done in this paper to contingency theory. By linking theories of technology evolution, organizational ecology and transaction cost alignment, the identification of contingencies is readily identified ex ante and can be generalized across populations. In sum, we find support for our misalignment hypothesis, H1, and strong support for our mature technology sub-population hypothesis, H2. We find that the misalignment variable in Model 1 has the predicted positive and significant effect on the failure rate. However, as we identify sub-populations by technology choice, we find that those operating in the mature technology as seen in Model 3 do benefit at the expense of inefficiently aligned rivals. Surprisingly, although H3 was not supported we find results consistent with the impact of sub-population or technologically localized competition on firm survival. Inefficiently aligned firms now present a threat to survival, perhaps because the still-evolving technology mandates timely investments in innovation.
326
LYDA S. BIGELOW
There are, of course, limitations to the interpretation of these results. First, our misalignment measure is necessarily a rough proxy for transaction cost economizing. We do not have access to the kind of survey data or other financial data that has been used in cross-sectional studies of transaction cost alignment, which led to the development and use of a proxy. Second, our measures of sub-population aggregate alignment are relatively coarse. It is, however, a first step. Future work will investigate this important comparison of competing risks from individual vs. sub-population misalignment. This paper makes contributions to both the transaction cost literature and the ecology literature by building on the recent effort to integrate these theories in order to provide greater insight into organizational structure and population dynamics. As stated above, the empirical and theoretical impediments to undertaking this work are non-trivial. We are however, able to provide an empirical test of the performance implications of transaction cost alignment. Further we see intriguing evidence of the importance of including measures of the state of technological evolution and the level of competitive intensity generated by technologically localized sub-populations. Further analysis may help unpack the qualifications and implications of this effect.
NOTES 1. Although creating a generalizable definition of the centrality or criticality of a transaction is beyond the scope of this paper, it is arguable that these criteria could be used to define a central transaction within other manufacturing industries. 2. Williamson (1996, p. 240) argues that yes, history matters, but this does not imply ‘‘that only history matters.’’ 3. These issues and results are elaborated in a separate paper (Bigelow, 2001). 4. The issue of how long a period of time it may take before inefficiently organized firms are selected out of a population is a current subject of interest (see Nickerson & Silverman, 2003; Argyres & Bigelow, 2005). In this paper, we assume that these efficiency-driven selection pressures may vary over time, but a weak-form selection environment holds throughout our period of observation. Greater examination of the tolerance of the selection environment and speed at which firms adapt is a topic worthy of its own paper. 5. See Carroll, Bigelow, Seidel, and Tsai (1996) and Bigelow (1999) for a complete description of the dataset collection process. 6. Recall that a firm may produce more than one model. A model in this dataset by definition must include a different engine type, i.e. styling changes alone do not constitute a different model. For single model firms, obviously Model 1 represents all of their production. However, for multi-model firms there is no significance to a model’s designation as Model 1. 7. Production data for this study is collected at the firm level, not the model level.
Technology Choice, Transaction Alignment, and Survival
327
8. See Carroll and Hannan (2000) for further discussion of mortality as performance measure. 9. Splits using slightly different time periods led to similar results. 10. These technical specifications include engine material, position of cylinders, number of cylinders, engine displacement, measurements of bore and stroke, and horsepower. The selection of which dimensions are most likely to determine the uniqueness of an engine was developed with the help of experts at the Behring Automotive Collection, Danville, CA and the National Automotive Historical Collection, Detroit, MI. 11. Using the concept of uniqueness as a proxy for asset specificity is consistent with Walker and Weber’s 1984 characterization of asset specificity, too. And uniqueness is used as a synonym in Riordan and Williamson (1985). 12. These estimates are available in a supplement. See Table 5 for select specifications relevant to constructing the misalignment measure used in this paper.
ACKNOWLEDGMENTS The author would like to thank Jackson Nickerson and Todd Zenger for insightful comments and suggestions. The author would also like to thank Glenn Carroll, Jackson Nickerson, and Oliver Williamson for providing critical comments on research that generated this paper. All errors and omissions remain the responsibility of the author.
REFERENCES Abernathy, W. J. (1978). The productivity dilemma: Roadblock to innovation in the automobile industry. Baltimore, MD: Johns Hopkins University Press. Abernathy, W. J., & Utterback, J. (1978). Patterns of industrial innovation. Technology Review, 80(7), 40–47. Alchian, A. (1950). Uncertainty, evolution and economic theory. Journal of Political Economy, 58, 211–221. Allison, P. D. (1984). Event history analysis. Newbury Park, CA: Sage. Anand, B. N., & Khanna, T. (2000). Do firms learn to create value: The case of alliances. Strategic Management Journal, 21(3), 295–316. Argyres, N., & Bigelow, L. (2005). Do transaction costs matter for survival at all stages of the industry lifecycle? Working paper, Washington University. Argyres, N. S., & Liebeskind, J. P. (1999). Contractual commitments, bargaining power, and governance inseparability: Incorporating history into transaction cost theory. Academy of Management Review, 24, 49–63. Armour, H., & Teece, D. (1978). Organizational structure and economic performance. Bell Journal of Economics, 9, 106–122.
328
LYDA S. BIGELOW
Baldwin, N., Georgano, G. N., Sedgwick, M., & Laban, B. (1987). The world guide to automobile manufacturers. New York: Facts on File. Barnett, W. P. (1997). The dynamics of competitive intensity. Administrative Science Quarterly, 42, 128–160. Barnett, W. P., & Carroll, G. R. (1995). Modeling internal organizational change. Annual Review of Sociology, 21, 217–236. Baum, J., & Mezias, S. (1992). Localized competition and organizational failure in the Manhattan hotel industry, 1898–1990. Administrative Science Quarterly, 37, 580–604. Bigelow, L. S. (1999). Transaction cost alignment and organizational survival in the early American automobile industry. Ph.D. dissertation, University of California, Berkeley. Bigelow, L. S., & Argyres, N. (2006). The impact of transaction costs on make-or-buy decisions: A population-wide test, forthcoming. Journal of Economic Behavior and Organization. Bigelow, L. S., Carroll, G. R., Seidel, M.-D. L., & Tsai, L. (1997). Legitimation, geographical scale, and organizational density: Regional patterns of foundings of American automobile producers, 1885–1981. Social Science Research, 26(4), 377–398. Boerner, C. S., & Macher, J. T. (2004). Transaction cost economics: An assessment of empirical research in the social sciences. Working Paper. Georgetown University. Carroll, G., & Wade, J. (1991). Density dependence in the organizational evolution of the American brewing industry across levels of analysis. Social Science Research, 20, 271–302. Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in populations of organizations. American Journal of Sociology, 90, 1262–1283. Carroll, G. R. (1997). Long-term evolutionary change in organizational populations: Theory, models and empirical findings from industrial demography. Industrial and Corporate Change, 6, 119–143. Carroll, G. R., Bigelow, L. S., Seidel, M.-D., & Tsai, L. (1996). The fates of de novo and de alio producers in the American automobile industry 1885–1981. Strategic Management Journal, 17, 117–137. Carroll, G. R., & Hannan, M. T. (1995). Organizations in industry: Strategy, structure, and selection. New York: Oxford University Press. Carroll, G. R., & Hannan, M. T. (2000). The demography of corporations and industries. Princeton, NJ: Princeton University Press. Christensen, C., Suarez, A., & Utterback, J. (1998). Strategies for survival in fast-changing industries. Management Science, 44, 207–220. Dobrev, S. D., & Carroll, G. R. (2003). Size and competition among organizations: Modeling scale-based selection among automobile producers in four major countries, 1885–1981. Strategic Management Journal, 24(6), 541–599. Dobrev, S. D., Kim, T.-Y., & Carroll, G. R. (2002). The evolution of organizational niches: U.S. automobile manufacturers, 1885–1981. Administrative Science Quarterly, 47(2), 233–265. Flammang, J. M. (1989). Standard catalogue of American cars 1976–1986 (2nd ed.). Iola, WI: Krause. Friedman, M. (1953). Essays in positive economics. Chicago: University of Chicago Press. Georgano, G. N. (Ed.). (1982). The new encyclopedia of motor cars: 1885 to present (3rd ed.). New York: E.P. Dutton. Gulati, R. (1998). Alliance and networks. Strategic Management Journal, 19(4), 293–318. Gunnell, J. (1987). Standard catalogue of American Cars 1476–1975 (3rd ed.). Iola, WI: Krause.
Technology Choice, Transaction Alignment, and Survival
329
Hannan, M., & Freeman, J. (1977). The population ecology of organizations. American Journal of Sociology, 82, 929–964. Hounshell, D. (1984). From the American system to mass production 1800–1932. Baltimore, MD: Johns Hopkins University Press. Kimes, B. R., & Clark, H. A. (1989). Standard catalogue of American cars 1805–1942 (2nd ed.). Iola, WI: Krause. Klein, P. G., & Shelanski, H. A. (1995). Empirical research in transaction cost economics: A review and assessment. Journal of Law, Economics, and Organization, 11, 335–361. Klepper, S. (1996). Entry, exit, growth, and innovation over the product life cycle. American Economic Review, 86, 562–583. Klepper, S. (2002). Firm survival and the evolution of oligopoly. RAND Journal of Economics, 33, 37–61. Lester-Steele, D. (1960). Automobile specifications 1915–1945. Columbus, OH. Masten, S. E. (1993). Transaction costs, mistakes, and performance: Assessing the importance of governance. Managerial and Decision Economics, 14(2), 119–129. Masten, S. E., Meehan, J. W., & Snyder, E. A. (1991). The costs of organization. Journal of Law, Economics, and Organization, 7, 265–273. Mayer, K., & Nickerson, J. (2005). Antecedents and performance implications of contracting for knowledge workers: Evidence from information technology services. Organization Science, 1(3), 225–242. Meyer, M., & Gupta, V. (1994). The performance paradox. Research in Organizational Behavior, 16, 309–370. Nickerson, J. (1997). Toward an economizing theory of strategy: The choice of strategic position, assets, and organizational form. Ph.D. thesis, University of California at Berkeley. Nickerson, J., Hamilton, B., & Wada, T. (2001). Market position, resource profile, and governance: Linking Porter and Williamson in the context of international courier and small package services in Japan. Strategic Management Journal, 22, 251–274. Nickerson, J., & Silverman, B. (2003). Why firms want to organize efficiently and what keeps them from doing so: Inappropriate governance, performance, and adaptation in a deregulated industry. Administrative Science Quarterly, 48, 433–465. Peli, G., & Nooteboom, B. (1999). Market partitioning and the geometry of the resource space. American Journal of Sociology, 104, 1132–1153. Petersen, T. (1995). Analysis of event histories. In: G. Arminger, C. C. Clogg, & M. Sobel (Eds), Handbook of statistical modeling for the social and behavioral sciences (pp. 453–517). New York: Plenum Press. Poppo, L., & Zenger, T. (1998). Testing alternative theories of the firm: Transaction cost, knowledge-based, and measurement explanations for make-or-buy decisions in information services. Strategic Management Journal, 19, 853–877. Rae, J. B. (1959). American automobile manufacturers, the first forty years. Philadelphia, PA: Chilton. Rae, J. B. (1984). The American automobile industry. Boston, MA: G.K. Hall & Co.. Raff, D. (1991). Making cars and making money in the interwar automobile industry: Economies of scale and scope and the manufacturing behind the marketing. Business History Review, 65, 721–753. Riordan, M., & Williamson, O. (1985). Asset specificity and economic organization. International Journal of Industrial Organization, 3, 365–378.
330
LYDA S. BIGELOW
Sampson, R. (2004). The cost of misaligned governance in R&D alliances. Journal of Law, Economics, and Organization, 20, 484–527. Silverman, B. S., Nickerson, J. A., & Freeman, J. (1997). Profitability, transactional alignment, and organizational mortality in the U.S. trucking industry. Strategic Management Journal, 18, 31–52. Sorensen, J. (1999). Stpiece subroutine for use in event-history analysis with stata. College Station, TX: Stata Corp. Tuma, N. B., & Hannan, M. T. (1984). Social dynamics: Models and methods. New York: Academic Press. Van Witteloostuijn, A. (1998). Bridging behavioral and economic theories of decline: Organizational inertia, strategic competition, and chronic failure. Management Science, 44(4), 501–519. Villalonga, B., & McGahan, A. (2005). The choice among acquisitions, alliances and divestitures. Working Paper. Harvard University. Walker, G., & Poppo, L. (1991). Profit centers, single-source suppliers and transaction costs. Administrative Science Quarterly, 36, 66–87. Williamson, O. E. (1985). The economic institutions of capitalism. New York, NY: Free Press. Williamson, O. E. (1991). Comparative economic organization: The analysis of discrete structural alternatives. Administrative Science Quarterly, 36, 269–296. Williamson, O. E. (1994). The Politics and Economics of Redistribution and Inefficiency. Unpulished manuscript, University of California, Berkeley. Williamson, O. E. (1996). The mechanisms of governance. New York, NY: Oxford University Press. Williamson, O. E. (1999). Strategy research: Governance and competence perspectives. Strategic Management Journal, 20(12), 1087–1108. Winter, S., Kaniovski, Y., & Dosi, G. (2003). A baseline model of industry evolution. Journal of Evolutionary Economics, 13, 355–383. Yamaguchi, K. (1991). Event history analysis. Newbury Park, CA: Sage.
Technology Choice, Transaction Alignment, and Survival
331
APPENDIX Producing a measure of misalignment that can be compiled over hundreds of firms for 18 annual observations requires first constructing a measure of asset specificity that, too, can be coded across firms and over time: a non-trivial task. Clearly, the conventional operationalization of asset specificity via survey (e.g. of engineering effort or cost data) is not feasible. Instead, we utilize information on the technical specifications of each engine to construct a proxy for asset specificity.10 We then use this measure of asset specificity in a Probit model to test whether high levels of asset specificity increase the likelihood of vertical integration as predicted by transaction cost economics. Finally, we use the results of the Probit model and interact this with the actual governance decision to determine the degree of misalignment (Misalign 1). Asset specificity is constructed using data on the engine displacement or cc variable. The cc variable was chosen in order to distinguish those engines that are particularly unusual in their size and power, and from this we label our proxy for asset specificity unique-eng. The measure used here is intended to capture this physical and technical uniqueness.11 Starting with the entire distribution of cc for a given year, the values are standardized and a dummy variable for the upper portion of the distribution is created. This variable is coded zero if the standardized measure of cc for that engine for that year is less than 2, 1 if the standardized measure of cc for that engine for that year is greater than or equal to 2. In other words, engines that are at least 2 standard deviations above the mean have a value of 1 for the unique-eng variable, zero otherwise. Since engines that fall in this category must either be physically larger or more unusual in terms of their bore and stroke measures, this is a rough proxy for asset specificity. It also allows for the maximum number of factors that can be rearranged in order to produce similar levels of cc. In other words, the lower end of the distribution is not considered highly unique because these low levels generally utilize the same number of cylinders (4), whereas the high end of the distribution could use (and does use) 6, 8, 12, or 16 cylinders to produce the same level of cc. Given that the interconnectedness of other drivetrain components depends in part on the number of cylinders, the implication is that, redeployability is greatly reduced for engines at this high end (right tail) of the distribution compared to those in the middle or left tail of the distribution. It is important to emphasize that the distribution from which this assetspecificity variable is derived includes the entire population of automobile manufacturers. Again, relying on the distribution of technical characteristics of engines in the dataset, it is possible to identify those engines that are
332
LYDA S. BIGELOW
extremely unusual on a given dimension, as compared to other engines (controlling for year, and market segment as well). In other words, although there may be other possible technical configurations of engines, we assert that such configurations are hypothetical ideals, not realistic alternatives. Thus, the problem of sample selection bias is mitigated in the construction of the scale for determining asset specificity. We then include the variable unique-eng in a Probit model where the dependent variable represents the likelihood of vertically integrating a given component. In this model, we utilize only those variables that are theoretically pertinent from a transaction cost perspective (e.g. variables that control for transaction characteristics and economies of scale and scope), thus the model took the following form: ProbðMAKEijt Þ ¼ aijt þ bjt UNIQUEENGINEjt þ gjt FIRMSIZEjt þ djt MULTIMODELit þ ijt
where Prob (MAKEijt) takes a value of zero if firm j purchases engine i from an outside supplier at time t, and the value of 1 it produces engine i in-house at time t. UNIQUE ENGINEjt is our cubic displacement-based measure, FIRM SIZEit measured by sales, and MULTIMODELit takes the value 1 if the firm produced more than one car model and zero otherwise, and eijt a vector of residuals. We expected positive and significant signs on b, g, and d. Both the UNIQUE ENGINE and FIRM SIZE variables carried positive and statistically significant coefficients, indicating that the behavior of auto firms with regard to engine make-or-buy decision during this period was generally consistent with predictions from transaction cost theory. The coefficient on MULTIMODEL was not significant, indicating that economies of scope from multi-engine production, if they existed, were not important in engine make-or-buy decisions. To check that the estimates from this Probit model were robust, we estimated several other models that included other, non-transaction cost-oriented covariates, as well as other engine characteristics.12 UNIQUE ENGINE and FIRM SIZE retained their statistical significance in all these additional regressions. Table A presents a set of results, which come directly from Bigelow (2002), a paper that elaborates and explains more fully these tests. Note that the coefficient for unique-eng is consistently positive as expected. Finally, to construct the misalignment measure used in the paper above, we utilize the residual from the Probit model with the actual firm structure, where 1 ¼ integration, 0 ¼ outsourcing. The predicted values (p) from the model tell us what the model predicts concerning the probability that a given engine will be produced in-house by a given firm in a given year, based on the values of
333
Technology Choice, Transaction Alignment, and Survival
the three explanatory variables. Therefore, if a given engine was in fact produced in-house, the greater is p, the more ‘‘aligned’’ is the producing firm for that particular engine transaction. On the other hand, if a firm in fact bought an engine from an outside supplier, the greater is p, the more ‘‘misaligned’’ is the firm for that transaction. We therefore created a variable, ALIGNMENTjt, for each engine–year transaction that takes the value of p if the firm made the given engine in-house, and 1 p if the given engine was purchased from outside suppliers. We then calculated the inverse of ALIGNMENTjt to create our misalignment measure, MISALIGNMENTjt for each engine in each year. Misalignment is thus a continuous measure. So, as Misalign 1 increases we posit that the likelihood of failure should increase. Table A.
Probit Models of Integrated Engine Manufacture. Adjusted for Clustering on Firm Model 1
Constant Bore Unique-eng
0.459 (0.687) 0.113 (0.215) 1.065 (0.504)
Stroke
Model 2
Model 3
2.175 (2.515) 0.618 (0.796) 1.215 (0.560) 0.009 (0.187)
3.446 (2.480) 0.802 (0.761) 1.26 (0.593) 0.129 (0.176) 0.064 (0.011)
Firm age Firm size Number of models
Number of observations Log likelihood Prob4w2 Pseudo R2
1158 779.24 0 0.022
1158 778.47 0 0.023
1158 778.46 0 0.023
po0.10. po0.05. po0.01 (adjusted standard errors in parentheses).
Model 4
Model 5
3.143 (2.498) 0.354 (0.776) 1.166 (0.596) 0.313 (0.187) 0.045 (0.012) 2.09E-5 (8.22E-6)
2.626 (2.546) 0.223 (0.792) 1.149 (0.578) 0.309 (0.187) 0.049 (0.013) 2.02E-5 (7.90E-6) 0.145 (0.135)
1158 697.86 0 0.12
1158 653.78 0 0.18
This page intentionally left blank
334
EXPLORING THE TAIL OF CREATIVITY: AN EVOLUTIONARY MODEL OF BREAKTHROUGH INVENTION Lee Fleming and Mark Szigety ABSTRACT Where are the ultimate sources of technological breakthroughs? What makes a firm more likely to invent a breakthrough or to exploit external breakthroughs? We develop an evolutionary model of invention as a process of recombinant variation and selection. Our contributions are to highlight the skewed outcome distributions resulting from evolutionary search and to develop theory that can be tested by modeling the higher moments of search processes. Recent methodological and data collection advances make such testing possible. We motivate further research, develop our model’s strategic implications, and discuss how managers might create and respond to breakthroughs.
INTRODUCTION While much has been written about innovation, less has been written about invention. Schumpeter defined innovation as the commercial application or Ecology and Strategy Advances in Strategic Management, Volume 23, 335–359 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23011-0
335
336
LEE FLEMING AND MARK SZIGETY
adoption of an invention, arguing that ‘‘the making of the invention and the carrying out of the corresponding innovation are, economically and sociologically, two entirely different things’’ (Schumpeter, 1939, p. 85). Invention deserves study because it provides the basis for innovation; until someone creates a new technology or idea, there can be no creation of wealth. Breakthrough inventions cause particularly wide-reaching and ecological change. Without breakthroughs, there would be little discontinuous or competencydestroying change (Tushman & Anderson, 1986), few disruptive technologies (Christensen, 1997), and far less interesting population dynamics (Hannan & Freeman, 1977; Bigelow, 2006). This paper focuses exclusively on the process of invention – on breakthroughs in particular. We refer the reader interested in innovation to the large literature on commercialization, product development, and technology strategy. (For entry points, see Tushman & Moore, 1980; Wheelwright & Clark, 1992; Utterback, 1996.) We follow a classical stream of literature on technological change and develop an evolutionary model of invention and creative search (Schumpeter, 1939; Tushman & Anderson, 1986; Basalla, 1988; Utterback, 1996). Our contribution is to highlight the asymmetric and skewed outcome distributions of evolutionary processes and to argue for explicit consideration of the higher moments in theory and estimation. Recent advances in econometrics (King, 1989; Weesie, 1999) make it possible to estimate influences on the variance of continuous and count distributions. Since these advances enable testing of hypotheses, they greatly increase the value of theory that predicts higher moments. Most importantly, they provide the opportunity for more penetrating investigation of inventive breakthrough, a phenomenon which has so far received little attention, largely because it occurs so rarely and is so difficult to predict (Tushman & Anderson, 1986; Utterback, 1996; Fleming, 2002). Due to the challenges of collecting comprehensive datasets of inventive ecologies over time, much work on breakthroughs has suffered from the inherent selection bias of studying known breakthroughs after the fact. Recent advances in computation and availability of archival data make it possible to avoid some of the selection issues. Thus, the combination of new econometrics and better data provides an opportunity to increase our understanding of inventive breakthroughs. Though complexity and some organizational theory have begun to investigate higher moments (March, 1991) and extreme distributions such as power laws (Bak & Chen, 1991), surprisingly little thinking on breakthroughs has framed the issue around the processes that generate and influence highly skewed distributions. Our research question is, therefore, the following: Given that breakthroughs are the far right outliers of skewed distributions, what processes
Exploring the Tail of Creativity
337
create and influence the shape of these distributions? In particular, what influences increase the probability that inventors, firms, and industries will invent an extreme outlier that constitutes a breakthrough? Our answer adopts an evolutionary approach that builds on the classical epistemology of Campbell (1960), the philosophy of science (Hull, 1988), the evolution of technology (Basalla, 1988), and recent work in psychology (Simonton, 1999). The theory requires multiple levels of analysis because technological breakthroughs cause far-reaching and unpredictable changes across firms and industries, yet the source of creativity is ultimately the mind of the individual inventor. As with all evolutionary theories, analysis must proceed across levels, as variation occurs within individual minds and is selected at progressively higher levels of analysis such as organizations and industries (Weick, 1979; McKelvey, 1997). We will explore these ideas in four sections. The first section reviews definitions and models of generative creativity and invention. The second builds a model of invention as a process of recombinant search and considers psychological and technological influences upon the distribution of outcomes. The third extends this model to organizational and managerial contexts in order to predict the most likely sources of technological breakthroughs. The fourth discusses managerial strategies that can be applied to invent, recognize, and exploit breakthroughs.
AN EVOLUTIONARY MODEL OF GENERATIVE CREATIVITY AND INVENTION To develop an evolutionary analogy for invention, we build on the work of scholars from many disciplines who have contributed to evolutionary epistemology and have proposed that knowledge results from an evolutionary process of combinatorial search. Darwin (1859) developed the most influential evolutionary argument that biological species evolve through a process of variation, selection, and retention. In the natural world, this process begins with the mutation and recombination of old (and the introduction of new) genetic material in a population. New genetic combinations confront a selection environment, which suits some of them more than others. Organisms with fitter genetic combinations reproduce more prolifically; hence, their traits are increasingly represented in subsequent generations. Building on Darwin’s ideas, Campbell (1960) argued that all learning follows a blind variation–selection–retention framework, from the process by which an
338
LEE FLEMING AND MARK SZIGETY
organism evolves sense organs to the processes by which humans generate knowledge and think creatively. He proposed that individuals learn and create new knowledge by (a) seemingly random generation of new combinations of ideas, (b) selection of the best of those ideas using a variety of criteria, and (c) retention of a small fraction of those combinations as bases for further generative search. Unfortunately, almost all combinations fail to function as intended and thus do not provide any basis for future recombination. This process of variation and selection occurs first within the inventor’s mind, where most new ideas are rejected through internal thought experiments immediately after formulation. Simonton (1999) extends Campbell’s arguments, but devotes greater attention to the retention stage and the definition of creativity. He studies genius across many spheres of creative endeavor and argues that creativity is best measured by its impact on culture and on further search. Creative genius, in Simonton’s definition, must be recognized by others in order to be influential: ‘‘Unrecognized genius becomes an oxymoron’’ (Simonton, 1999, p. 5). Many scholars of technology propose similar mechanisms for the invention and evolution of technology. The economist Schumpeter (1939) and his contemporary, the sociologist Gilfillan (1935), agreed that technological novelty consists of original combinations of extant inventions, components, and applications. Inventors either import new artifacts or components from the natural world or identify them through scientific research (Fleming & Sorenson, 2004). Biological evolution, according to classical theory, is typically only valid within a given species,1 but technological variation can mix any technological components, regardless of provenance (Kroeber, 1948). Nelson and Winter (1982) agree that the ultimate source of creativity lies in new combinations; they model creativity as a stochastic draw. Basalla (1988) develops the most comprehensive analogy of technological change to biological evolution, from an anthropological and historical perspective. Not only does he agree that the recombination of existing technologies and artifacts provides variation, but he makes an argument, similar to Simonton’s, that the ultimate value of a technology can only be judged by its popularity as the basis for future recombinant search.2 He further states that few inventions ever attain such popularity; society retains only a small subset of variants because cultural, technical, economic, and political factors must simultaneously be satisfied (Basalla, 1988; Pinch & Bijker, 1987). These evolutionary views of invention imply a definition for breakthrough: A new combination that provides the basis for a disproportionate share of future generative search, both directly and indirectly. This definition is consistent with historical accounts and with observed distributions of
Exploring the Tail of Creativity
339
creative outcomes. On the one hand, most inventions confer little advantage and are never used in any further recombination. On the other, a very few inventions provide the generative basis for a great deal of further search. Note that in this model, all new combinations qualify as inventions, regardless of their success. This definition avoids the assumption of success that is almost endemic in the common usage of the term innovation and thus avoids the severe selection bias that commonly handicaps the study of technological change. As a result, an evolutionary and recombinant view lends itself much more readily to scholarly research and empirical disconfirmation. Armed with this evolutionary conception of invention, a researcher can now (a) study how an inventor or organization selects components, (b) observe the processes of recombinant search, and (c) measure the success of the outcomes. Rather than studying a single breakthrough ex post, a researcher may now identify variance on both sides of the equation that models breakthroughs on one side and their causes on the other. If invention is indeed a process of recombinant search, then what are the combinatorial components? If one adopts an artifactual view of technology, then any natural or manufactured material or object is a potential component (Basalla, 1988).3 Of all the possible combinations of all the possible components, most would seem bizarre, given typical sets of technologies and commonly accepted associations. But perceptions of ‘‘appropriate’’ components and component sets are a cognitive and social construction and, theoretically, there is nothing stopping an inventor from attempting to combine ‘‘shoes and ships and sealing wax’’ – any components from the physical world. As an example of a strange but previously common combination, the Romans mixed blood and horsehair together to strengthen their concrete. Today, it seems natural for a medical diagnostician to work with high-speed computers, fluorescent quantum dots, and semiconductor fabrication technology. Yet the juxtaposition of such tools would have appeared rather odd only 20 years ago.
SOCIAL-PSYCHOLOGICAL AND TECHNOLOGICAL INFLUENCES UPON THE DISTRIBUTION OF EVOLUTIONARY OUTCOMES We have adopted an evolutionary perspective and a definition of invention that have both been grounded in historical research and are consistent with the observed distributions of creativity and invention outcomes. Since the
340
LEE FLEMING AND MARK SZIGETY
ultimate goal is to predict the outliers of such distributions, we will now develop arguments on how a variety of factors might influence the higher moments. We will focus on technological and social-psychological influences in this section4 and higher-level influences in the next. The organization is somewhat arbitrary. We start with a psychological model because we assume that creative variation, while obviously influenced by many factors at all levels of analysis, cannot occur outside an individual mind. For example, while two inventors are probably more creative if they work together and build on each other’s suggestions, each of the creative linkages will still occur within one mind or the other. Selection processes, however, take place both within the individual and in higher-level contexts. This can cause confusion because variation occurs at the level of the individual, selection occurs at multiple levels of analysis, and sources from various levels feedback to influence the variation stage (Weick, 1979; McKelvey, 1997, 1999). We begin by asking what influences an individual’s creativity and the probability of inventing a breakthrough. Simonton (1999) argues that ‘‘the generation of ideational variants depends on the richness of a person’s associative network. The larger the number of concepts in the network, and the more diffuse their interconnections, the greater is the potential supply of recombinations’’(p.152). Such combinatory play requires a suspension of judgment and has been reported to occur subconsciously with mainly visual associations and more often in mentally unstable individuals (Simonton, 1999). More creative individuals can evaluate the novel associations against a greater number of criteria, and thus increase the possibility of a productive ‘‘hit.’’ Hence, those who can simultaneously juxtapose, combine, and evaluate a divergent stream of uncombined input will be more creative. Many studies have also established a strong correlation between an individual’s total inventive output and his or her likelihood of creating a work of extremely high impact. For example, eminent scientists who have published a highly cited paper will generally have published many papers, including many that are poorly cited (Simonton, 1997). In our current context, this result implies that the most prolific inventor is the one most likely to invent a breakthrough. A one-hit wonder is very unlikely. The evolutionary model implies such a result directly, given the reasonable assumption that inventors have difficulty identifying and selecting which of their ideas are best for elaboration. Though this paper focuses on inventing or exploiting the rare breakthrough, it is important to keep in mind that there are no easy routes to an extreme outlier; Edison’s admonition about perspiration cannot be sidestepped. Applying these results to predict who is most likely to invent
341
Exploring the Tail of Creativity
a breakthrough identifies the most productive individual, by sheer number of inventions, as the best first-order approximation. While total productive output is much of the story, it is not the whole story. We also need to consider the distribution of success of the inventor’s creative output. This relationship between total productivity and breakthroughs is best understood by considering the histogram of the number of creative products of a particular quality. Fig. 1 displays a representative distribution of a creative person’s output. Generally, it illustrates the number of ideas and some indication of the quality of each idea; specifically, it might be the number of reprints of each of an author’s books, the number of performances of each of a composer’s pieces, or (as it actually does) the number of citations to an individual’s papers or patents. The illustration follows a highly right-skewed distribution that can be roughly characterized as negative binomial, Pareto, log normal, or exponential. Most of the creative works have almost no influence (as illustrated by the high bars on the
0.15
Density
0.1
0.05
0 0
20
40
60
80
100
quality
Fig. 1. Representative Histogram of Creative Works and their Quality. It Might Illustrate the Number of Reprints of each of an Author’s Books, the Number of Performances of Each of a Composer’s Pieces, or the Number of Citations to Each of an Individual’s Papers or Patents.
342
LEE FLEMING AND MARK SZIGETY
left), but a select few have exceedingly high influence (as illustrated by the occasional data point on the far right). While the output of all creative individuals will follow a skewed distribution, subtle differences will remain and motivate two questions for those interested in breakthroughs: What increases the total number of creative outcomes? What increases the mean, variance, and skew of an inventor’s creative output? Because most of the extant literature on creativity already considers the number of outcomes and the first moment (that is, the average), and because statistical estimators for the third moment have yet to be invented, we will focus on the second moment. Though methods for estimating the second moment have been recently improved5 (King, 1989; Weesie, 1999), there remains little theoretical or empirical work that considers the higher moments, with the exception of the organizations and technology search literatures. March and Simon (1958) made the earliest argument that local search increases the average outcome and decreases the chances of failure because it refines already known and previously discovered approaches. March (1991) elaborated the idea of local search theoretically, focusing explicitly on outliers and the second moment. He also advanced the argument that in severely competitive contexts, where only the extreme positive result is acceptable, the mean outcome becomes irrelevant; what really matters is the maximum of the distributional draw. Extending his argument to our severely competitive context of breakthroughs highlights the importance of drawing recombinant outcomes from a high-variance and highly skewed distribution. Such a distribution implies just the opposite of local search, which avoids high-risk outcomes in favor of more likely, if much less spectacular, success. Fleming (2001) demonstrates that the use of new components and combinations decreases the mean and increases the variance of inventive outcomes. Inventors can also seek or avoid variance by means of the modularity or interdependence of their components. Use of interdependent components – where a small change in one component can cause a large change in the ultimate success of the combination – provides a fertile but risky opportunity. Attempting to recombine interdependent components creates a rugged landscape, where local search algorithms quickly break down (Kauffman, Lobo, & Macready, 2000). Application of a simple ‘‘hill climbing’’ algorithm, for example, will usually strand the inventor on a local maximum and leave the global maximum undiscovered. Use of interdependent components increases the variance and decreases the mean of recombinant search outcomes, such that inventors create less useful combinations on average but also increase the chances of a breakthrough. The application of scientific
Exploring the Tail of Creativity
343
knowledge, however, ameliorates these influences and greatly improves the chances of a successful breakthrough, though the absolute probabilities remain low (Fleming & Sorenson, 2004). Almost no work in the psychology and sociology literatures has framed the question of breakthrough creativity as an investigation of the distributional moments of inventive output. Simonton (1997) comes closest when he establishes that the ratio of the number of major works to the number of minor works remains roughly constant over a productive career. Simonton calls this constant probability of breakthroughs the ‘‘equal odds’’ rule. As an example of how psychological research might leverage the higher moments, one might predict that perfectionists demonstrate higher means and lower variance because they only publish their best work (though the distribution would be far from uniform and concentrated, due to the difficulty of identifying the best work ex ante). In contrast, risk-seekers would publish anything in the hope that something might provide a breakthrough. As a result, the distribution of their creative outcomes would be fatter in both tails. Risk aversion should decrease with heightened aspiration and improved performance (March & Simon, 1958; Bromiley, 1991). Individuals who move to significantly different fields would demonstrate higher variance. (Again, the mean effect could be argued both ways, since new combinations can be worse on average but also give rise to outliers in both directions; see Fleming, 2001). Individuals with conventional and focused education might be expected to demonstrate higher-mean and lower-variance distributions, since they would have mastered a refined and previously organized body of knowledge and creative components. We are aware of no work from a sociological perspective on the distributional moments of creativity. Indeed, the entire subject of structural influences on creativity remains relatively unexplored (Burt, 2004; Obstfeld, 2005). Individuals who broker relationships (those who work with coinventors who do not work together without the central broker; see Burt, 2004) appear to be more creative because they span different areas of expertise and hence are more likely to create a new combination. But they probably also demonstrate higher variance because they lack depth in most areas and, on average, should be less able to judge creative work spanning multiple areas of expertise. Consistent with this argument of higher variance, cohesion should also provide for more rigorous selection. Cohesive inventors will probably question one another’s creativity more closely, so that the resulting quality is higher and less variable. Apart from the structural argument (Reagans, Zuckerman, & McEvily, 2004), demographic diversity should also increase the variance. Indeed, the conflicting results on
344
LEE FLEMING AND MARK SZIGETY
Table 1. Summary of Predictions for the Second Moment of Creative Outcome Distributions for Technological and Social-Psychological Influences. Impact on Variance of Distribution of Inventive Search Technology variables Local search New search components Interdependence (inverse of modularity) Scientific knowledge X interdependence
+ +
Social-psychological variables Risk aversion Individual’s change of creative fields Education Recent performance that exceeded expectations Unmet individual aspirations Perfectionism Brokerage of collaborative relationships Diversity of collaborators
+ + + + +
the relationship between diversity and many outcomes might be partially resolved by considering the higher moments. We summarize our arguments for the technological and psychological contexts in Table 1.
ORGANIZATIONAL AND MANAGERIAL INFLUENCES UPON THE DISTRIBUTION OF EVOLUTIONARY OUTCOMES Although invention and creative thought first occur within an individual, it is generally organizations that bring breakthroughs to fruition – and feel their impact. Just as scientists come up with idiosyncratic ideas which are then winnowed and refined by their colleagues and scientific colleges (McKelvey, 1999; Hull, 1988), inventors come up with new combinations which are then vetted by their managers and refined by their co-inventors within firms and technological communities. As previously cited, selection occurs at multiple levels of analysis and these levels influence the variation stage in turn (McKelvey, 1997; McKelvey & Baum, 1999). To account for
Exploring the Tail of Creativity
345
interactions across multiple levels of analysis, we will now extend our model to higher levels of analysis and then consider the strategic implications in our concluding section. We will refer to an inventive effort as ‘‘collective’’ whenever an inventor works with other inventors within a firm, a scientific laboratory, or an industrial or open innovation community. By this definition, an inventor almost always invents collectively and within an ‘‘ecology’’ of invention; indeed, we would be hard pressed to identify a completely isolated inventor. Before discussing the strategically valuable nuances that facilitate or hinder collective inventive effort, we offer a simple aggregated model of it. The assumption of independence between collective inventors (in both a statistical and substantive sense) implies that an organization’s total inventive output will simply be the sum of the individuals’ outputs. Under such an assumption, one could add all the research employees’ inventive distributions to arrive at the organization’s aggregate inventive distribution. Adding individual distributions – each characterized as Pareto or exponential, as described above – also results in right-skewed distributions.6 This fundamental mathematical result implies that the distribution of breakthroughs in a firm, industry, or community will be right-skewed. Indeed, this scale-free result should hold at any level and angle of analysis (we are not aware of any contrary evidence), whether one considers work teams, labs within a firm, firms, universities, industries, or a particular technological niche. Rightskewed distributions are also exactly the distribution we would expect to observe if technological invention followed power law processes of punctuated equilibrium (Tushman & Anderson, 1986). It is easy, given the importance of outliers for breakthroughs, to overlook the first-order effects. For organizations, as for individuals, the more prolifically they invent, the more likely they are to invent a breakthrough (the argument sets a baseline under the discussion of whether small or large firms are more innovative; see Acs & Audretsch, 1988). There are many secondorder effects, but the first-order effect implies that large firms are the likely sources of most breakthroughs simply because they invest more in technological search and as a result, invent more, at least on an absolute scale.7 This arithmetic assumes, however, that inventors invent independently, an assumption which is obviously unrealistic because both the variation and selection stages are sensitive to collaborations, information flows, incentives, and the managerial context. The multiple and interdependent feedback influences remain difficult to identify; the best ecological image of the process is that of a messy and tangled web of individual inventors, embedded in various organizations, institutions, and incentive structures, and exposed to
346
LEE FLEMING AND MARK SZIGETY
a wide variety of amounts, types, and distributions of information. We will organize this ecological mess by looking at three types of inventive organizations: firms; science laboratories, both public and private; and communities, both open by design and open despite proprietary boundaries. Though much has been written on what makes an organization more creative, we will focus on breakthroughs and reconsider the question from a distributional perspective. While hard work will obviously increase the probability of development success, its influence on the second moment is unclear. Pressured engineers probably take greater risks, but they are also more likely to build kludges and stopgap solutions. Hence, simple managerial pressure probably decreases the chance of a breakthrough. Hard work can also be elicited through financial incentives (such as stock options); cynical manipulation of recently graduated engineers; or a sense of mission, either technological (build the fastest widget in history) or societal (cure cancer). To the extent that hard work can be intrinsically motivated, the firm’s inventors should be more creative and likely to invent a breakthrough (Amabile, 1988). Such efforts may only work for a short time, however, before inventors burn out. The area of the inventors’ work efforts can also influence a firm’s inventive productivity. Extending the arguments presented above, firms that focus on previously searched combinatorial opportunities are much less likely to invent a breakthrough. Similarly, firms that work with well-defined and modular components (which often are already industry standards) will be less likely to invent a breakthrough. Firms with non-technical management will often favor exploitation over exploration, due to financial modeling and planning techniques that promote proximal and certain gains (though that is changing with the slow adoption of options theory) and an MBA education that favors control and certainty over creativity and risk. As industries mature, their firms become demographically older and more insular, particularly if firms do not hire new employees and/or individuals do not change employers. As a result, inventors increasingly recycle the same set of components. Indeed, demography itself – educational backgrounds and tenure in the industry or organization – should explain much of the second moment of the creative output of a group of inventors. As a result of these influences, motivation for demographic change (such as poor performance; see Boone, Olffen, Witteloostuijn, & De Brabander, 2004) should be observable, with delay, in greater variance of output. Firms in regions lacking diversified industries will also become insular and stale, particularly if the region’s legal infrastructure and culture does not support the movement of inventors across firms.
Exploring the Tail of Creativity
347
A firm’s attitude toward risk will often reflect managerial concerns, which are legitimate but should be consciously managed. Managers send a very strong signal by how they reward or punish failure. Individual inventor incentives must be specified with consideration of the organization’s aggregate portfolio goals; if inventors are punished for failure or rewarded only for steady, incremental work, it is very unlikely that they will individually take the risks likely to add up to a single breakthrough. Managers of firms that fail to attain their desired performance levels probably take greater risks, while managers who have achieved or slightly exceeded expectations will probably leave their risk strategy unchanged (Bromiley, 1991). Firms flush with recent success should also take greater risks, because they relax the control and scrutiny of inventor’s efforts (March & Simon, 1958). Risk preferences will also have indirect effects on inventive distributions. Because selection takes place at multiple – and ultimately social – levels of analysis, firms can harness the knowledge of groups of inventors to minimize the types of errors they collectively commit. As Sah and Stiglitz (1985) note, the proper mixing of hierarchies (in which every layer has to agree to fund a given project and which only an authoritative regime can install) and polyarchies (in which only one member needs to support a project for it to be funded) can reduce the probabilities of common decision errors.8 Thus, by installing a decision-making process that employs a mixture of hierarchies and polyarchies, a firm might be able to increase the weight in the right tail of its inventive distribution, or perhaps change the form of its distribution entirely. Up until this point we have focused most of our discussion on firms, even though two other entities have recently entered the inventive landscape: universities and open innovation communities. With passage of the Bayh–Dole Act in 1980, universities have begun to actively seek breakthrough inventions. Of course, a great deal of invention has always occurred in universities; what has changed is that universities now perceive the need to commercialize their intellectual property. While there is still controversy over the universities’ role as guardians of public science as opposed to widget peddlers (Thursby & Thursby, 2003), they have greatly increased their patenting, though it should be noted that some evidence indicates that their patenting bar has fallen and that less influential universities are attempting to patent as well. Regardless of the extrinsic motivations for patenting and technology transfer, universities provide an important collaborative context for the invention of breakthroughs, mainly because they maintain links to scientific research. As Rosenberg (1996) argued, breakthrough technological inventing is best done in conjunction with active science research. Science
348
LEE FLEMING AND MARK SZIGETY
can provide a map for the combinatorial search space (Fleming & Sorenson, 2004), predict fertile or dead-end outcomes before a combination is tried, and help explain unexpected results. Individual researchers will also attempt to maximize creativity when the incentive is first authorship of a publication in a top-ranked science journal. Such incentives become dulled in a corporate environment, particularly if the corporation does not encourage or allow external publication. University inventors are also, due to the primacy requirements of publication (Merton, 1948), much less likely to invest effort in refinement of a previous technology. University selection processes for invention quality are also less likely to be rigorous, due to inexperience, lack of technical expertise, and greed. All of these influences would increase the generative variance of the distribution of a university’s inventions, compared to that of a private firm. Continuing with a non-commercial focus, we consider when the output of a technological community will demonstrate lesser or greater variance (since the boundaries of communities and industries correlate strongly, the arguments probably hold for latter as well). Similar to scientific colleges but predating them by hundreds and possibly thousands of years, technological and industrial communities emerged whenever more than one inventor or craftsman began to specialize in a particular technology. They also existed and continue to exist without corporate approval, since engineers trade technical information and career gossip across corporate boundaries (von Hippel, 1988). Some technological and industrial communities are more dynamic and inventive than others, and while this tends to correlate with the ‘‘age’’ of the community, our arguments for the benefits of associative richness imply that age simply correlates with an unobserved causation. To motivate this line of reasoning, we first ask why some communities undergo rebirth. We propose that this is more likely to occur when community members import new components and work with personnel from other communities. For example, road bicycle racing technology progressed incrementally from the invention of the safety bicycle until the 1970s, when artisans and enthusiasts began to tinker with radical architectures, mountain biking components, and materials from the aerospace industry. Breakthroughs would therefore be most likely to emerge from the fertile boundaries between communities, where inventors could draw on experience and components from both (Bijker, 1987; Fleming, 2001; von Hippel, 2005). Such open communities would be much more likely to witness multiple breakthroughs and technological rebirth. If the outcomes of such fertile recombinant search followed a skewed distribution as we have argued, extreme outliers would not necessarily become less frequent over time. Only in
Exploring the Tail of Creativity
349
closed and insular communities would breakthroughs become less frequent over time, as the community exhausted its recombinant search space (Tushman & Anderson, 1986). Open source communities are the second new collective entity in technological invention. The rise of electronic networking technology has fostered contact among people who were previously constrained by physical proximity, corporate communication policies, or lacked computing and electronic networking resources. It remains to be seen, however, whether (or which) open innovation communities can invent breakthroughs. As individual reputation is gained only through personal contribution (just as in scientific communities), open innovation communities provide powerful incentives. Some communities, however (such as the most popular example, Linux), also rely heavily on modularity. This modularity, while enabling rapid incremental progress, constrains the combination of truly novel components and probably limits generative variance. (Indeed, the best contributions of open source communities are their open and high-powered debug and selection processes.) Other open innovation communities, such as the Internet Engineering Task Force (or Cambia), do not rely on modularity and have invented exceptionally fertile technologies, such as Internet protocols (Bradner, 1999) and genetic sequencing technologies (Broothaerts et al., 2005). Understanding how open innovation communities contribute to technological change and breakthroughs offers a rich area for further research. Table 2 summarizes our arguments at higher levels of analysis.
STRATEGIC IMPLICATIONS OF AN EVOLUTIONARY MODEL OF BREAKTHROUGHS Schumpeter was the first to consider the strategic implications of technological breakthroughs. While his colorful imagery of gales of destruction has earned him an exceptional number of citations over the years, he gave conflicting answers to the question of whether breakthroughs come from small entrepreneurial firms or from established R&D laboratories sponsored by large successful firms. We reframe the question as: What causes highly skewed outliers among prolific inventive distributions? We have thus bypassed the Schumpeterian contradiction by asking what factors influence the output distributions of the most prolific sources of inventions. As mentioned above, these sources can still be firms but increasingly include universities and open innovation communities. All of these collective efforts occur
350
LEE FLEMING AND MARK SZIGETY
Table 2. Summary of Predictions for the Second Moment of Creative Outcome Distributions for Organizational and Managerial Influences. Impact on Variance of Distribution of Iinventive Search Organizational variables Demographic diversity Insular technological communities Location at community boundary Work with previously explored component sets Managerial variables Extrinsic incentives (e.g., managerial pressure, financial incentives, punishment of failure) Personnel turnover Well-defined industry standards Slack resources MBA management Academic researcher’s affiliation with a top university Sloppy organizational patenting procedures (e.g., university) Organization located within region with low inventor mobility
+ + + + + +
within ecologies connected by personnel, technology, and information flows. We now discuss how a manager might profit in such a mixed environment. A firm’s chance of inventing a breakthrough remains roughly proportional to its amount of search. If we assume that this proportional argument holds at an industrial level of analysis, then firms will invent breakthroughs in rough correspondence to their proportion of search investment within the industry.9 This implies that even the largest organization in an industry will probably not be the source of a particular breakthrough. Consider, for example, a firm that does 20% of the search within a larger industrial and scientific community. All else being equal, that firm remains four times less likely to invent a breakthrough than the rest of the community (and this ignores sources of breakthroughs from outside the community). Hence, a firm is far more likely to need to respond to a breakthrough than it is to generate one. Managers (and more importantly, their inventors) must, therefore, devote considerable time to scan their environment. Fortunately, scanning improves the firm’s ability to create a breakthrough, because it identifies recent developments and external opportunities for recombination
351
Exploring the Tail of Creativity
Optimal mix of search activities
with internal technology (Cohen & Levinthal, 1990). Thus, the value of scanning may depend on the level of search activity within a firm, relative to the level of search activity outside of a firm. Given a firm’s budget constraint and R&D spending within its industry, the optimal mix of scanning for a given level of total search will vary. A firm devoting too much attention to internal search could miss the next external breakthrough. However, devoting too much attention to external scanning could undermine a firm’s absorptive capacity and thus diminish the productivity of all that scanning. These considerations motivate Fig. 2. As can be seen from the figure, when a firm accounts for a great percentage of its industry’s total search investment, it probably wants to limit its scanning activities. Alternatively, a small player likely wants to devote most of its limited resources to scanning external sources.10 Note that these propositions stand in contrast to models of invention, which posit that the dominating firms often have less incentive to fund R&D than smaller competitors. One can imagine that the convexity of our function increases or decreases with the difficulty of the industry’s inventive activity (Cohen & Levinthal, 1990), while the slope is never strictly positive.11 Note that the change in the optimal mix is not the result of a decrease in the value of a dollar spent on inventing; rather, it is the relative increase in importance of scanning as the probability of an external source of a breakthrough increases.12 This observation begs an important question: What factors influence the relative value of search across all levels of external
EXTERNAL SCANNING
INTERNAL SEARCH
Proportion of industry R&D spending outside the firm
Fig. 2.
Stylized Illustration of the Optimal Ratio of External to Internal Focus as a Function of Total Industry Investment in R&D.
352
LEE FLEMING AND MARK SZIGETY
scanning and thus shift the entire optimal mix line up or down? Certainly, the ease with which a breakthrough invention is copied or reverse engineered has an impact. If a breakthrough relies on highly originator-specific or tacit knowledge, then the relative value of scanning decreases. Likewise, if intellectual property laws offer strong protection (as in pharmaceuticals), then firms should allocate scarce resources to internal search. In the opposite case where breakthroughs are easily assimilated, the line may shift down (but never down to zero, as stipulated by absorptive capacity arguments). Thus, the optimal mix line’s movement up or down directly affects the total probability that an industry will invent a breakthrough in a given amount of time. Since these resource allocation decisions are conditional on a breakthrough being invented somewhere, there is a paradox: A high relative value of scanning makes an industry relatively dull compared to the level of inventiveness it could otherwise achieve (though a high relative value of scanning also decreases the possibility of duplicative search). Although the ultimate outcome appears consistent with classical economic models of innovation, the root cause is subtly different. Firms in our model do not decrease investment because they fear they will not be able to capture rents from invention, but because they recognize the wealth of creativity beyond their borders.13 These issues strike us as ripe for formal game-theoretic modeling and empirical validation. Scanning processes and strategies for identification of potential breakthroughs should occur early, before a potential breakthrough product hits the market. Scanning should be done by the technical professionals in their own areas of expertise and should involve technical and scientific literature, patent applications and grants, personal contacts, convention attendance, university ties, customer interaction, and reverse engineering. Unfortunately, inventive ecologies show no respect for industrial boundaries or competitive clusters. While pertinent breakthroughs are more likely to emerge from within the industry that can profit from them (per amount of inventive search), they can certainly emerge from outside the industry as well. For example, several computer firms, such as IBM, perform research in semiconductors – an industry in which they are less able to profit directly. Thus, firms must be cognizant of the total amount of money spent on pertinent research inside and outside competitively defined boundaries. Monitoring the pertinent scientific literatures should provide some defense against ‘‘outside’’ breakthroughs, since the literatures are organized by discipline and problems. This disconnect between source and application provides the basis for the classic argument that breakthroughs originate from ‘‘outside’’ an industry (Tushman & Anderson, 1986; Utterback, 1996).
Exploring the Tail of Creativity
353
A manager or researcher will be under constant pressure to prove the value of scanning, because it takes valuable time away from internal search. Fortunately, scanning has a variety of benefits beyond the early identification of breakthroughs. It enriches and improves the inventor’s own creativity. Because it tends to encompass market research, it gives inventors a better understanding of the customer. It enables the inventor to alert other inventors within the firm to potentially valuable technology. In many ways, scanning resembles membership in a scientific community, although scanning pays more attention to markets and technology than the discourse within a scientific community. Like scientists, inventors need to build, to maintain, or, at the very least, to be aware of external professional networks. Many inventors will not be able to understand the technical literature, so better educated inventors may need to adopt the ‘‘gatekeeper’’ role (Allen, 1977). To defend scanning against immediate deadline pressures, management might reserve a certain amount of time each week for scanning activities (Friday morning meetings, for example, perhaps followed by show-and-tell lunches). If a manager discovers a breakthrough outside her firm, she has three options. First, she can ignore the breakthrough and continue to refine her own research. This is probably not an effective strategy, although it may very well be a common one. Second, she can abandon her current technological efforts completely and rely on second sourcing and other marketing arrangements. Third, she can attempt to be a fast technological follower and either invent around or leapfrog over the breakthrough. We will discuss various criteria for making each of these choices and the idiosyncratic implementation issues that are likely to challenge the manager. A firm will be far more likely to ignore a potential breakthrough if its personnel lack background and experience in the area (Cohen & Levinthal, 1990). Breakthroughs remain difficult to identify because they are often technologically immature and rely on highly uncertain estimates of manufacturability and market potential. Firms with a background in the breakthrough technology and commercialization paths will probably see the potential in the rough initial combinations, though they will also need to overcome the inertia of internal politics, refined processes, and customer linkages (March & Simon, 1958; Hannan & Freeman, 1977; Christensen, 1997). Firms that have completely missed the breakthrough opportunity should consider withdrawing from the technological front-end in order to focus on second sourcing and marketing strategies or on exploiting their complementary assets (Teece, 1986). The third strategy would be to embrace the breakthrough – and go one better. Leapfrogging is a particularly viable
354
LEE FLEMING AND MARK SZIGETY
strategy if the firm already has a wide background in the constituent components. An organization that learns well (and can balance exploration and exploitation adroitly) may be uniquely positioned to use the breakthrough as a jumping-off point for a new field of research (Levinthal & March, 1993; Levitt & March, 1988). But this strategy should not proceed without a legal analysis of the breakthrough’s patent protection. Nor should it proceed when the breakthrough is of a cost-cutting nature; that is, when markets and customers will not change much. Breakthroughs go through many revisions between invention and market. They often prove difficult to manufacture and often require customers to search for new modes of use. New markets often emerge after this search and provide fertile opportunities for fast followers and new creative combinations. Thorough scanning provides the additional possibility of identifying a breakthrough that remains hidden from the inventing organization. In this case, managers might buy the intellectual property or hire the inventors, though care should obviously be exercised when expressing interest in hiring another firm’s employee, particularly in regions covered by non-competes or judges sympathetic to trade secrets law. Greater patenting and the development of intellectual property markets in general makes this strategy more viable since firms must make their discovery public in order to protect it. Better data analysis capabilities also make it possible to identify individual inventors. As such, scanners can identify and begin to map the social topology of their field, which can then be used to identify the most likely sources of future breakthroughs. Scanning firms can also hire more effectively, both for future breakthroughs and for development of current breakthroughs.
CONCLUSION Following a path worn by previous scholars of technology, we developed an evolutionary model of invention. Our contribution was to highlight the asymmetric distribution of inventive outcomes and focus on the highly skewed outliers. Recent econometric advances have made it possible to test predictions about the variance as well as the mean. We can therefore gain additional insight into breakthroughs by considering the observable implications of our arguments upon the second moment. Consistent with evolutionary epistemologies of creativity, we proposed that inventors create and discard numerous ideas in their recombinant search processes. For individuals, these processes result in highly right-skewed distributions of creative
355
Exploring the Tail of Creativity
output. Aggregation of these distributions results in similarly skewed outputs for firms, communities, and industries and is consistent with power law phenomena and with the common intuition that breakthroughs are extremely rare. We sketched a variety of testable arguments, which we hope will motivate additional research in the area. For ecological strategists, identification of the most likely sources of technological breakthroughs in an industry requires answering three questions. First, where among various firms, universities, and open innovation communities are inventors trying new combinations of components that might impact the field of interest? Second, what are the relative magnitudes of the inventive productivity of all these sources? Since the number of breakthroughs is roughly proportional to the number of inventions in general, sheer productivity provides the most likely indication of inventive breakthrough. Finally, in addition to the sources with the highest absolute magnitudes of inventive efforts, what sources demonstrate the most rightskewed distributions of inventive outcomes? These places are the most likely sources of a rare breakthrough outlier, conditional on their total inventive productivity. Having used these questions to identify a relevant set of possibilities, a manager must then focus her strategic efforts on a smaller number of inventive competitors. These strategic efforts should emphasize continual scanning of the identified sources in order to respond quickly and effectively to industry-changing, breakthrough inventions. Finally, while our analysis built on a psychological model of invention, it highlights the importance of an external and ecological view of invention. Even as managers strive for a breakthrough within their organization, they must maintain an awareness of their inventive ecology – for any particular firm is unlikely to be the eye of the next Schumpeterian storm.
NOTES 1. Recent thinking has changed drastically on this topic. Natural scientists now believe that gene transfer between species has been surprisingly commonplace, even before the development of genetic engineering (Smith, 1988). 2. It remains possible that some technology could be very valuable and yet remain undeveloped by future recombinant search. Forks, for example, have had similar functionality for many years. Yet even for that prosaic example, recent designs have incorporated different materials, such as grippy handles for toddlers. 3. Following the literatures on psychology and on the epistemology of knowledge, one could also include steps in a manufacturing process or bits of knowledge in the definition of component (Fleming & Sorenson, 2004).
356
LEE FLEMING AND MARK SZIGETY
4. Ignoring and obviously doing an injustice to the research on creativity, we only highlight a few ideas pertinent to our arguments. We refer the interested reader to Amabile (1988, 1996) and Simonton (1999) for general references. 5. While these models are promising, they remain relatively crude and in need of elaboration, for example, fixed effects and panel data models would be desirable, as would an exploration of the inherent violation of homoscedasticity. 6. For ease of mathematics, if we assume that the distribution of individual inventive outcomes follows an exponential, then the sum of exponential distributions also follows an exponential distribution (Pittman, 1993). 7. Ahuja and Lampert’s (2001) work on breakthroughs and unreported regressions supports the argument. A negative binomial regression of the number of breakthroughs (patents in the top first or fifth percentile in terms of forward citations of those patents issued in the same year t) yields a substantial and highly significant positive coefficient on the size of the firm (as measured by sales, R&D budget, employment, or total patents). 8. These errors come in two flavors: type I and type II. In our context, a type I error is committed when an inventing agent decides against pursuing an idea that would turn out to be a novel breakthrough, while a type II error is committed when an inventing agent decides to pursue an idea that turns out to be a waste of time. 9. Note that a firm’s total research investment ideally should include not only funds for R&D, but also funds for process improvements and other business activities related to invention. 10. A startup that has already identified a breakthrough should focus on exploiting the immediate opportunity. The budget constraint and the need for revenue loom large for these companies. 11. The function in Fig. 2 could easily have been piecewise linear (a line of negative slope followed by one of far less negative or zero slope); we are agnostic about the exact specification. 12. This model is consistent with and extends Chesbrough’s (2003) concept of open innovation by providing a quantifiable metric (ratio of external scanning to total industry funding) by which a firm can decide how much ‘‘openness’’ to pursue. 13. In fact, to the extent that the industry duplicates inventive effort, it is possible that a decrease in the level of inventive funding could actually bring the industry closer to the social optimum, at the very least in terms of the distribution of effort of the industry’s firms.
ACKNOWLEDGMENTS We would like to acknowledge the feedback of Francesco Gino, Richard Lai, Jose Lobo, Kwang-Hui Lim, Santiago Mingo, Dan Snow, the editing help of John Elder, and support from the Harvard Business School Department of Research. Errors and omissions remain ours.
Exploring the Tail of Creativity
357
REFERENCES Acs, Z., & Audretsch, D. B. (1988). Innovation in large and small firms: An empirical analysis. American Economic Review, 78, 678–690. Ahuja, G., & Lampert, C. (2001). Entrepreneurship in the large corporation: A longitudinal study of how established firms create breakthrough inventions. Strategic Management Journal, 22, 521–544. Allen, T. (1977). Managing the flow of technology. Cambridge, MA: MIT Press. Amabile, T. (1988). A model of creativity and innovation in organizations. Research in Organizational Behavior, 10, 123–167. Amabile, T. (1996). Creativity in context. Boulder, CO: Westview Press. Bak, P., & Chen, K. (1991). Self-organized criticality. Scientific American, 264, 46–53. Basalla, G. (1988). The evolution of technology. Cambridge: Cambridge University Press. Bigelow, L. S. (2006). Technology choice, transaction alignment and survival: The impact of sub-population organization structure. In: J. A. C. Baum, S. D. Dobrev, & A. van Witteloostuijn (Eds), Advances in strategic management: Vol. 23. Ecology and strategy. Oxford, UK: JAI. Bijker, W. (1987). The social construction of Bakelite: Toward a theory of invention. In: W. Bijker, T. Hughes, & T. Pinch (Eds), The social construction of technological systems: New directions in the sociology and history of technology (pp. 159–190). Cambridge, MA: MIT Press. Boone, C., van Olffen, W., van Witteloostuijn, A., & De Brabander, B. (2004). The genesis of top management team diversity: Selective turnover among top management teams in the Dutch newspaper publisher industry (1970–1994). Academy of Management Journal, 47(5), 633–656. Bradner, S. (1999). The Internet engineering task force. In: C. DiBona, S. Ockman, & M. Stone (Eds), Open sources: Voices from the open source revolution (pp. 47–52). Sebastapol, CA: O’Reilly. Bromiley, P. (1991). Testing a causal model of corporate risk taking and performance. Academy of Management Journal, 34(1), 37–59. Broothaerts, W., Mitchell, H. J., Weir, B., Kaines, S., Smith, L. M. A., Yang, W., Mayer, J. E., Roa-Rodriguez, C., & Jefferson, R. A. (2005). Gene transfer to plants by diverse species of bacteria. Nature, 433, 629–633. Burt, R. (2004). Structural holes and good ideas. American Journal of Sociology, 110(2), 349– 399. Campbell, D. (1960). Blind variation and selective retention in creative thought as in other knowledge processes. Psychological Review, 67(6), 380–400. Chesbrough, H. (2003). Open innovation: The new imperative for creating and profiting from technology. Boston: Harvard Business School Press. Christensen, C. (1997). The innovator’s dilemma: When new technologies cause great firms to fail. Boston: Harvard Business School Press. Cohen, W., & Levinthal, D. (1990). Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35, 128–152. Darwin, C. (1859). On the origin of the species by means of natural selection. London: J. Murray. Fleming, L. (2001). Recombinant uncertainty in technological search. Management Science, 47, 117–132.
358
LEE FLEMING AND MARK SZIGETY
Fleming, L. (2002). Finding the organizational sources of breakthrough invention: The story of Hewlett Packard’s ink jet invention. Industrial and Corporate Change, 11(5), 1059–1084. Fleming, L., & Sorenson, O. (2004). Science as a map in technological search. Strategic Management Journal, 25, 909–928. Gilfillan, S. (1935). Inventing the ship. Chicago: Follett. Hannan, M. T., & Freeman, J. (1977). The population ecology of organizations. American Journal of Sociology, 82, 829–864. Hull, D. (1988). Science as a process. Chicago: University of Chicago Press. Kauffman, S., Lobo, J., & Macready, W. (2000). Optimal search distance on a technology landscape. Journal of Economic Behavior and Organization, 43, 141–166. King, G. (1989). Variance specifications in event count models: From restrictive assumptions to a generalized estimator. American Journal of Political Science, 33(3), 762–784. Kroeber, A. (1948). Anthropology. New York: Harcourt Brace Jovanovich. Levinthal, D. A., & March, J. (1993). The myopia of learning. Strategic Management Journal, 14, 95–112. Levitt, B., & March, J. (1988). Organizational learning. Annual Review of Sociology, 14, 319–340. March, J. (1991). Exploration and exploitation in organizational learning. Organization Science, 2, 71–87. March, J., & Simon, H. (1958). Organizations. Cambridge, MA: Blackwell. McKelvey, B. (1997). Quasi-natural organization science. Organization Science, 8(4), 352–380. McKelvey, B. (1999). Toward a Campbellian realist organization science. In: J. A. C. Baum, & B. McKelvey (Eds), Variations in organization science: In honor of Donald T. Campbell. Thousand Oaks, CA: Sage. McKelvey, B., & Baum, J. (1999). Donald T. Campbell’s evolving influence on organization science. In: J. A. C. Baum, & B. McKelvey (Eds), Variations in organization science: In honor of Donald T. Campbell. Thousand Oaks, CA: Sage. Merton, R. K. (1948). Social theory and social structure. In: The bearing of empirical research on sociological theory (pp. 156–171). New York: Free Press. Nelson, R. R., & Winter, S. G. (1982). An evolutionary theory of economic change. Cambridge, MA: Belknap Press. Obstfeld, D. (2005). Social networks, the tertius iungens orientation, and involvement in innovation. Administrative Science Quarterly, 50, 100–130. Pinch, T., & Bijker, W. (1987). The social construction of facts and artifacts. In: W. Bijker, T. Hughes, & T. Pinch (Eds), The social construction of technological systems: New directions in the sociology and history of technology (pp. 17–50). Cambridge, MA: MIT Press. Pittman, J. (1993). Probability. New York: Springer. Reagans, R., Zuckerman, E., & McEvily, B. (2004). How to make the team: Social networks vs. demography as criteria for designing effective teams. Administrative Science Quarterly, 49, 101–133. Rosenberg, N. (1996). Uncertainty and technological change. In: R. Landau, T. Taylor, & G. Wright (Eds), The mosaic of economic growth. Stanford, CA: Stanford University Press. Sah, R. K., & Stiglitz, J. (1985). Human fallibility and economic organization. American Economic Review, 75(2), 292–297. Schumpeter, J. (1939). Business cycles. New York: McGraw-Hill. Simonton, D. (1997). Creative productivity: A predictive and explanatory model of career trajectories and landmarks. Psychological Review, 104(1), 66–89.
Exploring the Tail of Creativity
359
Simonton, D. (1999). Origins of genius: Darwinian perspectives on creativity. New York: Oxford University Press. Smith, G. (1988). Homologous recombination in prokaryotes. Microbiological Reviews, 52, 1–28. Teece, D. J. (1986). Profiting from technological innovation: Implications for integration, collaboration, licensing, and public policy. Research Policy, 15, 285–305. Thursby, J., & Thursby, M. (2003). University licensing and the Bayh–Dole Act. Science, 301, 1052. Tushman, M., & Anderson, P. (1986). Technological discontinuities and organizational environments. Administrative Science Quarterly, 31, 439–465. Tushman, M., & Moore, W. (Eds). (1980). Readings in the management of innovation (pp. 148–163). Cambridge, MA: Ballinger. Utterback, J. (1996). Mastering the dynamics of innovation. Boston: Harvard Business School Press. von Hippel, E. (1988). The sources of innovation. New York: Oxford University Press. von Hippel, E. (2005). Democratizing innovation. Cambridge. MA: MIT Press. Weesie, J. (1999). Regression analysis with multiplicative heteroscedasticity. Stata Technical Bulletin, 42, 28–32. Weick, K. (1979). The social psychology of organizing. New York: McGraw-Hill. Wheelwright, S., & Clark, K. (1992). Revolutionizing product development. New York: Free Press.
This page intentionally left blank
360
PART VI: COMPETITIVE STRATEGY
361
362
THE COMPETITIVE DYNAMICS OF VERTICAL INTEGRATION: EVIDENCE FROM U.S. MOTION PICTURE PRODUCERS, 1912–1970 Giacomo Negro and Olav Sorenson ABSTRACT We investigate the competitive consequence of vertical integration on organizational performance using a comprehensive dataset of U.S. motion picture production companies, which includes information on their vertical scope and competitive overlaps. Vertical integration appears to change the dynamics of competition in two ways: (i) it buffers the vertically integrated firms from environmental dependence and (ii) it intensifies competition among non-integrated organizations. In contrast to the existing literature, our results suggest that vertical integration has implications well beyond both the level of the individual transaction and even the internal efficiency of the integrated firm.
Ecology and Strategy Advances in Strategic Management, Volume 23, 363–398 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23012-2
363
364
GIACOMO NEGRO AND OLAV SORENSON
INTRODUCTION Organizational ecology’s contributions to strategy have been at least threefold. First and foremost, organizational ecologists, sometimes also known as corporate demographers, have drawn attention to the positive and negative interactions among populations of firms drawing on the same resources (e.g., raw materials, types of employees and customers), investigating not just how these forces influence organizations but also how they change over an industry’s life cycle. Second, this perspective has emphasized the importance of demographic rates, primarily the births and deaths of firms, as measures of industry vitality. And finally, organizational ecologists have brought a truly enormous amount of data into the picture to answer these questions, avoiding many of the methodological problems inherent in crosssectional studies or in small samples of unrepresentative firms (e.g., the Fortune 500). Although a large number of questions have fallen under the purview of ecologists (as one can see in this volume), much open territory remains. Consider the issue of vertical integration – that is, the degree to which a firm encompasses two or more activities where output from one serves as input for another. By and large, the analysis of vertical integration in the strategy literature has relied on the (efficiency) logic of transaction cost economics (TCE) (Williamson, 1975; for exceptions, see Pfeffer & Salancik, 1978; Sorenson, 2003). This perspective argues that firms should (and do) integrate vertically when the need for specialized investments coupled with the tendency for independent actors to behave opportunistically renders prohibitively expensive the writing of a contract. Though this logic yields many useful insights, it also leaves us with a relatively limited understanding of the consequences of vertical integration because it focuses almost exclusively on factors internal to the firm, or, even more narrowly, the transaction. Here, we apply an ecologist’s lens to the question. This perspective leads us to ask not how vertical integration influences the efficiency of the transaction itself, but rather, how does it alter the interactions between organizations in an industry? Our primary claim is that vertical integration offers organizations privileged access to resources, buffering them from the negative effects of competition while simultaneously intensifying competition among non-integrated rivals. We argue that integration guarantees organizations access to a larger share of the resources required for operation. In essence, when facing constraints in the scale of a production activity, integrated firms first supply the internal demand for these goods and services before they offer it to other potential buyers. As a result, integrated firms
The Competitive Dynamics of Vertical Integration
365
face a lower risk either of an interruption in production or of slowed growth as a result of an inability to obtain internally produced resources. The intensity of competition among non-integrated firms simultaneously rises because these firms must compete over the portion of resources not locked up in the integrated producers. These processes may also influence the relative positioning of integrated and non-integrated firms. In particular, we suspect that the protection from competition afforded by vertical integration might allow and encourage integrated firms to exploit broader niches. To corroborate these speculations, we analyzed an original dataset of motion picture production companies in the United States from 1912 to 1970. Our analysis of the exit rates of these firms revealed several interesting results: (i) integrated production companies exit the population at a lower rate than non-integrated firms, even after accounting for differences in scale; (ii) integrated production companies exhibit less susceptibility to diffuse competition both from other integrated firms and from non-integrated firms; (iii) integrated production companies also appear unaffected by localized competition with non-integrated firms; and (iv) non-integrated production companies exhibit greater susceptibility to competition from integrated producers than from other non-integrated firms. Integration also appears to limit the negative consequences of broad scope, thereby allowing integrated firms to maintain more variety in their product lines. In contrast to the dominant antitrust logic, then, our results suggest that vertical integration has a positive effect – the provision of greater product variety – that may offset its negative effect on pricing (through the exercise of market power).
THE CONSEQUENCES OF VERTICAL INTEGRATION One of the most important questions in both the management and study of organizations has been: What is the appropriate scope of the firm (i.e., Where should the firm draw its boundaries)? The critical factor in answering this question is an understanding of how changes in firm boundaries (scope) alter the behavior and performance of organizations. Scope varies along multiple dimensions – geographic, horizontal and vertical – all of which have important consequences for firm performance and industry dynamics. We nonetheless focus here on a single dimension: vertical scope. Researchers typically label expansions in vertical scope as vertical integration. In particular, vertical integration refers to situations where an organization encompasses two or more stages of a production process. In other words, cases in which a firm produces an output that
366
GIACOMO NEGRO AND OLAV SORENSON
becomes an input to another segment of the firm’s operations. Examples of vertical integration would include automobile manufacturers that either produce the engines used in their cars or sell cars to consumers through their own dealerships, and soda manufacturers that also bottle and distribute their soft drinks. In our context, vertical integration refers to those motion picture production companies that also distribute their own films. Researchers have devoted substantial attention to understanding numerous features of the processes related to vertical scope extension. One might cluster this research into two streams: (i) the reasons why firms decide to integrate vertically and (ii) the consequences of being integrated (for recent reviews of theory and empirical studies, see Joskow, 2005: Klein, 2005). When separated along these dimensions, the former question has received a great deal more attention than the latter. Such a characterization, however, strikes us as misleading. To the extent that efficiency logic has dominated the analysis of the antecedents of vertical integration, the two issues have been conflated: These perspectives contend that firms integrate vertically when doing so would improve their economic performance. Hence, the two issues become one and the same.
Efficiency Logics Perhaps the most prominent approach to understanding the function of vertical integration is TCE. TCE maintains that engaging in transactions exposes actors to a series of potential costs, contractual and organizational hazards, that depend on the nature of the transactions themselves (Williamson, 1971, 1975). Examples include the cost of writing complete contracts covering all contingencies, monitoring performance following the contract, and bargaining over unexpected events. These ex ante and ex post costs vary as a function of several factors including the uncertainty, frequency and complexity of the transactions, but asset specificity plays the central role. Asset specificity refers to the degree to which sunk investments have alternative uses outside of the existing buyer–seller relationship. Specific investments have few alternative uses. This inflexibility provides an incentive for the party not making these investments to behave opportunistically, attempting to renegotiate terms to their advantage once the investments have been made. As a result, parties facing such risks tend to underinvest in specific investments in the first place. In these circumstances, vertical integration therefore often represents a superior solution for organizing transactions.
The Competitive Dynamics of Vertical Integration
367
Whereas TCE underscores the relevance of ex post costs and their connection to ex ante investment incentives in contractual arrangements, property rights theory highlights the inefficiencies that can arise when split ownership produces a misalignment of incentives (Grossman & Hart, 1986; Hart & Moore, 1990). Again, the argument turns on the value of specific assets; non-integrated firms tend to underinvest in specific assets when the costs of developing them would fall unduly on one party. Joint ownership avoids underinvestment in these specific assets by aligning the incentives of the two parties. Though pointing to a different mechanism, like TCE, property rights theory focuses on the efficiency of the transaction as the consequence (and also the antecedent) of vertical integration. The empirical support for these efficiency logics remains somewhat thin. The bulk of research to date corroborating the TCE perspective has involved cross-sectional correlations between asset specificity and vertical integration. Although these studies overwhelmingly find positive correlations, their cross-sectional nature does not allow one to discern whether firms integrated because of the perceived value of developing specific assets or because integration in an earlier period shifted the incentives for future specific investments. Moreover, these studies do not really consider whether integration improved performance. Rumelt (1974), in fact, finds that vertically integrated firms underperform non-integrated firms, a result that seems difficult to reconcile with the notion that these decisions have been made to promote economic efficiency. The most convincing evidence for an efficiency logic comes from a study of the for-hire trucking industry (Silverman, Nickerson, & Freeman, 1997; Nickerson & Silverman, 2003). It found that companies employing an ‘‘inappropriate’’ governance of certain labor and capital market transactions experienced higher failure rates, though these effects appeared weak relative to the importance of age, size and density dependence on failure rates. Adaptation Although the idea that governance arrangements influence the efficiency of transactions has received substantial attention the effects of vertical integration on other organizational features and outcomes has been less extensively explored. For instance, the focus on specific investments does not account for the fact that while vertically integrated structures may reduce opportunism, they also impose structural costs and constraints. Williamson (1975), for example, asserted that vertical integration represents an inferior strategy for obtaining, processing and employing various types of information, such as
368
GIACOMO NEGRO AND OLAV SORENSON
price/cost structures and technical evaluations. These disadvantages, however, would accrue to the firm as a whole rather than to the focal transaction. A more dynamic perspective on the effects of vertical integration has been developed by Sorenson (1997, 2003). He argued that vertical integration engenders interdependence among a firm’s activities and that this interdependence has countervailing short- and long-term effects. In the short term, integrated firms can often benefit from interdependence through synergy by producing goods of higher quality or with novel features. They can also eliminate the search costs associated with locating external exchange partners because they rely on internal resources to manage their transactions. Indeed, Sorenson’s empirical analysis of the exit rates of computer workstation manufacturers from 1980 to 1996 found that firms integrated into the production of components enjoyed an approximately 33 percent lower hazard of exit than non-integrated producers. In the longer term though, vertical integration can become a disadvantage because interdependence limits the rate of organizational learning. Organizational knowledge resides in routines, and increasing levels of interdependence, such as those found in integrated firms, tend to obscure specific cause–effect relationships. Integration therefore stymies the identification of effective routines. Moreover, even when firms do discover more effective procedures, interdependence also increases the costs and difficulties associated with implementing them because the interactions between operations typically produce a cascading series of unintended consequences within the firm in response to any change. As a result, integrated firms benefit less from learning through cumulative experience. Consistent with this logic, Sorenson (1997, 2003) found that a history of vertical integration increases the exit rates of workstation producers. Unlike efficiency perspectives, this approach does not assume that firms integrate to maximize performance. Indeed, defining optimal performance in any dynamic setting is difficult because it depends crucially on the discount rate. Firms (or managers) that discount the future heavily prefer to integrate for short-term gains, but those with more distant horizons remain unintegrated to maximize the future benefits of learning. On the other hand, this perspective, like the efficiency-based accounts, focuses on the effects of vertical integration on the internal operations of the firm. Competitive Interaction Although a great deal of research has considered the effect of integration on internal processes, relatively less attention has been given to how vertical
The Competitive Dynamics of Vertical Integration
369
integration might influence processes external to the organization. A clear exception (discussed below), however, appears in the theories expounded both in antitrust case law and in an accompanying (largely game theoretic) literature in economics. To address the question of how integration affects processes external to the organization, we adopt an ecological perspective. Organizational ecologists have highlighted the importance of two processes – legitimation and competition – in industry evolution. Our principle insight with respect to this perspective is that integrated firms differ from nonintegrated firms because integration alters the nature of competition for scarce resources. The density dependence model relates the processes of legitimation and competition to changes in population vital rates. By definition, all firms within a population could draw on the same set of resources to sustain their survival. Legitimation increases the viability of firms by easing access to these resources, while competition for limited resources reduces firm viability. By assuming that these processes both relate to the number of organizations in a population (i.e., its density) – legitimation increasing linearly (or less than linearly) with density while competition increases as a function of its square – organizational ecologists produce an expectation of a non-monotonic relationship between population density and vital rates. Specifically, entry into the population should follow an inverse U-shaped curve, first increasing and then decreasing with greater population density, and exit from the population should follow a U-shaped function. Dozens of studies have found empirical support for these expectations (Carroll & Hannan, 2000). How then does vertical integration modify competition and competitive interaction? Increased vertical scope reduces an organization’s dependence on the external environment (Thompson, 1967; Pfeffer & Salancik, 1978). Consider a case of forward integration, where a firm engaged in manufacturing activities extends its scope into distribution. As a producer, the firm in question can better control schedules and smooth deliveries by integrating into distribution. As a distributor, the firm foregoes the costs of coordinating operations, monitoring the market and responding to unstable supply. This forward integration has two effects. On the one hand, internally, it stabilizes inputs for the distribution side of the business and secures demand for the manufacturing side, thereby mitigating some of the uncertainty typically associated with these activities. On the other, externally, it reduces the supply of goods available to non-integrated distributors and the availability of distribution outlets to non-integrated manufacturers. Even when integrated firms do seek to access resources from outside the firm, they likely enjoy an advantage relative to non-integrated rivals. To the
370
GIACOMO NEGRO AND OLAV SORENSON
extent that vertical integration reduces the uncertainty inherent in the coordination of production processes across firms, external parties perceive less risk in dealing with integrated firms and hence prefer them in negotiations. For example, an actor knows that an integrated production company will almost certainly distribute to theaters any film it produces, while an independent producer on the other hand may fail to negotiate a distribution agreement. To the extent that the actor’s compensation depends on the revenue it produces, he would then have a higher expected value for the film being produced by the integrated company. And even if his compensation did not depend on box office performance, he would likely prefer the film with guaranteed distribution as its greater likely visibility increases his odds of being considered for roles in future projects. Owing to this privileged access to external resources, vertical integration might reduce the volume of resources available to non-integrated firms by an even greater amount than the sum of resources internalized by integrated firms. We see these processes potentially influencing density dependence in a variety of ways. At the most basic level, integrated firms should exhibit less sensitivity to diffuse competition. Because integrated firms have internalized a portion of the resources they require, they should compete less intensely both against non-integrated firms and vis-a`-vis other integrated firms. This process should enhance the survival chances of integrated firms, particularly as the competition for resources associated with increasing density becomes more intense. Hypothesis 1. Vertical integration reduces the positive effect of density on exit rates. A more nuanced picture, however, arises from considering the potential differential effects of density both within and across sub-populations. Here, we expect that vertical integration would engender asymmetric competition. First consider the effects of the density of integrated firms. Integrated firms should exert more competitive pressure on non-integrated firms than on other integrated firms. This effect reflects the advantageous position of integrated firms relative to non-integrated firms with respect to accessing external resources. On the other hand, non-integrated firms should compete less intensely with integrated firms than with their non-integrated peers. Two factors underlie this effect. First, the internalization of inputs among integrated firms reduces the number of fronts on which they must compete with non-integrated firms for critical resources. Second, even when the two sub-populations do vie for resources, integrated firms enjoy an advantageous position compared to non-integrated firms.
The Competitive Dynamics of Vertical Integration
371
Although vertical integration may introduce asymmetry into the strength of competitive pressures across sub-populations, it is less clear whether these groups should differ in terms of density dependence within their subpopulations. Legitimation likely operates within each sub-population to some degree. On the one hand, both integrated and non-integrated firms might benefit mutually from the joint legitimation of the enterprise. The prevalence of each particular sub-form may nonetheless help to establish the legitimacy of a particular vertical scope of operations. With respect to competition, however, one might suspect that vertically integrated firms would compete less intensely with other integrated firms than non-integrated firms do with others of similar vertical scope because non-integrated firms compete on a day-to-day basis over a much larger range of the resources they require to sustain their operations. Hypothesis 2a. Integrated producer density increases the exit rates of non-integrated producers more than non-integrated producer density. Hypothesis 2b. Non-integrated producer density increases the exit rates of integrated producers less than integrated producer density. A more fine-grained consideration of competitive intensity yields additional evidence for the effect of vertical integration on competitive dynamics. Although the density dependence model assumes that all firms interact with one another, in practice firms typically respond more directly to interactions occurring within specific subsets of the resource space (Hannan & Freeman, 1977; McPherson, 1983; Baum & Singh, 1994). One observes this general insight across a variety of dimensions: Organizations of similar size compete more intensely with each other than with firms of either larger or smaller scale (Baum & Mezias, 1992). Firms overlapping to a greater extent in the nature of the services they provide experience higher failure rates (Baum & Singh, 1994; Dobrev, Kim, & Hannan, 2001). Organizations that recruit from the same population of potential employees exhibit greater interdependence in their vital rates (Sørensen, 1999, 2004). And geographic proximity, which increases the degree of overlap both in terms of inputs and in terms of potential buyers, dramatically intensifies competition (Sorenson & Audia, 2000; Stuart & Sorenson, 2003). Just as we expect vertical integration to modify the nature of density dependence across and within integrated and non-integrated firms, we also anticipate that it should alter the effects of direct overlap in organizational niches. We argued above that integrated firms should exhibit less sensitivity to diffuse competition over inputs and/or outputs compared to non-integrated
372
GIACOMO NEGRO AND OLAV SORENSON
firms due to benefits from internalization and asymmetric positions. Here we expect that increasing overlap hurts all firms, but that overlap between integrated firms and non-integrated firms should primarily disadvantage the non-integrated firms. Two mechanisms produce this expectation. First, integrated firms depend less on the resources they share with non-integrated firms even when their niches overlap substantially. Second, they also have an advantage in accessing these resources when they compete for them against nonintegrated firms. Hypothesis 3a. Competitive overlaps with integrated producers increase the exit rates of non-integrated producers more than competitive overlaps with non-integrated producers. Hypothesis 3b. Competitive overlaps with non-integrated producers increase the exit rates of integrated producers less than competitive overlaps with integrated producers. These insights parallel closely those garnered from game theoretic models of vertical integration (cf. Krattenmaker & Salop, 1986; Salinger, 1988; Ordover, Saloner, & Salop, 1990). In general, this literature demonstrates that firms can often increase their profitability by integrating vertically. These benefits in turn typically come at the expense of rivals that did not integrate. Non-integrated firms find themselves paying more for critical inputs, thereby reducing their profitability. To the extent that exit rates mirror profits then this stream of research would appear to suggest that vertical integration would reduce the exit rate of firms that integrate and raise the exit rate of non-integrated firms. We nonetheless have built our theory independent of this literature because several factors raise questions regarding the applicability of these analytical models to our setting. First, these models have been limited to markets with very small (and fixed) numbers of firms (often only two in any one stage). By contrast the motion picture industry has hundreds of active firms through most of our window of study. Though the intuitions of these game theoretic models may extend to populations with large numbers of firms, this robustness has not yet been demonstrated. Moreover, since these models always assume a fixed population of producers, it has been silent with regard to the competitive effects of population density. Second, integrated firms benefit in these models through their ability to manipulate prices. But in the film industry, production and distribution companies almost always simply split the revenue from a particular firm, to share the risks associated with it. As a result, the transfer price is determined ex post.
The Competitive Dynamics of Vertical Integration
373
Similarly, theaters charge consumers uniform prices across films. Price therefore does not appear to offer a meaningful mechanism for gaining a competitive advantage in this context.
THE MOTION PICTURE INDUSTRY IN THE UNITED STATES We analyze whether vertical integration influences the relationship between competitive interaction and organizational performance by studying the population of feature film production companies in the United States from 1912 to 1970. The film industry offers a particularly appropriate context for studying the effects of vertical integration on competitive interaction for four main reasons. First, a clear sequential interdependence of activities links the production and distribution of films: production companies assemble creative and technical inputs to make a film and then license the negative print to a distributor, which proceeds to market the right to use positive prints of a film to exhibitors. Second, this sequential interdependence nevertheless does not require either party to sink a specific investment into the transaction. On the contrary, production companies can consider distributors fungible because they do not offer differentiated services (Conant, 1960). As a result, we can separate the competitive effects of integration from the competitive consequences of differentiation that might arise from specific investments. A third feature making the motion picture industry amenable to the study of vertical integration is that we can trace every exchange between production and distribution companies, thereby allowing us to develop a far more nuanced picture than the typical study of vertical integration. Fourth, the geographic concentration of the film industry in Hollywood allows us to isolate vertical integration from other relevant scope dimensions such as localization. The feature film industry in the United States began in 1912, following the appearance on screen of longer films imported from Europe. Feature-length films (usually defined as longer than four reels, where each reel runs 1,000 feet or about 10 min) altered the nature of production because they involved multiple reels, thereby requiring more articulate production processes and higher levels of capital investment than the ‘‘shorts’’ that preceded them (Bordwell, Staiger, & Thompson, 1985). Distribution and exhibition also became more complex businesses as the industry transformed into one of differentiated products. Production companies had sold shorts by the foot, implicitly considering a minute of film a commodity, but bids across films
374
GIACOMO NEGRO AND OLAV SORENSON
began to diverge with the advent of these longer motion pictures. In fact, historians have even interpreted the emergence of the star system as a means of signaling audiences to alert them to the qualitative differences across offerings (Kerr, 1990). Feature films enjoyed quick success. In 1914, roughly 14,000 exhibition outlets in the United States played feature films more or less exclusively (Lewis, 1933). Our data indicate that in the same year, 121 production companies released 327 films. Consumption continued to grow in the following years: in 1921, theaters sold roughly 40 million tickets each week, and the number continued to grow throughout the decade (Lewis, 1933). The fact that feature films could build on the established demand for short film entertainment almost certainly contributed to its rapid adoption. Despite this moderate substitutability on the demand side, the dramatic differences in the organization and process of production meant that new populations of firms arose to meet this demand; very few producers of shorts successfully transitioned to producing feature length films (Mezias & Boyle, 2005). During this same period, the number of production companies and films, on the other hand, declined: 244 production companies released close to 650 films in 1921 down from a peak of 926 in 1917. Fig. 1 shows the historical density of production companies over our observation period (1912–1970). In the drive to differentiate their films, production companies created longer and more expensive films that spent increasingly long periods in the theaters. These rising costs erected barriers to entry for those interested in producing movies, which may account in part for the diverging patterns of demand and density. Throughout the industry’s history, production companies have varied in their choice of whether or not to integrate into distribution (see Fig. 2). When theaters first began exhibiting shorts to audiences, they would contract directly with production companies to purchase the films that they played. Hence, in this early stage, all companies essentially engaged in both production and distribution. Independent distribution companies emerged out of a recognition of inefficiencies in this system. Films typically had a limited life in any given theater; audiences demanded novelty. As a result, rental companies, the precursors of distributors, arose to rent shorts to theaters so that they could share content acquisition costs. As the feature film industry evolved toward more differentiated products, these companies expanded the scope of their activities to include not just the purchase and rental of film reels, but also the marketing of motion pictures and eventually the copying of the negative. In some cases, integration also extended into exhibition. During the second half of the 1920s, a group of integrated firms, called the ‘‘Majors’’ and
The Competitive Dynamics of Vertical Integration
375
Density of feature film producing companies
350
300
25
20
15
10
50
0 1912 1916 1920 1924 1928 1932 1936 1940 1944 1948 1952 1956 1960 1964 1968 1914 1918 1922 1926 1930 1934 1938 1942 1946 1950 1954 1958 1962 1966 1970
Year
Fig. 1.
Density of Feature Film Production Companies.
composed of Paramount, Loew’s-MGM, Fox, Warner Bros and RKO emerged and became dominant players in the industry (Balio, 1993). When the Great Depression struck, theater attendance declined by more than 30 percent, forcing more than 4,000 theaters to close in the space of three years, dramatically increasing the concentration of ownership. The production and distribution of films also became increasingly concentrated. During the 1930s, the Majors produced and distributed more than half of all domestic features. A niche nonetheless remained for independent production companies because none of the Majors had sufficient production capacity to meet the demand of its own theater circuit. Integrated firms therefore depended on a residual external supply of feature films from other Majors and independent producers. This organization remained in place well into the 1940s. In 1948, a decision by the Supreme Court in the United States v. Paramount Pictures antitrust case found the eight largest organizations guilty of restricting competition in the market for exhibition, and forced them to discontinue several practices that the Court considered illegal. In addition, the decision forced the separation of the five Majors into production/distribution businesses on the one side, and exhibition on the other side. Following this
376
GIACOMO NEGRO AND OLAV SORENSON
Density of integrated feature film producing companies
50
40
30
20
10
0 1912 1916 1920 1924 1928 1932 1936 1940 1944 1948 1952 1956 1960 1964 1968 1914 1918 1922 1926 1930 1934 1938 1942 1946 1950 1954 1958 1962 1966 1970
Year
Fig. 2.
Density of Vertically Integrated Feature Film Production Companies.
decision, the Majors reduced their production of films by more than 50 percent and concentrated increasingly on distribution and film financing (Christopherson & Storper, 1989). In their wake, the industry experienced a proliferation of independent production and distribution companies. From 1946 to 1970, the demand for motion pictures declined. Attendance fell rapidly from 98 million tickets per week to 65 million in 1950 and to 44 million in 1955. Radio and more significantly television began to compete with motion pictures for consumers’ leisure time, gradually eroding film attendance. After 1955, the number of tickets sold continued to decline gradually until the early 1970s.
ANALYSIS OF PRODUCER EXIT RATES Data We analyzed an original database that includes all motion picture production companies in the United States. We began the observation period in
The Competitive Dynamics of Vertical Integration
377
1912, the year of release of the first American-produced feature film, and ended it in 1970, the last year covered by the primary data source. We reconstructed the life histories of production companies through the release dates of the films they made. Production companies enter the population with the release of their first film, and exit it the first day following the release date of their last film (Mezias & Mezias, 2000, adopt a similar strategy). As with other corporate demography studies that use product-level information to time entry, these data do not account for the duration of preproduction processes. The short period of time necessary to produce a film, however, should limit the importance of these pre-production periods in industry dynamics. On average, production companies can complete a film in 6–18 months. The American Film Institute Catalog of Motion Pictures (AFI, 1989–1999) serves as the primary data source. This directory comprises reviews of all motion pictures distributed in the U.S. between 1893 and 1970 and provides detailed information on each film – including the production company, distributor, release date, length and genre. In addition to this primary source, we also collected industry-level statistics and supplemental information from The Motion Picture Year Book, the Motion Picture Almanac and Moving Picture World, a trade journal. Although experts generally consider the AFI catalog the most complete and comprehensive source on the film industry (Mezias & Mezias, 2000), it has one shortcoming: It lists feature films released in the periods 1911–1950 and 1960–1970, but has not yet documented the period from 1951 to 1960. To fill this missing window, we consulted several additional sources: (i) A.G. Ferrow’s filmography covering film production in the U.S. in the 1950s, which reviews 3,069 movies (Ferrow, 1999); (ii) the Motion Picture Catalog of the Library of Congress, which provides a list of all films that received copyright protection from 1950 to 1959 along with their respective production and distribution companies (Copyright Office, 1951–1971); and (iii) The Motion Picture Guide 1927–1982, a 12-volume reference set edited by J.R. Nash and S.R. Ross (1983) that provides comparable information. We restricted the population of production companies in several ways. Films produced and released for non-commercial purposes, such as those commissioned by government agencies do not enter the data. We also excluded imported films as these come from non-U.S. based production companies. In cases of international coproduction, we included only those films in which an American producer was the majority stakeholder. Finally, we excluded films from the late 1960s that provided no information on genre. An examination of the titles and casts suggests that the majority of these
378
GIACOMO NEGRO AND OLAV SORENSON
films contained pornographic content or represented unauthorized re-releases of earlier material. The final dataset consists of life histories for 4,089 production companies. Measures Our dependent variable is exit from the motion picture industry. Exit offers an attractive means of estimating performance for at least two reasons. First, it is less susceptible to manipulation than accounting-based measures and therefore provides a more reliable measure of performance. Second, production companies in most cases are not required to, and do not, report financial information. When available, moreover, firms typically aggregate the financial information of production activities together with nonproduction activities, making it impossible to isolate the profitability of the business of interest. Firms can exit industries in many different ways, including disbanding, bankruptcy, merger, acquisition, etc. For this study we could trace many instances of mergers, but only in cases where the merged entity produced a film following the merger. Moreover, we could not distinguish between other modes of exit so we do not attempt to analyze transition rates for these different types of events separately. Of the 4,089 producing organizations identified in the period, we observed 4,022 exits. Our key explanatory variables are density, overlap density and vertical integration. We update all of our covariates annually using 12-month moving averages. Knowledge that production in the motion picture industry requires an average of 12 months of pre-entry activity governed our choice of lag (Squire, 1992). Following other recent studies (Ruef, 2004; Lomi, Larsen, & Freeman, 2005), we use moving averages as a means of adjusting for inertia in the rate at which firms respond to changes in population density and environmental characteristics. The use of moving averages accounts for the fact that organizational vital rates do not reflect exclusively events that occurred precisely 12 months earlier and also reduces the effect of measurement error in our recording of the timing of events. We measured population density by counting the number of production companies operating each year. In addition to total density, we calculated counts of integrated producer density and non-integrated producer density to investigate asymmetry in the density-dependent effects produced by these two sub-populations. We measured vertical integration using a dichotomous variable set to one, if the producing organization also self-distributed at least one of its films in
The Competitive Dynamics of Vertical Integration
379
a given year, or set to zero otherwise. In separate (unreported) analyses, we calculated a continuous ratio measure of integration by dividing the number of self-distributed films by the total number of films produced in a year. Analyses using that continuous measure produced qualitatively equivalent results to those we report here. We nonetheless prefer the dichotomous measure because it more clearly distinguishes scope from size, and also eases the interpretation of the interaction terms. We measured localized competition through niche overlap density, the sum of niche overlaps (in terms of genre space) between the focal organization and all other firms releasing films in a particular year (Baum & Singh, 1994). The overlap density for an organization in niche i at time t is given by X N it þ wij N jt (1) jai
where Nit is the number of production companies in niche i at time t, Njt the number of production companies in niche j at time t and wij the niche overlap weight for production companies in niche i with those in niche j. Our operationalization of organizational niche comes from the classification of films into genres. AFI classifies films into 72 different genres, but the classification we used includes only 53 categories because we grouped together genres that appear nearly synonymous (e.g., baseball, boxing and football group into a single ‘‘sports’’ genre). Genres represent socially constructed categorizations of conventions regarding content and form (DiMaggio, 1987). As social constructions, the rules for inclusion and exclusion remain somewhat fluid over time and a continuous logic of membership probably better represents the truth than a discrete one. These categories nonetheless provide meaningful if rough information on niche positions for two reasons: First, audiences respond to genre characteristics when they select films to view (Austin, 1989). Second, barriers do appear to constrain the easy movement from one genre to another to some extent. For instance, producing a Western requires different resources and strategies from a musical (lighting, sound equipment, props, choreography, etc.). As a result, production companies likely compete more intensely with other organizations operating in the same genres. Since we do not have a clear means of assigning distances between genres, our formulation of niche overlap, assumes that wij ¼ 0 and that overlap density is equal to X N it , (2) i
380
GIACOMO NEGRO AND OLAV SORENSON
in other words, to the sum of production companies operating in all niches where producer i competes (less one). In addition to these variables of interest, our analyses include several controls. To account for age dependence, we estimate a piecewise exponential specification of time dependence that we describe in greater detail below. By including time-varying information on size, we can isolate the effects of age from scale. Research on time-varying size and life chances has conceptualized organizational size in two ways: capacity and scale of operations. Our measure here, the (log of) annual volume of film production (i.e., a count of films released in the previous 12 months) focuses on the absolute scale of operations. Companies engaged in more projects may survive longer thanks to their ability to diversify away some of the project-specific uncertainty plaguing each individual film (De Vany, 2004). In addition to absolute size, we also control for relative size effects, calculated as the ratio of an organization’s absolute size to the largest size observed in the population for a given year. Scale-based processes can affect survival in complex ways, and the largest organizations may enjoy positional gains by way of scale advantages in politics, production costs, etc. (Dobrev & Carroll, 2003). We also included several industry-level variables to control for changes in carrying capacity and industry structure. First, we included weekly attendance in terms of millions of admissions per year. Over the observed period, motion pictures went from being the primary form of visual entertainment (1912–1946) to being an alternative to television (1947–1970). Second, we introduced a measure of market concentration to control for potential barriers to entry and the alternative explanation of resource partitioning (Carroll, 1985; Mezias & Mezias, 2000). We calculated a Hirschman–Herfindahl index, obtained by squaring and then summing the market share of each producer based on the number of films produced (multiplied by 100).1 Third, we included two dichotomous variables to capture period effects. One variable controls for the potential effects associated with the establishment of the Hollywood studio system (RKO, the youngest of the Majors, began producing films in 1928); our measure takes a value of one between 1928 and 1947, and zero otherwise. Historical analyses of the industry suggest that total integration might have reduced the viability of specialized producers because the Majors controlled access to a significant proportion of first-run theaters (Conant, 1960; Balio, 1985). A second variable, postParamount, captures the potential impact of antitrust actions on the structure of the industry. In 1948, a government suit against the eight largest firms (United States v. Paramount Pictures, 334 U.S. 131) culminated in a Supreme Court decision that imposed divestiture of the exhibition chains
381
The Competitive Dynamics of Vertical Integration
Table 1.
Descriptive Statistics.
Variable
Mean
Std dev
Min
Max
Tenure Weekly attendance (/1,000,000) Studio system period Post-paramount period Market concentration (/1,000) Absolute size Relative size Density Density2 (/100) Vertical integration Density integrated Density2 integrated (/100) Density non-integrated Density2 non-integrated (/100) Overlap density Overlap density integrated producers Overlap density non-integrated producers Niche width
3.94 45.23 0.25 0.48 3.76 0.5 0.08 177.75 357 0.13 22 5.73 155.42 278.89 34.97 11.39 23.58 2
7.15 22.6 0.43 0.5 2.75 1.92 0.15 64.06 243.05 0.34 8.54 4.17 61.11 203.87 46.21 19.25 32.33 2.63
1 17.9 0 0 0.51 0 0 2 0 0 0 0 2 1 0 0 0 0
56 87.25 1 1 58.5 4.44 1 310 961 1 42 18 269 723 326 174 152 24
owned by the Majors. The variable takes the value of one from 1948 to 1970 and zero otherwise.2 Table 1 reports descriptive statistics for the variables used in the regressions.
Estimation We estimate the effects of vertical integration and competitive interaction on organizational performance using survival (event history) analysis. More specifically, we estimate the hazard of exit from the market, defined as mðtÞ ¼ lim
Dt!0
prðt T t þ DtjT4tÞ Dt
(3)
where T is a random variable representing the time of exit, t denotes the amount of time that producer i has been in operation and Prð:Þ represents the probability of exit over the interval (t, t+Dt) given that the production company still had ongoing operations at the beginning of the interval. Age dependence is a crucial feature of these models. We adopt the piecewise exponential specification, which allows the base rate of exit to vary flexibly with organizational age. In particular, this approach splits time into pieces according to the tenure of the organization. The base failure rate remains
382
GIACOMO NEGRO AND OLAV SORENSON
constant within each timepiece, but these base rates can vary across pieces. As a result, the piecewise model does not require any strong assumption about the exact form of duration dependence (for more information on this approach, see Barron, West, & Hannan, 1994). We define the P pieces according to break points 0 t1 t2 . . . tp
(4)
and with tP+1 ¼ N. Our exploratory research on the population found the best fit using five break points at 1, 2, 4, 10 and 20 years (intervals open on the right). The first segment then includes events occurring within the first year of tenure in the industry and cases that enter and exit within the same year. The second segment includes events that occur within the first and second years of tenure, and so forth. We specify producer exit rates r(u,t) as a function of firm tenure in the industry, u, a set of time-varying covariates, X, and a set of time-invariant covariates, Z X X ln ri ðuÞ ¼ mp þ ax xit þ bz z i (5) where mp denotes tenure-specific effects, and a and b respectively scale the effects of time-varying and time-invariant effects. To estimate rate models with time-varying covariates, we constructed split-spell data breaking observation periods with durations of more than one year to allow for the updating of annual covariates.
Results Tables 2 and 3 report the results of our analysis. Model 1 provides a baseline density-dependence model. The baseline shows that film production follows the expected form of density-dependent evolution, with exit rates having the predicted U-shaped relationship to density. The results also suggest that film production companies suffer from a liability of newness and experience decreasing exit rates over time. The extremely high exit rate associated with the first year of tenure may actually stem from a specific feature of the movie industry, where individuals frequently organize to produce a single motion picture and then intentionally disband. Or, this liability may reflect the risky nature of movie-making, where 8 out of 10 products fail at the box office (De Vany, 2004). The benefits associated with very long tenures may stem either from the development or selection of effective production routines
383
The Competitive Dynamics of Vertical Integration
Table 2. Piecewise Exponential Regressions: Competitive Effects of Vertical Integration and Density Dependence on Producer Exit Rates. Variable Tenure 0our1 Tenure 1our2 Tenure 2our4 Tenure 4our10 Tenure 10our20 Tenure uZ20 Weekly attendance Studio system period Post-paramount period Market concentration Absolute size Relative size Density Density2
Model 1 2.644 (0.181) 1.062 (0.191) 0.989 (0.192) 0.777 (0.193) 0.716 (0.213) 0.054 (0.295) 0.002 (0.002) 0.093 (0.086) 0.101 (0.047) 0.019 (0.014) 1.465 (0.053) 0.188 (0.442) 0.016 (0.002) 0.005 (0.000)
Vertical integration
Model 2 2.858 (0.186) 1.297 (0.197) 1.227 (0.197) 0.951 (0.199) 0.910 (0.218) 0.495 (0.299) 0.001 (0.002) 0.030 (0.087) 0.030 (0.048) 0.016 (0.015) 1.429 (0.053) 0.159 (0.452) 0.016 (0.002) 0.005 (0.000) 0.842 (0.063)
Density2 integration Density integrated
Model 3 2.820 (0.187) 1.260 (0.197) 1.189 (0.198) 0.971 (0.199) 0.910 (0.218) 0.482 (0.299) 0.002 (0.002) 0.035 (0.087) 0.035 (0.048) 0.015 (0.015) 1.430 (0.053) 0.156 (0.451) 0.016 (0.001) 0.005 (0.000) 0.652 (0.111) 0.001 (0.000)
Density2 integrated Density nonintegrated Density2 nonintegrated Log-likelihood Number of producers Number of producer-year spells Chi-square vs. null rate
Model 4 2.489 (0.201) 0.830 (0.212) 0.729 (0.212) 0.487 (0.234) 0.362 (0.215) 0.306 (0.335) 0.003 (0.003) 0.312 (0.109) 0.053 (0.067) 0.022 (0.022) 1.717 (0.062) 0.875 (0.613)
3.932 (0.924) 3.412 (0.942) 3.449 (0.947) 3.417 (0.952) 3.106 (1.001) 1.661 (1.040) 0.011 (0.011) 0.995 (0.517) 1.155 (0.491) 0.144 (0.101) 1.052 (0.121) 2.282 (1.255)
0.057 (0.014) 0.102 (0.027) 0.021 (0.002) 0.006 (0.001)
0.127 (0.594) 0.216 (0.106) 0.024 (0.007) 0.010 (0.002)
16712.2 4,089
16685.9 4,089
16595.8 4,089
13779.5 3,913
9,316
9,316
9,316
8,044
Standard errors in parentheses. po0.10. po0.05. po0.01.
228.75
232.79
Model 5
173.14
681.208 363 1,272
28.94
384
GIACOMO NEGRO AND OLAV SORENSON
Table 3. Piecewise Exponential Regressions: Competitive Effects of Vertical Integration and Niche Overlap on Producer Exit Rates. Variable Tenure 0our1 Tenure 1our2 Tenure 2our4 Tenure 4our10 Tenure 10our20 Tenure uZ20 Weekly attendance Studio system period Post-Paramount period Market concentration Absolute size Relative size Density Density2 Vertical integration Overlap density Overlap density integration Overlap of integrated producers Overlap of non-integrated producers Overlap of integrated integration Overlap of non-integrated integration
Model 6
Model 7
2.815
2.477
(0.187) 1.250 (0.197) 1.182 (0.197) 0.915 (0.218) 0.865 (0.199) 0.457 (0.300) 0.002 (0.002) 0.014 (0.088) 0.078 (0.050) 0.019 (0.015) 1.516 (0.593) 0.034 (0.452) 0.016 (0.002) 0.005 (0.000) 0.840 (0.063) 0.002 (0.000)
(0.194) 0.922 (0.204) 0.853 (0.204) 0.559 (0.225) 0.536 (0.207) 0.504 (0.307) 0.002 (0.002) 0.017 (0.086) 0.072 (0.057) 0.019 (0.014) 1.414 (0.053) 0.204 (0.447) 0.016 (0.002) 0.004 (0.000) 0.850 (0.063) 0.002 (0.001) 0.004 (0.001)
Model 8 2.876 (0.193) 1.313 (0.203) 1.244 (0.203) 0.964 (0.223) 0.929 (0.205) 0.507 (0.305) 0.002 (0.002) 0.025 (0.089) 0.074 (0.050) 0.026 (0.015) 1.507 (0.063) 0.505 (0.459) 0.016 (0.002) 0.005 (0.000) 1.102 (0.287)
0.005 (0.002) 0.004 (0.001) 0.001 (0.001) 0.003 (0.001)
385
The Competitive Dynamics of Vertical Integration
Table 3. (Continued ) Variable Log-likelihood Number of producers Number of producer-year spells Chi-square vs. null rate
Model 6
Model 7
Model 8
16592.1 4,089 9,316 240.19
16580.4 4,089 9,316 249.43
16650.9 4,089 9,316 290.52
Standard errors in parentheses. *po0.10. po0.05. po0.01.
and/or from having established relations with distributors or other critical resource holders (Sorenson & Waguespack, 2004). With respect to scale, increasing size appears to provide an effective means of reducing market risks. All models reveal negative size dependence. Since analyses of the industry typically do not find substantial cost savings associated with the simultaneous production of multiple films, these effects likely stem more from risk diversification than from economies of scale. The non-significant effect of relative size also supports such an interpretation. Consistent with this idea, market concentration does not significantly influence exit rates in any of the models (hence cost advantages do not appear to push smaller production companies out of the market). Among the period effects, only the post-Paramount indicator variable ever shows a significant coefficient. Production companies experienced an increased hazard of exit following the divestiture of the Majors out of exhibition. Given that the justification for this decision revolved to some degree around the idea that the studio system had locked independent production companies and distributors out of the market, this result seems somewhat surprising. It might nonetheless stem from two factors. On the one hand, by reducing the barriers to entry, the Paramount agreement may have increased the number of firms with limited resources entering the industry, thereby increasing the exit rate. Or, it may reflect a riskier environment. Around the same time as the Paramount decision, production companies faced the arrival of television, which both increased the competition for leisure time and for many of the types of personnel required for motion picture production. Model 2 introduces the main effect of vertical integration. Being integrated reduces the instantaneous hazard of exit from the population. The magnitude of the effect, moreover, is large: integrated producers have a 57
386
GIACOMO NEGRO AND OLAV SORENSON
percent lower rate of exit than specialized producers (exp 0.842ffi0.43). This result is consistent with Sorenson’s (2003) analysis of vertical integration in the computer workstation industry; he found that integrated manufacturers enjoyed a survival advantage of roughly 33 percent compared to nonintegrated organizations. Model 3 then tests to what extent this effect stems from integration’s ability to buffer firms from the competitive effects of density dependence. We find some support for this hypothesis. Models 4 and 5 decompose density into two sub-counts, one for integrated production companies and another for non-integrated firms, and analyze the impact of these measures on the exit rates of the two subpopulations. For non-integrated producers, non-integrated density continues to exhibit a U-shaped relationship with exit rates, first falling and then rising with increasing density. The density of integrated firms, however, operates quite differently. Though the effect of integrated firm density on non-integrated producer exit appears first to rise and then to fall, exit rates peak at a point close to the maximum of the observed integrated density range (n ¼ 29, where the multiplier of the rate is equal to 2.22). It therefore appears that integrated firms have a purely competitive effect on nonintegrated producers. Though it seems that integrated firms do not help to legitimate the non-integrated form, it is also possible that the form had already been legitimated before production companies began integrating into distribution. As expected, integrated producer density increases the exit rates of non-integrated producers more than non-integrated producer density. For example, at their mean values, the density of integrated producers increases the hazard rate of exit by 100 percent (multiplier ¼ 2.001 for n ¼ 22), while the density of non-integrated producers decreases the chance of exit by 83 percent (multiplier ¼ 0.165 for n ¼ 155). The densities of both integrated producers and non-integrated producers exhibit U-shaped relations to the exit rates of integrated producers. As one can see from the coefficient estimates, each integrated firm contributes much more to both the legitimating (first-order term) and competitive (quadratic term) effects of density. Consistent with our expectations, then, nonintegrated producer density increases the exit rates of integrated producers less than integrated producer density. At the mean of their observed ranges, for example, integrated producer density decreases the rate of exit by 80 percent (multiplier ¼ 0.203 for n ¼ 22), while non-integrated producer density decreases the hazard rate by only 74 percent (multiplier ¼ 0.258 for n ¼ 155). Table 3 reports our analysis of the effects of niche overlap. Models 6–8 address the effects of overlap. Models 6 and 7 test the main effect of overlap
The Competitive Dynamics of Vertical Integration
387
density and its interaction with vertical integration. As expected, increasing overlap with production companies operating in the same film genres reduces organizational survival. Integrated production companies, on the other hand, do not appear to suffer from this overlap. Model 8 decomposes niche overlap according to vertical scope, showing that genre overlaps with integrated and non-integrated production companies show roughly equivalent effects, and includes the interactions between these overlap densities and vertical integration. The interaction between integration and integrated firm overlap is not significant, suggesting that integrated firms impose roughly equivalent competitive pressure on both integrated and nonintegrated firms (hence failing to support Hypothesis 3a). The interaction between integration and non-integrated firm overlap, on the other hand, is negative, significant and roughly equal in magnitude to the main effect of the overlap. Integrated firms therefore do not appear to suffer from competition with non-integrated firms (consistent with Hypothesis 3b). In sum, we find support for Hypotheses 1, 2 and 3b, but not Hypothesis 3a.
INTEGRATION AND NICHE WIDTH The results of the preceding section imply that vertical integration buffers integrated firms from competitive pressures, particularly from those generated by non-integrated firms. Clearly, this buffering influences survival rates, but it might also affect other aspects of organizational behavior? Here, we investigate how vertical integration may interact with niche width in determining firm viability. Film historians have analyzed production trends among the Hollywood Majors and their accounts suggest that each integrated producer developed similarly wide, but unique product strategies: MGM became prominent in the production of sophisticated dramas called ‘‘prestige films,’’ Paramount invested in comedies starring actors recruited from vaudeville, radio and stage, RKO introduced musicals, and distributed successful animated films created by Walt Disney (Bordwell et al., 1985; Balio, 1993). The main reason that we suspect a link between these two dimensions is that buffering from competition may provide an incentive for vertically integrated firms to expand their horizontal scope. Broader scope offers the organization a greater ability to weather shifts in the environment, for example, as a result of changing consumer preferences. Previous studies have shown that firms with broad niches benefit from risk spreading and economies of scale (Baum & Singh, 1994; Dobrev et al., 2001). This reduction in
388
GIACOMO NEGRO AND OLAV SORENSON
risk nevertheless comes at a price because as organizations expand their scope they come into competitive contact with a larger number of rivals. To the extent that vertically integrated firms can avoid this competition, scope offers a relatively greater advantage to these firms. On the other hand, one might anticipate a positive relation between vertical and horizontal scope even in the absence of these incentives. Expansion in both directions might simply reflects the routines that reside at the core of the organization (Sorenson, McEvily, Ren, & Roy, 2006). Firms that expand their boundaries in one direction reveal operating procedures for and values favoring the expansion of firm scope. The very act of engaging these routines moreover may strengthen them as they become embedded within the firms, operations and routinized among employees. The key difference between these two accounts concerns the benefits of expanded scope for integrated firms relative to non-integrated firms. The first argument implies that vertically integrated firms benefit more from broad scope than non-integrated firms, while the second argument does not. We therefore began our investigation by estimating the effects of niche width on organizational exit. We measured horizontal scope with niche width, calculated as the number of film genres in which the producing organization engages. Table 4 presents the regression results of a model estimating the effect of niche width, measured as the number of genres in which a producer operates, and integration on exit rates. As we expected, exit rates rise with niche width, but only for non-integrated organizations. Integrated firms should therefore prefer wider niches. Consistent with this finding and with the historical accounts, vertically integrated firms appear to maintain a broader scope than non-integrated firms for most of the industry’s history. Fig. 3 illustrates scope relative to size – the ratio of the number of genres represented to the number of films made – for both sub-populations (we exclude companies with only a single film in a year from these calculations because they trivially have a ratio of one).
DISCUSSION AND CONCLUSION Our results strongly suggest that vertical integration influences not just the internal operations of organizations, but also the competitive interactions between firms. We see both specific and broader implications of our study. At a more specific level, whereas previous research primarily emphasized the role of integration as an organizational arrangement to gain efficiency in
The Competitive Dynamics of Vertical Integration
389
Table 4. Piecewise Exponential Regressions: Competitive Effects of Vertical Integration and Niche Width on Producer Exit Rates. Variable Tenure 0our1 Tenure 1our2 Tenure 2our4 Tenure 4our10 Tenure 10our20 Tenure uZ20 Weekly attendance Studio system period Post-Paramount period Market concentration Absolute size Relative size Density Density2 Vertical integration Niche width Niche width integration Log-likelihood Number of producers No. of producer-year spells Chi-square vs. null rate Standard errors in parentheses. *po0.10. po0.05. po0.01.
Model 9 3.142 0.196) 1.578 (0.206) 1.510 (0.206) 1.224 (0.227) 1.191 (0.208) 0.681 (0.309) 0.001 (0.002) 0.002 (0.087) 0.046 (0.048) 0.023 (0.015) 1.233 (0.074) 0.513 (0.457) 0.016 (0.002) 0.005 (0.000) 1.324 (0.099) 0.371 (0.057) 0.244 (0.058) 16561.3 4,089 9,316 266.94
390
GIACOMO NEGRO AND OLAV SORENSON
1
Producer scope
0.8
0.6
0.4
0.2
Scope of integrated producers
0
Scope of non-integrated producers
1912 1916 1920 1924 1928 1932 1936 1940 1944 1948 1952 1956 1960 1964 1968 1914 1918 1922 1926 1930 1934 1938 1942 1946 1950 1954 1958 1962 1966 1970
Year
Fig. 3.
Integrated, Non-Integrated Producers and their Scope.
transactions, we offer and support an ecological perspective on the vertical integration question. Vertical integration buffers firms from environmental uncertainty and competitive pressures through the internalization of input and output activities. Hence we find that integrated firms exhibit less sensitivity to diffuse competition, particularly that produced by non-integrated firms. This buffering also appears to allow integrated firms to pursue broader product niches, which connotes its own advantages to these firms. In addition to buffering integrated firms from the environment, vertical integration may also increase the intensity of competition experienced by non-integrated firms. Integrated firms enjoy privileged access to the resources they control directly. And owing to the greater stability they offer to resource providers, integrated firms may also receive preferential access to external resources. Because these ecological benefits result from the relative competitive strength of integrated firms vis-a`-vis non-integrated firms, they likely offer the greatest benefits when the number of integrated firms is few relative to the number of non-integrated firms; when competing against other integrated rivals, integrated firms find themselves on level ground. The combined analysis of integration and niche width moreover offers a more nuanced view of competitive processes, and the resulting evidence allows us to address other possible explanations for the observed patterns.
The Competitive Dynamics of Vertical Integration
391
First, one might argue that, contrary to our prediction, vertically integrated firms would compete more intensely with each other because their structural overlap increases. Not only do the general findings of less-intense densityand overlap-based competition run counter to this account, but also the fact that integrated producers enjoy lower exit rates when they expand their niche seems inconsistent with the expectations of this thesis. Second, we also believe that we can rule out the advantages of head-to-head competition as an alternative explanation for our findings (Klemperer, 1992). In cases with high search costs for alternative suppliers, competing head-to-head benefits firms and we would consequently expect higher exit rates for integrated producers. Again, however, we find the opposite effect. Finally, one might attribute our cross-competition effects to differences in the relative efficiency of integrated and non-integrated firms (and hence a consequence of differing internal processes rather than external ones). Though we cannot bring direct evidence to bear on this possibility, Corts’ (2001) recent analysis of the release decisions of integrated and non-integrated firms finds evidence inconsistent with this hypothesis. Despite the advantage of coordinating release dates to avoid product cannibalization, non-integrated producers and distributors actually exhibited less-clustered releases than integrated producers. From where such internal efficiency would arise therefore remains unclear in this setting. Despite the support for the hypotheses we have developed, this study suffers from four main limitations. First, we have modeled and analyzed the impact of integration on film producers alone. To improve the validity of our results, we should observe similar patterns in the performance of film distributors. Second and partly related to this, film producers do not develop specific investments with distributors but distributors may develop specific commitments with exhibitors, making the final outcome of integration more complex. Moreover, other types of sunk costs play a role here, raising the exit barriers that integrated firms face. For instance, advertising expenses crucially influence distributor performance and excluding them from the picture might lead to inappropriate inferences regarding the effects of integration. Third, we lack information on various forms of quasiintegration (e.g., strategic alliances) and hence cannot speak to whether or not these structures create dynamics similar to those of vertical integration. Finally, the focus on film genres as a measure for organizational niche means that we have analyzed realized rather than fundamental niches. The problem with the use of realized niche measures is that we cannot separate organizations’ abilities to procure and exploit resources from the outcomes of their competitive interactions. Alternative niche measures therefore might
392
GIACOMO NEGRO AND OLAV SORENSON
provide a better basis for future empirical research on vertical integration and niche-width dynamics. Though this study helps to clarify how horizontal competition affects and is affected by the vertical organization of industries, our results seem relevant to several other issues and literatures as well. For example, the analysis of horizontal and vertical scope can connect to niche width theory (Freeman & Hannan, 1983). In its original formulation, this theory considered the relationship between the width of an organization’s fundamental niche and its capacity for resource utilization, holding that specialist strategies prevail over generalist strategies in fine-grained environments (i.e., environments where resource distributions shift rapidly in time and/or space). If we believe the film industry occupies a fine-grained environment, with frequent fluctuations in demand, we expect to find a relative advantage to specialism. One might then expect that advantage to lead to the extinction of the other sub-form (Gause, 1934). Our results indicate, however, that integration can help to reduce the inferiority in the fitness of generalists, and concomitantly to explain the coexistence of generalists and specialists in industries. Our results may also inform strategy research, particularly studies of multimarket competition (MMC). MMC occurs when firms encounter the same rivals in multiple markets. In these circumstances, competitive behavior may differ from that of single-point rivals because a firm that meets a rival in multiple markets can respond to an attack not only in the besieged market, but also in other markets in which both firms compete. MMC can therefore foster tacit cooperation, and reduce the intensity of competition among rivals (Baum & Korn, 1996; Gimeno & Woo, 1999). Our analysis of the film industry suggests that, if producers ‘‘meet’’ rivals in the same genres, integration allows distinct strategies to reduce the effects of niche overlap. This effect could interfere with the incentive structure that allows tacit collusion to emerge in MMC, and therefore could mediate its effect on competitive intensity. We therefore see reason to incorporate information and theory on vertical integration in future studies of MMC. Finally, our study provides evidence consistent with game theoretic and industry structure perspectives on integration. The basic argument in this literature is that vertical integration can increase profitability by internalizing the downstream or upstream profit margin and augmenting the integrated firm’s market power through market foreclosure. Our study corroborates the intuition that vertical integration benefits the integrated firms and hurts the non-integrated ones. Most models that address the vertical foreclosure effect of integration nonetheless focus on the competitive
The Competitive Dynamics of Vertical Integration
393
exclusion of downstream firms rather than upstream competitors (Rey & Tirole, 2003). This becomes particularly important when economic theory maintains that vertical integration reduces consumer surplus and total welfare. By focusing on downstream market outcomes, these studies overlook the potential positive effects – that we find here – related to increased differentiation in the upstream market (see also Chipty, 2001). What our study leaves unanswered is why firms choose different levels of vertical integration in the first place. Here, however, we believe that the companion paper by Gimeno, Chen, and Bae (2006, this volume) offers a nice complement. They demonstrate that firms actively manage their output markets, resource endowments and strategic postures as a means of adjusting to (and thereby generally reducing) pair-wise competition. Though not considered in their analysis, our results suggest that we might usefully consider vertical integration another dimension of strategic action (and another means of reducing ecological interdependence) that managers vary in their responses to rivals. As our results imply that these moves have asymmetric consequences, however, the manner in which managers use integration as a response to pair-wise competition might well vary from that the actions considered by Gimeno, Chen, and Bae. We also see potential policy implications in our study. The antitrust decision that forced the Majors to divest of their theaters turned on the notion that vertical integration allowed these firms to gain undue access to the market, to the detriment of both rival production and distribution companies and consumers. Some scholars have nonetheless maintained that the hierarchical structure of the Major studios and the long-term contracts they used to book films into theaters actually represented a more efficient solution to the problems associated with identifying demand, controlling principal–agent problems, and maintaining the flexibility necessary to adapt to new technologies (De Vany, 2004). Our results seem more consistent with the first view, but we see a novel twist. In particular, we find substantial evidence that forced divestiture might not have served public interest. First, in all of our models, the external reorganization required by the antitrust decision increased the exit rates of film production companies. One possible reason for this effect is that vertical integration may have helped to stabilize the industry as a whole, even if the majority of the advantages accrued to the integrated firms. Another possibility, of course, is that the combination of competition with, and the potential for ancillary markets in, television disrupted established routines around the time of the Paramount agreement. More research is required, but our findings nonetheless raise questions regarding the standard antitrust wisdom.
394
GIACOMO NEGRO AND OLAV SORENSON
Maximum number of film genres
25
20
15
10
5
0 1912 1916 1920 1924 1928 1932 1936 1940 1944 1948 1952 1956 1960 1964 1968 1914 1918 1922 1926 1930 1934 1938 1942 1946 1950 1954 1958 1962 1966 1970
Year
Fig. 4.
The Maximum Number of Film Genres in which Producing Organizations Operated.
Second, our investigation of product scope suggests that variety increases with vertical integration. At least two factors might account for this fact. On the one hand, vertical integration buffers firms from competition, thereby reducing the costs associated with broader scope. On the other hand, guaranteed outlets also decrease the risks associated with producing novel genres. Before these new types of films gain legitimacy, distributors and exhibitors likely have a reluctance to commit to these untested products. Fig. 4 depicts the evolution of the maximum number of film genres in which producers operated. Consistent with our speculation, maximum diversity coincided with the period with the highest density of integrated production companies; following the Paramount case, it declined. One proposition deriving from this argument is that reduced competition among vertically integrated firms has greater positive welfare consequences relative to competition among specialized producers because of the additional product variety supported by integration. Finally, at a more general level, we view these results as a call for greater ecological research. In the late 1970s, with the emergence of organizational
The Competitive Dynamics of Vertical Integration
395
ecology, institutional theory and resource dependence, organizational scholars witnessed an explosion of interest in incorporating the environment more seriously into our understanding of the operations of organizations. Despite substantial progress in these research programs, the dominant mode of analysis with respect to many organizational questions remains a focus on the internal efficiency of the firm. Vertical integration is but one example. Extant empirical research has focused primarily on whether ownership offers a more efficient solution to contracting in a market (Williamson, 1975). Even ‘‘dynamic’’ analyses of this phenomenon have been framed in terms of whether firms appear to move toward this ‘‘equilibrium.’’ We claim that in addition to these internal effects, the consequences of many structural features of the organization, such as vertical integration, reverberate throughout the dynamics of a population. The effects ripple through the population because these structural features of the organization influence firms’ resource requirements and concomitantly their interactions with other firms either in the same or in overlapping niches.
NOTES 1. In unreported analyses, we also estimated models including interaction terms between size and concentration to capture the potential effects of consolidation on industry dynamics (Dobrev, Kim, & Carroll, 2002). This interaction term, however, had no significant effect on producer exit rates. 2. Several studies have argued that the nature of competition varies over an industry’s lifecycle (Baum, 1995; Hannan, 1997; Sorenson, 2000). Though we tested these alternative specifications, none of them substantively modified the reported results.
ACKNOWLEDGMENTS We would like to thank the volume editors, Jerker Denrell, Carolin Haeussler and Michael Hannan for comments on previous versions of the paper and helpful discussions in the development of this research. Negro received additional financial support through a grant from SDA Bocconi’s Claudio Dematte` Research Center.
REFERENCES American Film Institute (AFI). (1989–1999). The American Film Institute Catalog of motion pictures produced in the United States: Feature films, 1911–1920,
396
GIACOMO NEGRO AND OLAV SORENSON
1921–1930, 1931–1940, 1941–1950, 1961–1970. Berkeley, CA: University of California Press. Austin, B. A. (1989). Immediate seating: A look at movie audiences. Belmont, CA: Wadsworth. Balio, T. (1985). The American film industry. Madison, WI: University of Wisconsin Press. Balio, T. (1993). Grand design: Hollywood as a modern business enterprise 1930–1939. Berkeley, CA: University of California Press. Barron, D. N., West, E., & Hannan, M. T. (1994). A time to grow and a time to die: Growth and mortality of credit unions in New York, 1914–1990. American Journal of Sociology, 100, 381–421. Baum, J. A. C. (1995). The changing basis of competition in organizational populations: The Manhattan hotel industry, 1887–1990. Social Forces, 74, 177–205. Baum, J. A. C., & Korn, H. J. (1996). Competitive dynamics of interfirm rivalry. Academy of Management Journal, 39, 255–291. Baum, J. A. C., & Mezias, S. J. (1992). Localized competition and organizational mortality in the Manhattan hotel industry, 1898–1990. Administrative Science Quarterly, 37, 580–604. Baum, J. A. C., & Singh, J. V. (1994). Organizational niches and the dynamics of organizational mortality. American Journal of Sociology, 100, 346–380. Bordwell, D., Staiger, J., & Thompson, K. (1985). The classical Hollywood cinema. Film style and mode of production to 1960. London: Routledge. Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in populations of organizations. American Journal of Sociology, 90, 1262–1283. Carroll, G. R., & Hannan, M. T. (2000). The demography of corporations and industries. Princeton: Princeton University Press. Chipty, T. (2001). Vertical integration, market foreclosure, and consumer welfare in the cable television industry. American Economic Review, 91, 428–453. Christopherson, S., & Storper, M. (1989). The effects of flexible specialization on industrial politics and the labor market: The motion picture industry. Industrial and Labor Relations Review, 42, 333–347. Conant, M. (1960). Antitrust in the motion picture industry. Economic and legal analysis. Berkeley, CA: University of California Press. Copyright Office. (1951–1971). Catalog of copyright entries: Motion pictures 1912–1939, 1940– 1949, 1950–1959, 1960–1969. Washington: Library of Congress. Corts, K. S. (2001). The strategic effects of vertical market structure: Common agency and divisionalization in the U.S. motion picture industry. Journal of Economics and Management Strategy, 10, 509–552. De Vany, A. (2004). Hollywood economics. New York: Routledge. DiMaggio, P. J. (1987). Classification in art. American Sociological Review, 52, 440–455. Dobrev, S. D., & Carroll, G. R. (2003). Size (and competition) among organizations: Modeling scale-based selection among automobile producers in four major countries, 1885–1981. Strategic Management Journal, 24, 541–558. Dobrev, S. D., Kim, T.-Y., & Carroll, G. R. (2002). The evolution of organizational niches: U.S. automobile manufacturers, 1885–1981. Administrative Science Quarterly, 47, 233–264. Dobrev, S. D., Kim, T.-Y., & Hannan, M. T. (2001). Dynamics of niche width and resource partitioning. American Journal of Sociology, 106, 1299–1337. Ferrow, A. G. (1999). Feature films 1950–1959: A United States filmography. Jefferson, NC: McFarland and Company, Inc.
The Competitive Dynamics of Vertical Integration
397
Freeman, J. H., & Hannan, M. T. (1983). Niche width and the dynamics of organizational populations. American Journal of Sociology, 88, 1116–1145. Gause, G. F. (1934). The struggle for existence. Baltimore, MD: Williams and Wilkins. Gimeno, J., Chen, M. -J., & Bae, J. (2006). Dynamics of competitive niche posititioning. In: J. A. C. Baum, S. D. Dobrev, & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 399–441). Oxford UK: JAI. Gimeno, J., & Woo, C. Y. (1999). Multimarket contact, economies of scope, and firm performance. Academy of Management Journal, 43, 239–259. Grossman, S., & Hart, O. (1986). The costs and benefits of ownership: A theory of vertical and lateral integration. Journal of Political Economy, 94, 691–719. Hannan, M. T. (1997). Inertia, density and the structure of organizational populations: Entries in European automobile industries. Organization Studies, 18, 193–228. Hannan, M. T., & Freeman, J. H. (1977). The population ecology of organizations. American Journal of Sociology, 82, 929–964. Hart, O., & Moore, J. (1990). Property rights and the nature of the firm. Journal of Political Economy, 98, 1119–1158. Joskow, P. (2005). Vertical integration. In: P. Menard, & M. M. Shirley (Eds), Handbook of new institutional economics. New York: Springer. Kerr, C. (1990). Incorporating the star: The intersection of business and aesthetic strategies in early American film. Business History Review, 64, 383–410. Klein, P. (2005). The make or buy decisions: Lessons from empirical studies. In: P. Menard, & M. M. Shirley (Eds), Handbook of new institutional economics. New York: Springer. Klemperer, P. (1992). Equilibrium product lines: Competing head-to-head may be less competitive. American Economic Review, 82, 740–755. Krattenmaker, T. G., & Salop, S. C. (1986). Anticompetitive exclusion: Raising rivals’ costs to achieve power over price. Yale Law Journal, 96, 209–293. Lewis, H. T. (1933). The motion picture industry. New York: D. Van Nostrand Company. Lomi, A., Larsen, E. R., & Freeman, J. H. (2005). Things change: Dynamic resource constraints and system-dependent selection in the evolution of populations of organizations. Management Science, 51, 882–903. McPherson, J. M. (1983). An ecology of affiliation. American Sociological Review, 48, 519–535. Mezias, S. J., & Boyle, E. (2005). Blind trust: Market control, legal environments, and the dynamics of competitive intensity in the early American film industry, 1893–1920. Administrative Science Quarterly, 50, 1–34. Mezias, J. M., & Mezias, S. J. (2000). Resource partitioning, the founding of specialist firms, and innovation: The American film industry, 1912–1929. Organization Science, 11, 306–322. Nash, J. R., & Ross, S. R. (1983). The motion picture guide 1927–1982. New York: Cinebooks. Nickerson, J. A., & Silverman, B. S. (2003). Why firms want to organize efficiently and what keeps them from doing so: Inappropriate governance, performance, and adaptation in a deregulated industry. Administrative Science Quarterly, 48, 433–465. Ordover, J. A., Saloner, G., & Salop, S. C. (1990). Equilibrium vertical foreclosure. American Economic Review, 80, 127–142. Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations. New York: Harper and Row. Rey, P., & Tirole, J. (2003). A primer on foreclosure. In: M. Armstrong, & R. Porter (Eds), Handbook of industrial organization (Vol. 3). New York: North-Holland.
398
GIACOMO NEGRO AND OLAV SORENSON
Ruef, M. (2004). For whom the bell tolls: Ecological perspectives on industrial decline and resurgence. Industrial and Corporate Change, 13, 61–89. Rumelt, R. P. (1974). Strategy, structure and economic performance. Boston: Division of Research, Graduate School of Business Administration, Harvard University. Salinger, M. A. (1988). Vertical mergers and market foreclosure. Quarterly Journal of Economics, 77, 345–356. Silverman, B. S., Nickerson, J. A., & Freeman, J. H. (1997). Profitability, transactional alignment, and organizational mortality in the U.S. trucking industry. Strategic Management Journal, 18(Special Issue), 31–52. Sørensen, J. B. (1999). Executive migration and interorganizational competition. Social Science Research, 28, 289–315. Sørensen, J. B. (2004). Recruitment-based competition between industries: A community ecology. Industrial and Corporate Change, 13, 149–170. Sorenson, O. (1997). The complexity catastrophe in the computer industry: Interdependence and adaptability in organizational evolution. Unpublished Ph.D. dissertation, Stanford University. Sorenson, O. (2000). The effect of population-level learning on market entry: The American automobile industry. Social Science Research, 29, 307–326. Sorenson, O. (2003). Interdependence and adaptability: Organizational learning and the longterm effect of integration. Management Science, 49, 446–463. Sorenson, O., & Audia, P. G. (2000). The social structure of entrepreneurial activity: Geographic concentration of footwear production in the U.S., 1940–1989. American Journal of Sociology, 106, 424–462. Sorenson, O., McEvily, S., Ren, C. R., & Roy, R. (2006). Niche width revisited: Organizational scope, behavior and performance. Strategic Management Journal (forthcoming). Sorenson, O., & Waguespack, D. (2004). Social networks and exchange: Self-confirming dynamics in Hollywood. Working paper. UCLA Anderson Graduate School of Management. Squire, J. E. (1992). The movie business book. New York: Fireside. Stuart, T. E., & Sorenson, O. (2003). The geography of opportunity: Spatial heterogeneity in founding rates and the performance of biotechnology firms. Research Policy, 32, 229–252. Thompson, J. D. (1967). Organizations in action. Social science bases of administrative theory. New York: McGraw-Hill. Williamson, O. E. (1971). The vertical integration of production: Market failure considerations. American Economic Review, 61, 112–123. Williamson, O. E. (1975). Markets and hierarchies: Analysis and antitrust implications. A study in the economics of internal organization. New York: Free Press.
DYNAMICS OF COMPETITIVE REPOSITIONING: A MULTIDIMENSIONAL APPROACH Javier Gimeno, Ming-Jer Chen and Jonghoon Bae ABSTRACT We investigate the dynamics of competitive repositioning of firms in the deregulated U.S. airline industry (1979–1995) in terms of a firm’s target market, strategic posture, and resource endowment relative to other firms in the industry. We suggest that, despite strong inertia in competitive positions, the direction of repositioning responds to external and internal alignment considerations. For external alignment, we examined how firms changed their competitive positioning to mimic the positions of similar, successful firms, and to differentiate themselves when experiencing intense rivalry. For internal alignment, we examined how firms changed their position in each dimension to align with the other dimensions of positioning. This internal alignment led to convergent positioning moves for firms with similar resource endowments and strategic postures, and divergent moves for firms with similar target markets and strategic postures. The evidence suggests that repositioning moves in terms of target markets and resource endowments are more sensitive to external and internal alignment considerations, but that changes in strategic posture are subject to very high inertia and do not appear to respond well to alignment considerations. Ecology and Strategy Advances in Strategic Management, Volume 23, 399–441 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23013-4
399
400
JAVIER GIMENO ET AL.
Firms in most industries occupy different market niches, implement various strategies, and develop and deploy distinct organizational resources. Understanding the antecedents, extent, and consequences of intra-industry heterogeneity has provided a fruitful area of research for both strategic management and ecology scholars since the 1970s (Hannan & Freeman, 1977; Hatten & Schendel, 1977), and remains so today. Over time, similarities between these research traditions have come to exist. Both streams have tended to conceptualize strategic heterogeneity as differences in ‘‘positions’’ within the industry. Both streams have developed typologies of strategies, whose performance depends on fit with environmental opportunities and the competitive context of other interacting firms (Barnett & Hansen, 1996; Carroll, 1985; Carroll & Swaminathan, 1992; McGee & Thomas, 1986). Yet these perspectives have also differed in their fundamental assumptions, such as the extent of strategic change that is possible, and whether strategic changes respond effectively to changes in environment opportunities and competitive context (Barnett & Carroll, 1995; Dobrev, Kim, & Hannan, 2001; Hannan & Freeman, 1984). Over the last three decades, however, the strategy and ecology streams have sufficiently relaxed their initial assumptions to allow some coherence and productive convergence. An area where further convergence is needed is the study of the dynamics of competitive repositioning. The evolution of niche width and market positions occupied by firms – i.e., competitive repositioning (e.g., Jovanovic, 2001) – is a critical empirical observation for both strategy and ecology research. Evidence on this domain informs the debate about the extent of strategic change, the responsiveness of strategic change to external and internal inducements, and the effectiveness of change (e.g., Baum & Singh, 1996; Dobrev, Kim, & Hannan, 2003). Researchers in ecology have recently started to shed light on this burgeoning research area. For example, the ecology literature views competitive positioning or competitive scope as niche width (Boone, Carroll, & van Witteloostuijin, 2004), and it examines the dynamics of competitive repositioning by invoking the competitive effects of niche overlap (Baum & Singh, 1996; McPherson, 1983; McPherson & Rotolo, 1996) or the resource-partitioning process (Carroll, 1985; Carroll & Swaminathan, 2000; Dobrev et al., 2001). Despite this progress, these studies have so far adopted a relatively narrow, one-dimensional interpretation of competitive positioning and the strategy that underlies it. To close the gap, this paper considers firm strategy as a multidimensional construct, an approach that corresponds with the conceptualization traditionally embraced in the strategic management field. Definitions of strategy in strategic management typically emphasize its multidimensional nature: strategy involves consistent decisions about which
Dynamics of Competitive Repositioning
401
product markets to target, which input resources to acquire or develop, and how to effectively deploy those resources in selected markets in ways that generate competitive advantage. For example, Hofer and Schendel (1978) suggest four components of strategy: product–market scope, resource deployment and distinctive competencies, competitive advantages, and synergy (p. 25). Porter’s (1980) popular typology of business strategies classifies them in terms of scope and competitive advantage. Chen (1996) and Mehra (1996) characterize strategies in terms of market and resource dimensions, while Gimeno and Woo (1996) compare strategies in terms of served markets and strategic posture. Given the multidimensionality of strategy, much existing work on the dynamics of competitive repositioning provides at best an incomplete picture. The multidimensional nature of strategy and positioning was once explored by scholars examining strategic groups (Cool & Schendel, 1987; Fiegenbaum & Thomas, 1990; Ketchen, Thomas, & Snow, 1993). These scholars typically searched for ‘‘configurations’’ based on multiple-strategic dimensions, and assumed that these dimensions converged into discrete strategic groups or taxonomies. As a consequence, heterogeneity in multiple dimensions of strategy was collapsed into a simple measure of strategic distance (typically, whether firms belonged to the same or different strategic groups). However, this data-reduction approach caused significant debate within the literature, for scholars recognized that firms pursuing similar strategies might compete in different markets (Chen, 1996; Hatten & Hatten, 1987) or deploy different resources (Mehra, 1996). The main contribution of this paper is the examination of the dynamics of competitive repositioning at the business level in a way that reflects the multidimensional nature of the strategy construct. To this end, we characterize a firm’s strategy in three dimensions: a description of the markets targeted by the firm, its resource endowments, and the strategic posture for transforming those resources into advantageous positions in the served markets, all of which are interrelated (and often empirically confounded) but theoretically distinct. Firms can serve the same markets using different strategic postures and different resources. Firms can also deploy the same underlying resources in different ways, leading to different strategic postures in different markets. Our approach is to build an integrated model that explains the dynamics of competitive repositioning in, and the interplay among, multiple dimensions of firm strategy: served markets, resource endowments, and strategic postures. When evaluating a firm’s competitive positioning in these three dimensions, we define the firm’s competitive position relative to its reference population (e.g., Bothner, 2003; McPherson, 1983; Podolny, Stuart, & Hannan, 1996) and examine how firms change their relative competitive positioning
402
JAVIER GIMENO ET AL.
vis-a`-vis rivals over time. This approach is rooted in the view that an industry is a social organization in which firms define their roles in relation to other producers (Leifer, 1985; Porac, Thomas, Wilson, Paton, & Kanfer, 1995; White, 1981). Our approach views competitive positioning as a transformation process that organizes activities in a firm’s value chain to distinguish the firm from rivals (Porter, 1985; White, 1981), and it uncovers the nature of positioning by examining specific decisions made by firms in both input and output markets, as well as in the configuration of their activities. Specifically, the mechanism behind competitive repositioning is as follows: first, a firm’s decision in the product/factor market arises from the incentive to be aligned with the external context (in this case, the intensity of rivalry experienced by a firm and its learning from better-performing organizations). Second, a firm’s decision in one dimension of its strategy will lead to changes in other dimensions insofar as the internal alignment of various dimensions of the strategy improves the performance of the strategy as a whole. We assume that firms do change even in the short run, although change is subject to inertial forces and, therefore, is difficult and costly to implement. Moreover, firms are assumed to be purposeful on the direction of strategic change, even though their existing positions constrain the extent of possible change. Hence, random variation or unintended change is not considered here. We also assume that firms’ perception of their competitive environments is shaped by their relative position in the industry. Therefore, their perception of change opportunities tends to be local and influenced by interactions with similar organizations in that industry. This paper first develops our conceptualization of competitive repositioning in terms of three dimensions of pair-wise similarity between firms: served markets, resource endowments, and strategic posture. It then proposes hypotheses about how these changes are influenced by external and internal alignment processes. These hypotheses are tested with a longitudinal sample of firms in the deregulated U.S. scheduled airline industry (1979–1995). The paper ends with a discussion of the results and the implications for strategy and ecology researchers.
CONCEPTUAL FRAMEWORK Competitive Positioning in Strategy and Ecology: A Relative Pair-Wise Comparison Prior work on relative competitive positioning in strategy has drawn mainly on the strategic group concept, while ecology work has leveraged the concept
Dynamics of Competitive Repositioning
403
of the niche. Although there has been much debate about the nature and definition of strategic groups (Barney & Hoskisson, 1990), the focus was on grouping firms with similar strategies, defined in terms of market scope, resource deployment activities, and competitive advantage (Cool & Schendel, 1987; Fiegenbaum & Thomas, 1990). However, those dimensions used to define strategic group are not necessarily correlated with one another. First, although firms in the same strategic groups should have similar market scope, they do not necessarily compete in the same markets. After a careful review of the literature, Hatten and Hatten (1987) concluded that ‘‘as strategic group research develops, therefore, it is clear that groups do not have to be composed of competing companies’’ (p. 333). Similarly, Mehra (1996) found that clusters of banks formed by strategic scope dimensions and resource endowment dimensions did not correlate highly, and that resource endowment clusters explained firm performance better. This finding is consistent with the resource-based view, and Mehra (1996) concluded that ‘‘strategists need to include rivals in input markets who often are not their competitors in the output market’’ (p. 318). This insight that market and resource dimensions may not be directly related was further developed by Chen (1996) and Peteraf and Bergen (2003). In sum, while the traditional measures of strategic similarity are effective at describing similarities in strategic postures, they do not necessarily reflect whether firms use the same resources or target the same markets. Moreover, strategic group research, though providing an essential foundation for the study of competitive positioning, focuses on intergroup heterogeneity, and tends to ignore heterogeneity between firms within a group, as firms jockey for favorable competitive positions. To remedy this deficiency, competitive dynamics research (Baum & Korn, 1996; Chen & MacMillan, 1992; Gimeno, 1999) advances an action-response conception of competition and a pair-wise comparison of firms for competitor analysis. Consistent with this research stream, the current study takes a relative pair-wise view of competitive positioning. In contrast, the notion of the niche used to define competitive positioning in ecology research is based not on the type of strategic postures embedded in the competitive activities but on the type of resources (broadly conceived, which include also markets and customers) that firms draw on for their survival (Hannan & Freeman, 1989; McPherson, 1983). Given that the organizational form, either core or peripheral, states the profile of actions undertaken by the firm (e.g., Haveman & Rao, 1997), organizational forms in the ecology literature are equivalent to strategic postures, i.e., a set of competitive activities that ‘‘transform’’ input resources into advantageous
404
JAVIER GIMENO ET AL.
positions in the served markets. Yet when examining niche positions, the ecology literature focuses mostly on the type of resources used by the organization, relying on the assumption of duality between niche and form (Hannan & Freeman, 1986). Accordingly, two populations are assumed to occupy a similar competitive position when their niches overlap, i.e., they draw on the same resources and customers for their survival. In particular, much research on niche overlap has tended to focus on product markets (Baum & Korn, 1996; Bothner, 2003; McPherson, 1983), although niche overlap could also be defined by input resource or technology space (Stuart & Podolny, 1996). Compared with the measures of similarity in strategic group research, niche overlap measures provide reliable indication of who competes with whom for market or resource space. Yet these similarity measures may not describe the profile of a firm’s competitive activities, e.g., commitments and investment decisions responsible for a given niche width and the resulting niche overlap with other firms. Thus, these measures are largely silent about how firms come to develop their niche width by deploying their resources to serve their selected markets. In this regard, niche overlap measures are complementary to, not substitutes for, traditional measures of strategic similarity in strategic management.
Market, Resource, and Strategic Similarity The discussion above highlights two theoretical propositions. First, firms that follow similar strategies do not necessarily compete in the same market or with the same resource endowments. Second, niche position (which markets and resources are used by the firm) only partially describes the strategy of a firm; firms with high niche overlap may still pursue different strategic postures within their niche (Child, 1972). Both propositions imply that the duality of niche and form (Hannan & Freeman, 1986, 1989) may not be systematically observed in all markets, and they require researchers to theorize distinct contributions of resources (niche) and transformation activities (form) to the strategic interactions between a focal firm and other industry participants when examining competitive repositioning. One way to address these tensions is to use simultaneously measures that include similarity of strategic posture and niche overlap in served market and resource endowments. Following parallel research efforts by Chen (1996) and Gimeno and Woo (1996), we conceptualize relative competitive positioning in terms of three distinctive dimensions of interfirm similarity: market
Dynamics of Competitive Repositioning
405
similarity, which captures the degree of product market overlap between two firms; resource similarity, the extent to which two firms share common resource endowments; and strategic similarity, the degree of similarity between the strategic postures of two firms. Unlike prior research on niche overlap, our approach deconstructs the intersection of fundamental niches between two organizations into two distinct domains: market similarity and resource similarity. We then integrate these two domains further with strategic similarity in a bid to assess the dynamics of competitive repositioning. Market similarity refers to the extent to which a focal firm competes in the same specific markets with another firm, and is, therefore, associated with the degree of product market interdependence between the two firms. As a result, firms are highly dependent on the competitive actions and strategies of other firms with high market similarity (Baum & Singh, 1994; Chen, 1996; McPherson, 1983). Such mutual dependence can strongly influence a firm’s attack and response behavior in dealing with a rival with high market similarity (Chen, 1996; Chen & MacMillan, 1992). In addition, market similarity may influence a firm’s awareness of other firms in the industry, because firms are more likely to monitor the actions and strategies of those firms with which they compete in output markets (Chen, 1996; Greve, 1995, 1996). Therefore, market similarity is a critical dimension for understanding the relative position of a firm and its behavior with respect to another industry incumbent. In fact, the expectation of energetic response by rivals with high market similarity may influence firms to avoid unilateral competitive attacks on such rivals, a behavior detected in studies of mutual forbearance by multipoint competitors (Baum & Korn, 1996; Gimeno & Woo, 1996). Resource similarity refers to the extent that a firm has the same or similar endowments of tangible and intangible resources as another industry incumbent (Chen, 1996). The importance of this dimension is underscored by the resource-based view of the firm, which predicts that competitive advantage is linked to heterogeneous resource endowments (Wernerfelt, 1984). Similarity of resource profiles may serve as an indicator of firms with similar resources and capabilities (Amit & Shoemaker, 1993; Mehra, 1996; Stuart & Podolny, 1996). A firm is likely to be vulnerable to actions made by rivals with similar resources, since the use of common resources may leave them exposed to identical selection environments (e.g., Chen, Su, & Tsai, 2006). Strategic similarity defines the extent to which firms follow similar strategic postures or business models within the industry. These strategic postures represent transformation activities that determine how firms deploy resources to gain competitive advantage within certain markets. These strategic
406
JAVIER GIMENO ET AL.
postures describe dimensions of the activity system (Siggelkow, 2001) or value-chain activities (Porter, 1985, 1996) that allow firms to create superior value within certain markets by enhancing customer value or reducing costs. Because strategic postures reflect differences in the organization of activities, they closely represent the organizational form chosen by the firm (Hannan & Freeman, 1977, p. 935). The value of this concept is that two firms may differ in strategic postures even though they occupy the same market or resource niches. The concept of ‘‘strategic distance’’ in early strategic group research (Caves & Porter, 1977) also reflected these differences in strategic postures for firms that could pursue the same market positions. Accordingly, strategic similarity may influence the dynamics of competitive repositioning, independent of market similarity. This multidimensional conceptualization of strategy in terms of market, resource, and strategic similarity provides some obvious advantages over previous conceptualizations.1 Empirically, these constructs provide more finegrained information about the similarities and differences between a firm and another firm in the industry, and a better identification of niches. This is particularly relevant for describing interfirm relationships when the three dimensions of similarity are not closely aligned (e.g., when firms with similar strategic postures occupy different markets, or when firms with dissimilar resources or strategies compete head-to-head in the same markets). Theoretically, the multidimensional conceptualization is also important because each of those dimensions may underpin distinct aspects of competitive interactions and performance outcomes (Chen, 1996; Gimeno & Woo, 1996; Mehra, 1996).
DYNAMICS OF COMPETITIVE REPOSITIONING The ecology literature on strategic change proposes that changes in strategic positioning are constrained by inertial forces (Hannan & Freeman, 1984). The introduction of a new organizational form may be disfavored, at least in the short run, by a majority of internal and external stakeholders who are not convinced of the value of the new form, and it may be costly and difficul to implement. While empirical findings are inconclusive (Haveman, 1992; Kelly & Amburgey, 1991), many studies show the difficulty of changing the core of organizational form (Barnett & Carroll, 1995). This discussion suggests that competitive positioning with respect to strategic posture will be highly inert. As a result of extensive interdependencies in a firm’s value chain and activity systems, change in strategy posture is difficult and risky (Porter, 1996; Rivkin & Siggelkow, 2003).
Dynamics of Competitive Repositioning
407
Inertial forces also affect repositioning in resource and market niches. Hannan and Freeman (1984) argue that a firm’s core technology and the markets it serves are core features of the organization, and therefore subject to inertia. Tangible and intangible resource endowments are the result of years of accumulated investments, since they usually depreciate over long cycles. Even if companies modify the direction of new investment flows, the resource endowment stocks change very slowly (Dierickx & Cool, 1989). In terms of served markets, even when barriers to entry are relatively low (as is arguably the case in the airline industry), initial positions provide a revenue base that cannot be quickly changed without substantial organizational disruption (Dobrev et al., 2003). Despite inertia in competitive positioning, changes often arise from the pressures of external and internal alignments.2 Pressures from external alignment are those that help the company move toward attractive business opportunities or avoid environmental threats (e.g., Haveman, 1992). In particular, we focus our analysis on the role of rivalry as an important environmental threat and the role of interfirm performance differences to facilitate vicarious learning about opportunities. As discussed below, we argue that these forces act in localized ways: firms search for business opportunities in areas that are proximate to their current position, and they try to manage competitive differentiation relative to those firms that are initially most similar. Pressures for internal alignment, on the other hand, influence the position in one dimension of strategy relative to the other dimensions. A firm’s decision in one dimension of its strategy will lead to changes in other dimensions insofar as the internal alignment of various dimensions of the strategy improves the performance of the strategy as a whole (Porter, 1996; Siggelkow, 2001). Because these dimensions of strategy are interrelated, we assume that relative choice in one-dimension influences changes in the others. This could involve a convergence toward some configurations (i.e., firms similar in some dimensions becoming more similar in other dimensions). It may also, however, involve development of new configurations that are different from those of the more similar competitors. Thus, we examine when internal alignment pressures are likely to result in convergence or divergence among firms.
External Alignment Competitive Differentiation. A firm’s position relative to other incumbents determines the competitive interdependence with those firms and the pattern
408
JAVIER GIMENO ET AL.
of rivalry (Chen, 1996; Gimeno & Woo, 1996). If a firm is able to establish an idiosyncratic position in the industry, it may benefit from reduced competition from other firms and possibly improve its performance. On the other hand, if the firm occupies a similar position as other industry incumbents, it is dependent on the competitive actions of those incumbents and may suffer from lower performance if rivalry erupts (Porter, 1980).3 The benefits of differentiation may be obtained not only by serving different markets, but also by taking different strategic postures in those markets or deploying heterogeneous resources. Similarly, the ecology literature also conceptualizes competitive interdependence between firms in terms of niche overlap. As two organizations compete for common resources or markets, neither can realize its fundamental niche; they deplete their respective growth opportunities rapidly (Hannan & Carroll 1992; McPherson, 1983; McPherson & Rotolo, 1996). Although the focus on niche overlap is typically seen in product–market positioning, it should also exist in resource space and strategic posture. Other things constant, a firm is less dependent on competitors if it serves a distinct market, employs a different strategic posture, and draws on idiosyncratic resources. Mutual dependence per se is not negative. Highly dependent firms may learn to live and let live by developing strategies to avoid direct conflict (Baum & Korn, 1996; Edwards, 1955; Gimeno & Woo, 1996; McGrath, Chen, & MacMillan, 1998). Yet it is also likely that high mutual dependence aggravates the negative consequences of intense rivalry (Chen & MacMillan, 1992). Economic research of product differentiation has emphasized that a principal incentive for firms to maximally differentiate is to reduce the intensity of price rivalry (D’Aspremont, Gabszewicz, & Thisse, 1979). This prediction also agrees with organizational models of localized competition (Baum & Haveman, 1997; Baum & Mezias, 1992; McPherson, 1983). By reducing interdependence with respect to the most similar competitors, differentiation strategies are likely to increase expected performance and reduce performance volatility under conditions of high rivalry. Accordingly, firms facing intense rivalry in their served markets will likely reposition themselves in a way that reduces their interdependence (Baum & Singh, 1996; Dobrev et al., 2001, 2003). In other words, niche overlap aggravates competition between firms, which in turn seek to avoid competition by differentiating themselves from one another (McPherson, 1983; McPherson & Rotolo, 1996). Since a firm’s interdependence with another firm is an increasing function of the initial level of similarity between them, the firm faced with intense rivalry would reposition itself away from the most similar
Dynamics of Competitive Repositioning
409
competitors. Therefore, we predict that firms experiencing intense rivalry will respond by reducing market, strategic, and resource similarity relative to those firms that are most similar on those dimensions. H1a. A firm is more likely to reduce market similarity to market-similar firms if it is experiencing intense rivalry. H1b. A firm is more likely to reduce strategic similarity to strategy-similar firms if it is experiencing intense rivalry. H1c. A firm is more likely to reduce resource similarity to resource-similar firms if it is experiencing intense rivalry. Mimetic search. Changes in strategic positioning may not only reflect attempts to differentiate strategy under conditions of rivalry, but also the search for business opportunities in the industry (Haveman, 1992; Teece, Pisano, & Shuen, 1997). Yet the process of identification and targeting of opportunities is a social process embedded in the context of the initial position of a firm (Baum & Dutton, 1996; White, 1981). Firms learn about emerging opportunities in multiple ways, but few are more powerful and more convincing than observing better-performing competitors, i.e., those displaying better fitness in a given selection environment (Boyd & Richerson, 1985). This form of mimetic or vicarious outcome-based learning may occur at both conscious and subconscious levels. At a conscious level, managers seeking more effective strategic positions in the industry may infer those strategies by observing successful firms, particularly in a context of high uncertainty about the strategy-performance relationship (Cyert & March, 1963). Mimetic processes may also work at a subconscious level, as comparison with better-performing organizations may lead to consideration of strategic choices that would not be considered otherwise (Festinger, 1954; Greve, 1995). In both cases, firms will be likely to observe the strategic positions of better-performing organizations, particularly if those organizations are already sufficiently similar.4 The motivation to imitate better-performing organizations is constrained by the firm’s awareness of the strategies of the reference organizations and the viability of effectively imitating their position (Baum, Li, & Usher, 2000; Haunschild & Miner, 1997; Miller & Chen, 1994). Both the awareness of reference strategies and the ability to effectively imitate them are likely to increase with higher levels of similarity. Benchmarking against another firm is easier and more helpful if the companies are similar in markets, strategy, and resources, because performance differences can then be more narrowly interpreted. Moreover, because the dimensions of strategy are
410
JAVIER GIMENO ET AL.
interdependent and complementary, imitation of strategies that are very different is more likely to fail (Rivkin, 2000) due to the complexity, causal ambiguity, and implementation difficulties. On the other hand, if the betterperforming firm is initially similar, an imitation move can be implemented more easily. It follows that mimetic search leads to imitation that is both trait based, as firms search locally among competitors that are relatively similar in the three dimension of strategy, and outcome based, as firms focus on betterperforming competitors (Haunschild & Miner, 1997). Therefore, a focal firm would imitate the strategies of better-performing firms that are initially similar in the market, strategy, and resource dimensions. H2a. A firm is more likely to increase market similarity to market-similar firms if those firms have better performance than the firm. H2b. A firm is more likely to increase strategic similarity to strategysimilar firms if those firms have better performance than the firm. H2c. A firm is more likely to increase resource similarity to resourcesimilar firms if those firms have better performance than the firm. Internal Alignment The previous hypotheses examine how changes in the three dimensions of strategy respond to business opportunities in the environment (identified by better-performing firms) and competitive threats (reflected in rivalry pressures). The two dimensions of external alignment already pose some tensions, since moving closer to better-performing firms may also increase rivalry with stronger organizations. Another important tension is that posed by the multidimensional nature of strategy. That is, competition in factor and product markets is not independent. The benefits or drawbacks of increasing similarity in one dimension may be contingent on the similarity in other dimensions. If those dimensions have strong complementarities, an increase of similarity in one dimension will be most effective if the firm is also similar in other dimensions. This would lead toward the emergence of configurations or clusters of firms with multiple-aligned dimensions. On the other hand, it is also possible that the dimensions are divergent, and therefore firms that are similar in one dimension may compensate by decreasing similarity in another dimension. Prior research suggests that firms that compete in the same markets may not use the same strategies (Hatten & Schendel, 1977), and that firms serving
Dynamics of Competitive Repositioning
411
the same market may not have similar resource endowments (Chen, 1996; Mehra, 1996). Therefore, it is not clear whether all three dimensions of strategy will always (or even frequently) operate in concert and align into configurations of complementary dimensions. Indeed, the three dimensions that characterize a firm’s competitive positioning may interact in complex ways. Some dimensions may tend toward alignment and convergence with other dimensions, while others may tend to diverge. In this regard, there are several interesting anecdotal examples in the airline industry. Southwest Airlines, which today is perceived as the innovator of the ‘‘low-cost carrier’’ strategy, actually imitated much of its strategic profile from Pacific Southwest Airlines (PSA), but implemented it in a different market niche (Southwest in Texas, PSA in California). Today, most of the successful imitators of Southwest (Ryanair and easyJet in Europe) do so in markets that are not served by Southwest. Indeed, it would be difficult to imitate a leader’s strategy in the same markets where that leader is established. As a consequence, firms will try to develop unique combinations of strategy and market positions. From these examples, we expect that market similarity and strategic similarity will have divergent tendencies, since a way to avoid excessive rivalry in the target markets is by differentiating the strategy. Therefore, firms are more likely to differentiate their strategic posture if strategically similar firms compete in the same target markets. On the other hand, firms are more likely to imitate firms with similar strategies if they do not compete in the same markets. This perspective contrasts with views in strategy and organization theory that emphasize convergence among members of strategic groups or sociocognitive communities (Fiegenbaum & Thomas, 1995; Porac et al., 1995; White, 1981). For instance, Fiegenbaum and Thomas (1995) found that strategic group members tended to adjust their strategies toward the group’s reference point. However, this tendency toward convergence is problematic when firms also compete in the same product markets. In that situation, the result of competitive repositioning would be unfettered rivalry. Therefore, we suggest that although firms tend to benchmark their competitive repositioning relative to firms with similar positions, firms that have simultaneously high market and strategy similarity will actually reposition to reduce such convergence by reducing market similarity or strategic similarity; over time the firms either serve the same markets with different strategies or use similar strategies in different markets. Which one of these repositioning actions is more frequently taken (repositioning markets or repositioning strategy) remains an interesting empirical question.
412
JAVIER GIMENO ET AL.
H3a. A firm is more likely to reduce market similarity to market-similar firms if those firms are also similar in strategic postures. H3b. A firm is more likely to reduce strategic similarity to strategy-similar firms if those firms also occupy the same markets. Whereas the previous logic of differentiation suggests that market and strategic similarity will tend to diverge, the opposite is proposed for the strategic and resource dimensions. The resource-based view suggests a strong alignment between strategy and resources: an effective strategy is one that optimally deploys the endowments of tangible and intangible resources held by a firm and builds future resource endowments. This implies that strategic postures and resource endowments are heavily interdependent. We assume that competitive interactions are more intense in market space than in resource space, and therefore the incentives for competitive differentiation are strongest among firms pursuing the same target markets, as suggested by the previous hypotheses. The motivations for competitive differentiation would not be as strong when dealing with repositioning in strategic postures and resource endowments, since firms may not deploy their resources and strategies in the same target markets. As a result, firms would be more likely to monitor and imitate the strategy and resource endowments of firms that are initially similar in both strategy and resources. Over the time, this would lead to convergence toward certain configurations of strategy and resource commitments in the industry. H4a. A firm is more likely to increase strategic similarity to strategysimilar firms if those firms also have similar resource endowments. H4b. A firm is more likely to increase resource similarity to resourcesimilar firms if those firms are also similar in strategic postures.
METHODS Sample The paper studies the dynamics of competitive repositioning in the context of the post-deregulation U.S. scheduled airline industry from 1979 to 1995. The Airline Deregulation Act of 1978 substantially modified the ability of airlines to change their competitive positioning in all three dimensions of served market, strategic posture, and resource endowments. Prior to deregulation, entry and exit from markets had to be approved by the Civil Aviation Board,
Dynamics of Competitive Repositioning
413
making it harder for firms to alter their market scope. The regulatory constraints also made entry difficult. Airlines were constrained in their pursuit of differentiated strategic postures, leading to substantial strategic homogeneity before deregulation. By focusing on the window starting immediately after deregulation, our study captures the competitive (re)positioning choices of established airlines and new entrants. The idiosyncratic characteristics of the airline industry provide a convenient setting for studying the dynamics of competitive repositioning. Changes in served markets were frequent (due to the relatively low costs of market entry and exit), and served markets were easy to track empirically. The industry is divided into multiple clearly defined markets (city-pair markets), with different firms serving different sets of these markets. The major capital commitment of an airline, its aircraft fleet, can be redeployed across markets in response to market incentives. In fact, proponents of contestability theory (Baumol, Panzar, & Willig, 1982) claimed that airline markets were nearly perfectly contestable.5 While ensuing empirical research has not supported the full predictions of that view (Graham, Kaplan, & Sibley, 1983; Hurdle, Johnson, Joskow, Werden, & Williams, 1989; Strassmann, 1990) – mainly due to entry barriers created by commitments to ground facilities and the development of hub-and-spoke systems – the costs of market entry in the airline industry are nevertheless substantially lower than in other industries in which resources are committed to specific markets and cannot be redeployed easily. The post-deregulation airline industry was also characterized by substantial intra-industry heterogeneity in strategic postures (Chen, 1996; Gimeno & Woo, 1996). Large firms with hub-and-spoke systems, such as American and United, found themselves competing with smaller firms using point-topoint traffic systems, such as Southwest and ValuJet. Some firms specialized in short routes with high-frequency service, while others focused on long routes with lower frequency, usually by using one-stop flights through a hub (Levine, 1987). The airline industry was also characterized by little diversification outside the industry, which controlled for the possibility of influences from outside the industry. The definition of a market in the airline industry corresponds to the citypair market: the demand for air transportation between two U.S. cities. City-pair markets have little demand cross-elasticity, because passengers wanting to fly a particular city-pair market would be unlikely to see another city-pair market as a substitute service. For the purpose of this study, we considered only city-pair markets in which both end cities were FAA-classified hubs: cities with passenger emplanements of at least 0.05% of the
414
JAVIER GIMENO ET AL.
overall U.S. traffic. Thus, our sample excluded relatively small cities, usually served by commuter airlines. Our sample included traffic between 134 cities. We also eliminated city-pair markets that had insignificant traffic (fewer than 10 passengers per day), thereby reducing the analysis to meaningful markets, and very short city-pair markets (less than 100 miles), to control for the substitution effect of ground transportation. Again, this eliminated most commuter airlines from our sample. Overall, the sample covered the markets of 54 airlines in 4,994 domestic city-pair markets over 17 years (1979–1995).6 We obtained data from four longitudinal databases with quarterly frequency collected by the Department of Transportation. The ticket price origin and destination survey data (DB1A data set) is a 10% sample of the tickets sold in the U.S., and includes information about the itinerary of the passenger, the airline used, and the price paid. A representative quarterly DB1A database contains millions of records. This information was aggregated to determine which companies were incumbent in which markets in each quarter, market share in those markets, and other firm-market-specific information. An airline was considered an incumbent if it had at least a 5% market share or carried at least 10 passengers per day. The service segment data sets provided the complete population of flight segments flown by an airline in a quarter, including the load factors (percentage of filled seats), frequency of flights, and total passengers transported in each segment. Since a city-pair market can be served by a combination of segments (through one-stop service), the service segment data provided the network structure of the firm supply used to fulfill city-pair market demand. We also used the schedule T-3 data, a report of the operations of each airline in each city, to determine the fleet structure of each airline. For each airline and city, the schedule T-3 provided a breakdown of departures with each model of aircraft, using a very detailed classification system of aircraft models. Aggregating this data at the firm level provided us with a detailed report of the type of aircraft used by each airline and the number of departures for each aircraft type. Finally, we used Form 41 accounting reports to obtain quarterly balance sheet and profit statement information for the airlines. All data sets were reported at a quarterly frequency. Since we were interested in assessing long-term repositioning dynamics rather than short-term seasonal adjustments, only data for the fourth quarter of every year (1979–1995) was used, with the exception of performance, which was measured yearly. The unit of analysis was the ordered firm-pair, and each firm-pair was observed over multiple years. One observation described the position of a focal firm relative to another firm for a particular year. Since our emphasis
415
Dynamics of Competitive Repositioning
was on the dynamics of pair-wise similarity, we considered only firm pairs in which both firms were active for at least two overlapping periods. The number of ordered firm-pairs in the sample was 1,729.7 The variables in the empirical model represent pair-wise variables: variables that describe the competitive position of a focal firm i relative to another firm j in the sample, at a period of time t. Subscripts i, j, and t are used throughout to represent the focal firm, other incumbent and time period, respectively. Independent Variables Market similarity was defined as the overlap between the focal firm and another firm in the multiple city-pair markets in the industry. Thus, we examine situations in which the focal firm and the other firm compete in the same markets. Following the ecology literature, we used a cosine measure of niche overlap (Hannan & Freeman, 1989; McPherson, 1983; Sohn, 2000). Specifically, we measured the extent of overlap in markets between the focal firm and the reference firm by weighting these contacts by the percentage of revenues obtained by the focal firm in that market. This measure is also used in the multimarket competition literature as the ‘‘salesat-risk’’ measure (Alexander, 1985; Feinberg, 1984, 1985). Empirically, the measure of market similarity between two firms for a given time period was constructed as 4994 P
Market Similarityði;t; j;tÞ ¼ m¼1
wimt min Iimt; Ijmt 4994 P
wimt Iimt
m¼1
where Iimt and Ijmt were dummy variables indicating whether firms i and j were present in market m at period t, and wimt represented the percentage of firm’s i revenues obtained in market m at period t. In this original scale, the measure ranged from 0 to 1, and was not symmetric – a small firm might have a large overlap with a large firm, while the large firm would have small overlap with the small firm. Resource similarity captured the similarity in the resource endowments of two airlines. As suggested by Chen (1996, p. 119), the fleet structures of airlines were vital investments for those companies. Although airplanes were tradable resources, the high cost and long life of these assets made them very ‘‘sticky’’ resources for the airlines. Changing the fleet structure required a very significant long-term commitment, not only in purchase costs but also
416
JAVIER GIMENO ET AL.
in retraining existing pilots and maintenance personnel. In addition, a firm’s fleet structure was substantially specialized to certain service applications. For instance, wide-body aircrafts with large passenger capacity and long ranges – such as the Boeing 747 – which were most efficient in providing point-to-point service on long-distance routes, were very expensive to operate on shorter routes and in hub-and-spoke systems, which were more efficiently served by planes such as the Boeing 737 (Bailey, Graham, & Kaplan, 1985, p. 50). Our measure of resource similarity described the similarity in fleet structure between a focal firm and another industry incumbent. From the schedule T-3 database we obtained the distribution of departures performed by an airline in a quarter based on our classification of aircraft in 26 homogeneous types of aircraft.8 For each firm i and time period t, we calculated the percentage of overall departures performed with each type of aircraft r, represented as airt. The measure of resource similarity was based on the Euclidean distance between the vectors reflecting the firm’s reliance on each type of aircraft, with a negative sign so that a lower distance represented greater resource similarity. Empirically, vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 26 uX Resource Similarityi;t; j;t ¼ t ðairt ajrt Þ2 r¼1
Strategic similarity captured the similarity in the strategic postures of two firms in the industry. Following Gimeno and Woo (1996), we identified several decision domains that described the strategic posture of an airline at a period of time. In the airline industry, firms’ strategic postures are mostly aimed at organizing their value chain activities so as to implement costdifferentiation positioning within those markets. In that regard, the choice of market distance was critical, since it influenced the type of aircraft that most efficiently served those markets and the ability to use hub-and-spoke networks (one-stop flights typically are less effective in short-distance markets). For instance, Southwest differed from most of the major airlines in its preference for short-distance markets. Some companies could also differentiate by the type of destinations, which influence the price sensitivity of customers and the ability to differentiate services. Companies such as American Trans Air and Carnival Air Lines specialized in tourist destinations. It should be noted that multiple firms could serve markets with similar characteristics (which would contribute to high strategic similarity) without serving the same specific target markets (which would keep market similarity low).
417
Dynamics of Competitive Repositioning
In terms of the operational network structures, firms chose either pointto-point systems (in which most passengers flew nonstop from origin to destination) or hub-and-spoke systems (in which most passengers changed planes in the firm’s hub). Some companies, especially those using point-topoint systems in highly dense markets, could differentiate themselves by offering a high frequency of daily flights. Seeking competitive advantage, some firms used low prices to attempt to dominate the markets they served. Other companies positioned themselves as niche players in dense markets, providing an exclusive service at a higher price. For example, MGM Grand Air operated in a small exclusive niche in the Los Angeles–New York market, charging prices well above those of other airlines. Southwest, on the other hand, was commonly the price leader and tended to dominate the markets it served. To measure differences in strategic posture, we created for each firm and year a vector of seven variables describing the above-mentioned decisions. This vector was calculated by obtaining a passenger-weighted average across the markets served by the airline of the following characteristics: (1) market density (demand size), (2) market distance, (3) whether the market was a tourist market (dummy variable),9 (4) percentage of direct flights, (5) average daily frequency, (6) market share, and (7) difference between the firm’s price per mile in the market and the market-average price per mile. Our measure of strategic similarity was based on the Euclidean distance between those vectors after standardization. The sign of the variable was changed so that higher values reflected more similar strategic postures. Empirically: vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 7 uX Strategic Similarityði;t; j;tÞ ¼ t ðzidt zjdt Þ2 , d¼1
where zidt represented firm i’s standardized level in strategic decisions d (d ¼ 1, y ,7) at time t. The rivalry experienced by firm i at time period t was calculated from a first-stage longitudinal analysis of market prices in the 4,994 markets. The rivalry measure showed whether the average market prices in the markets served by an airline were higher or lower than what would be expected under ‘‘normal’’ competitive conditions. To identify the benchmark for ‘‘normal’’ competition, the average price per mile (yield) in each market and year was regressed on several variables describing exogenous market characteristics that influence costs, such as market distance, scale of operations in the airports and the city-pair market, and input prices. These exogenous variables were market distance, distance squared, the scale category of the
418
JAVIER GIMENO ET AL.
origin and destination cities (large, medium, or small), the interaction between origin and destination city scales (which determine the scale of operations between the cities), and the interaction between distance and distance squared and the origin and destination city scales, plus a series of year dummy variables. These variables were exogenous in that they influenced market structure and competition, but were not influenced by them. The regression had 62,930 market-year observations and 42 independent variables (mostly dummy variables), and the R2 was 0.58.10 The predicted value of that regression was interpreted as the benchmark for ‘‘normal’’ prices under typical competitive conditions for a particular market. The residual of the regression therefore indicated whether the average prices in a market were higher or lower than the benchmark. To simplify interpretation, the residuals were multiplied by -1 so that higher values reflect higher rivalry. The rivalry experienced by a focal firm i, Rivalryit, was calculated as the passenger-weighted average of these negative residuals over all the markets served by the airline at time t. Note that our measure of rivalry reflects diffuse competition in the markets served by an airline, not direct competition with another rival. In this regard, this measure should be distinguished from that of market similarity, which indicates direct competitive overlap between two firms. Note also that we do not use the population density to measure diffuse competition but use the ‘‘below-the-normal’’ price. The reason is that as shown in the economic literature, the latter is largely redundant to the former in many observable markets. The performance gap between firm j and firm i was calculated as the difference in return on sales (ROS) between firm j and firm i, so that a positive value meant that firm j had higher performance than firm i in period t. Return on sales was defined as operating profit over total revenues. Since mimetic behavior was likely to lead to imitation only when the reference firm performed better than the focal firm, the variable was set to 0 when firm j’s performance was below firm i’s. In contrast with other measures in this paper that used only fourth-quarter data, ROS was measured over the whole calendar year to avoid the effects of year-end accounting charges. Empirically, Performance Gapðj;t; i:tÞ ¼ MaxðROSjt
ROSit; 0Þ.
Statistical Methodology Because we were interested in the dynamics of similarity across firms, a longitudinal dynamic design was required. A dynamic model uses lagged
419
Dynamics of Competitive Repositioning
values of the dependent variable to explain future values, and therefore focuses on changes in the variable over time. Yet modeling the dynamics of pair-wise similarity posed unique methodological challenges with respect to the specification of change and correlated error terms. Changes in similarity scores between two firms between period t and period t+1 do not provide unequivocal evidence about which firm changed, since changes can be influenced by either the repositioning of the focal firm or the repositioning of the reference firm. For instance, a focal firm might have been actively trying to imitate the served markets dimension of a reference firm by entering that firm’s markets, yet the market similarity might not have increased if the reference firm was simultaneously repositioning itself away from the focal firm. Since we are concerned with whether a focal firm increased or decreased similarity with other firms, we used a different measure of change. Specifically, we modeled the future change in pair-wise similarity implemented by the focal firm relative to the current-year position of the reference firm. Measured in this way, only the focal firm’s repositioning initiative would affect change. For example, to assess the change in market similarity, we compared Market Similarity(i,t+1,j,t) to Market Similarity(i,t; j,t). Differences between these two scores could be attributed only to changes in the market scope of firm i between period t and t+1 that increased or reduced the overlap with firm j, and not to changes in the market scope of firm j. Accordingly, the dependent variables were operationally defined as:
Market Similarityði;tþ1; j;tÞ ¼
Resource Similarityði;tþ1; j;tÞ ¼
4994 P
wim;tþ1 minðIim;tþ1 ; Ijmt Þ
m¼1 4994 P
wim;tþ1 Iim;tþ1 m¼1 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 26 uX t ða ajrt Þ2 ir;tþ1 r¼1
Strategic Similarityði;tþ1; j;tÞ ¼
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 7 uX t ðz zjdt Þ2 id;tþ1 d¼1
As noted above, the unit of analysis was the ordered ‘‘firm-pair’’, not an individual firm. That is, we did not aggregate these similarity measures at the individual firm level to uncover an attribute of the firm. Rather, we detected the distance between two firms in the Euclidean space spanned by
420
JAVIER GIMENO ET AL.
each of the three dimensions of strategy. In so doing, we treated each pair equally in this Euclidean space, independent of the size of the firms or the number of markets served. Since these equations included lagged dependent variables, interpretation of the model followed the standard interpretation of autoregressive models (Greene, 1997). In particular, the coefficient of the lagged-dependent variable reflected how quickly the dependent variable adapted to changes in independent variables or random shocks. Equilibrium considerations required that the coefficients of the lagged dependent variables should be between 0 and 1, with 0 representing quick adaptation and 1 representing high inertia. In addition, all the independent variables were lagged relative to the dependent variable, to avoid concerns about reverse causation. The hypotheses suggested interactions between lagged dependent variables and other independent variables. Interpretation of main and interaction effects was facilitated by centering the variables before calculating interactions (Aiken & West, 1991). This transformation allowed us to interpret the main effects as the marginal effects of a variable when the remaining interacting variables were at their mean levels. Since the different similarity variables have different scales, interpretation was also facilitated by standardizing all variables prior to the calculation of the interaction effects.11 These transformations facilitated interpretation and reduced multicollinearity in the empirical model. A serious estimation issue of modeling pair-wise similarity was the lack of independence of errors, or what network analysts refer to as the ‘‘common actor’’ problem (Lincoln, 1984). Models with dyadic relational data include multiple observations for each focal firm and reference firm. Errors in such models are unlikely to be independent, since they would represent firmspecific characteristics for each of the common actors. Repositioning decisions by firms relative to a reference firm might also affect position relative to other firms. For instance, a firm increasing similarity with a reference firm would also increase similarity to firms that were similar to the reference firm. We proposed two estimation solutions to the ‘‘common actor’’ problem. The first was to create a control for network autoregression. The second was to treat the issue as a misspecification problem (unobserved heterogeneity) leading to correlated errors among dyadic observations that had common actors, and allow a correlated error structure that modeled those correlated errors. In the first approach, we created a control for spatial autoregression as specified in the literature on spatial statistics and network autocorrelation (Cliff & Ord, 1981; Doreian, 1981). Spatial autoregression takes place because increasing similarity toward a reference firm also increases similarity
421
Dynamics of Competitive Repositioning
to other firms that are similar to the reference firm. These observations of dyadic similarity were not independent. To control this effect, a spatial autoregressive term was constructed. For a focal firm i and reference firm j, the spatial autoregressive term captured the average change in similarity that focal firm i had with all other firms in the sample, weighted proportionally to the distance between those firms and the reference firm j. If a focal firm increased similarity with respect to a reference firm simply because the firm had increased similarity to other firms proximate to the reference firm, then the spatial autoregressive term would be large and would capture the spurious effect. For a given similarity-dependent variable Y (where Y could represent market, strategic, or resource similarity), the spatial autoregressive term was calculated as: P wjk ðYði;tþ1; k;tÞ Yði;t;k;tÞ Þ kai;j P WYi;t; j;t ¼ , wjk kai;j
and
wjk ¼ 1=2 ðYðj;t; k;tÞ þ Yðk;t; j;tÞ Þ.
The second approach was to interpret the ‘‘common actor’’ problem as an error in specification due to unobserved heterogeneity associated with firmand pair-specific effects. In a similar context, Korn and Baum (1999) include two random effects in the specification of the error term to account for correlated errors among observation where the focal firm or reference firm are common actors. These random effects would allow correlations among error terms of observations that shared the same focal firm or reference firm. We went beyond that approach by specifying that common actor effects might not be constant over time. For example, their specification would imply that the correlation between the errors e(i,t;j,t) and e(i,t;k,t) was the same as that between errors e(i,t;j,t) and e(i,t0 ;k,t0 ). It is, however, more likely that correlations will be higher if the errors arise from contemporaneous observations. A richer specification of the error component model should take into account that the strongest correlations should exist among contemporaneous observations that share the same common actors, since these would reflect the same underlying repositioning moves by the firms. Empirically, we specified an error component model as follows: ði;t;j;tÞ ¼ mi þ ðmtÞit þ jj þ ðjtÞjt þ eði;t;j;tÞ , where (mt)it and (jt)jt represented the time-varying random effects for the focal and reference firm, respectively. According to this specification, for
422
JAVIER GIMENO ET AL.
instance, the correlation between e(i,t;j,t) and e(i,t;k,t) would equal ðs2m þ . 2 2 smt Þ ðsm þ s2mt þ s2j þ s2jt þ s2e Þ;.while the correlation between e(i,t;j,t) and e(i,t0 ;k,t0 ) would equal only ðs2m Þ ðs2m þ s2mt þ s2j þ s2jt þ s2e Þ: These models were estimated using the restricted maximum likelihood (REML) estimation, built in Proc MIXED procedure in SAS. In addition, in order to provide a benchmark for comparison of the effects of these statistical techniques, we also presented the models estimated without random effects (i.e., OLS estimation).
RESULTS Table 1 presents the descriptive statistics and correlation matrix of the variables in the study. Tables 2–4 present the results for the analysis of market, strategic, and resource similarity, respectively. All the independent variables were one-year lagged in a way that controls for reversed causality. Each table displays a baseline OLS model as a benchmark for comparison. The tables present two models for hypotheses testing: a model with random effects for focal and reference firm and with a spatial autoregressive term, and a random effects (or error components) model with constant and timevarying random effects for focal and reference firm. Because these models are nonnested models with different statistical assumptions, there are no easy statistics for comparing their relative support. However, one measure of fit to compare these nonnested models is the variance of the residual (s2e), which is lower for better-fitting models. The correlation table (Table 1) shows that all three dimensions of pairwise similarity (market, strategic, resources) had very high inertia. The level of correlation between similarity between two firms in period t and in period t+1 ranged from 0.96 to 0.97 for all three measures of similarity. These coefficients are akin to autocorrelation coefficients. Thus, the extent of change in similarity between two firms is very small relative to the range of the similarity variables across pairs. As expected, the evidence confirmed that firms changed their competitive positions only very slowly. The three dimensions of similarity were mildly correlated (ranging from 0.20 to 0.38). The strongest correlations were between strategic and resource similarity, the weakest ones between market and strategic similarity. These bivariate results confirmed the underlying assumption of the paper: that competitive position is a multidimensional construct. The correlations of the similarity measures with rivalry and performance gap were relatively low, below 0.12 in absolute values.
423
Dynamics of Competitive Repositioning
Table 1. Variable [1] Market similarity (i,t+l; j,t) [2] Market similarity (i,t; j,t) [3] Strategic similarity (i,t+1; j,t) [4] Strategic similarity (i,t; j,t) [5] Resource similarity (i,t+1; j,t) [6] Resource similarity (i,t; j,t) [7] Rivalry (i,t) [8] Performance gap (j,t; i,t)
Descriptive Statistics and Correlations. Mean Standard Deviation [1] 0.00 0.01 0.03 0.02 0.03 0.01 0.01 0.00
0.98 0.99 0.94 0.98 0.98 1.00 1.00 1.00
[2]
0.97 0.23 0.23 0.20 0.20 0.32 0.31 0.32 0.31 0.03 0.02 0.06 0.05
[3]
[4]
[5]
[6]
0.96 0.38 0.36 0.05 0.06
0.38 0.36 0.07 0.07
0.96 0.12 0.07
0.11 0.08
[7]
0.09
N ¼ 6,947.
Hypotheses 1a–1c stated that firms experiencing intense rivalry tend to increase differentiation relative to the most similar firms in the three strategic dimensions. These hypotheses implied that the interaction of lagged similarity with rivalry had a negative effect on future levels of similarity (i.e., similarity decayed faster among highly similar firms when they were exposed to high rivalry). The relevant coefficients were negative and significant in Tables 2–4. The results in all three tables provided consistent evidence in support of hypotheses 1a–1c at significance levels of 0.05 or above. The evidence was consistent across alternative model specifications, and was stronger for market and resource similarity dimensions than for strategic similarity. As an illustration, a firm with high market similarity with a rival would decrease that similarity by 7.6% each year if rivalry was low (one standard deviation below the mean), but would decrease it by 9.5% each year if rivalry was high (one standard deviation above the mean). The evidence supported the view that, despite high levels of inertia in positioning, firms exposed to intense rivalry incrementally changed their strategic position in order to differentiate from very similar firms. Hypotheses 2.a–2.c stated that firms were likely to increase similarity with respect to better-performing firms that were initially similar, i.e., positive interaction between lagged similarity and performance gap. We found support for the hypotheses relative to market similarity and resource similarity (Tables 2 and 4, respectively), but not for strategic similarity (Table 3). The results suggested that firms were more effective in imitating the market and resource dimensions of similar, yet better-performing firms. As an illustration, a firm with high resource similarity with another would reduce that similarity by 8.6% each year if that firm was low-performing (one standard deviation below the mean), but would reduce it by 6.3% if that firm was
424
Table 2. Dependent Variable: Market Similarity (i ,t+1;
Hypotheses
Analysis of Dynamics of Market Similarity. Model 1a
Model 1b
Model 1c
Benchmark: OLSa
Spatial autoregressive modelb
Error components modelc
j,t)
Coeffecient
Coefficient
(Standard error) (0.0102)
0.9505
0.9307
(0.0036)
0.9142
(0.0039)
0.0013 0.0037 0.0000 0.0227
(0.0029) (0.0029) (0.0032) (0.0033)
0.0008 0.0044 0.0075 0.0154
(0.0029) (0.0029) (0.0036) (0.0032)
0.0052 0.0073 0.0062 0.0166
(0.0057) (0.0047) (0.0049) (0.0036)
H1a( )
0.0101
(0.0031)
0.0101
(0.0029)
0.0096
(0.0030)
H2a(+)
0.0059
(0.0025)
0.0054
(0.0024)
0.0068
(0.0025)
H3a( )
0.0110
(0.0027)
0.0087
(0.0026)
0.0105
(0.0028)
0.0262
(0.0035)
0.0200
(0.0034)
0.0233
(0.0035)
0.4458
(0.0168)
0.0038
(0.0030)
JAVIER GIMENO ET AL.
0.0126
(0.0033)
Rivalry (i,t) Performance gap (j,t; i,t) Strategic similarity (i,t; j,t) Resource similarity (i,t; j,t)
Spatial autoregressive term (W market similarity)
(Standard error) (0.0072)
(i,t; j,t)
Market similarity (i,t; j,t) rivalry (i,t) Market similarity (i,t; j,t) performance gap (j,t; i,t) Market similarity (i,t; j,t) strategic similarity (i,t; j,t) Market similarity (i,t; j,t) resource similarity (i,t; j,t)
Coefficient
0.0218
Constant Market similarity
(Standard error)
0.0000 0.0019
s2j s2jt s2e R2 F REML log-likelihood
0.0494
0.0486
0.9414 12400.84 498.21
Ordinary least squares (OLS) regression. Random effects for focal and referent firm effects, and spatial autoregressive term. c Random effects for stable and time-varying effects for focal and referent firms. b
0.0026 0.0015
0.0564
(po0.05). (po0.01). (po0.001) a
0.0008 0.0042
s2mt
304.25
Dynamics of Competitive Repositioning
s2m
425
426
Table 3. Dependent Variable: Strategic Similarity (i,t+1;
Hypotheses
Model 2a
Model 2b
Model 2c
Benchmark: OLSa
Spatial autoregressive modelb
Error components modelc
j,t)
0.0124
(0.0035)
Coefficient
(Standard error)
Coefficient
(Standard error)
0.0139
(0.0239)
0.9237
(0.0036)
0.9642
(0.0030)
0.9343
(0.0040)
0.0022 0.0035 0.0228 0.0169
(0.0036) (0.0034) (0.0035) (0.0036)
0.0001 0.0005 0.0041 0.0016
(0.0026) (0.0022) (0.0025)y (0.0023)
0.0161 (0.0040) 0.0004 0.0016
H1b( )
0.0254
(0.0039)
0.0050
(0.0024)
0.0069
(0.0026)
H2b(+)
0.0002
(0.0036)
0.0000
(0.0021)
0.0012
(0.0025)
H3b( )
0.0516
(0.0028)
0.0004
(0.0018)
0.0011
(0.0020)
H4a(+)
0.0242
(0.0035)
0.0026
(0.0021)
0.0057
(0.0022)
0.9471
(0.0088)
Rivalry (i,t) Performance gap (j,t; i,t) Market similarity (i,t; j,t) Resource similarity (i,t; j,t)
(0.0173) (0.0053) (0.0028) (0.0026)
JAVIER GIMENO ET AL.
Spatial autoregressive term (W strategic similarity)
(Standard error)
(0.0064)y
(i,t; j,t)
Strategic similarity (i,t; j,t) rivalry (i,t) Strategic similarity (i,t; j,t) performance gap (i,t; j,t) Strategic similarity (i,t; j,t) market similarity (i,t; j,t) Strategic similarity (i,t; j,t) resource similarity (i,t; j,t)
Coefficient
0.0121
Constant Strategic similarity
Analysis of Dynamics of Strategic Similarity.
s2mt s2j s2jt s2e R2 F REML log-likelihood
0.0399 0.0016 0.0239
0.0234
2987.50
2395.16
0.9188 8724.43
(po0.1). Ordinary least squares (OLS) regression. b Random effects for focal and referent firm effects, and spatial autoregressive term. c Random effects for stable and time-varying effects for focal and referent firms. a
0.0030 0.0005
0.0724
(po0.05). (po0.01). (po0.001). y
0.0160
Dynamics of Competitive Repositioning
0.0002y
s2m
427
428
Table 4. Dependent Variable: Resource Similarity (i,t+1;
Hypotheses j,t)
Model 3c Error components modelc
(Standard error)
Coefficient
0.0242
(0.0036)
0.0203
(0.0073)
0.0116
(0.0118)
0.9221
(0.0037)
0.9594
(0.0026)
0.9255
(0.0036)
0.0109 0.0101 0.0119 0.0448
(0.0041) (0.0034) (0.0037) (0.0037)
0.0085 0.0045 0.0123 0.0130
(0.0040) (0.0026)y (0.0028) (0.0033)
0.0069 0.0094 0.0143 0.0249
(0.0096) (0.0064) (0.0038) (0.0051)
Hlc( )
0.0084
(0.0037)
0.0119
(0.0025)
0.0139
(0.0036)
H2c(+)
0.0064
(0.0034)y
0.0084
(0.0022)
0.0118
(0.0030)
0.0228
(0.0038)
0.0105
(0.0025)
0.0162
(0.0033)
0.0250
(0.0035)
0.0234
(0.0023)
0.0295
(0.0030)
1.1109
(0.0112)
H4b (+)
(Standard error)
JAVIER GIMENO ET AL.
Coefficient
Rivalry (i,t) Performance gap (j,t; i,t) Market similarity (i,t; j,t) Strategic similarity (i,t; j,t)
Spatial autoregressive term (W resource similarity)
Model 3b Spatial autoregressive modelb
(Standard error)
(i,t; j,t)
Resources similarity (i,t; j,t) rivalry (i,t) Resource similarity (i,t; j,t) performance gap (j,t; i,t) Resource Similarity (i,t; j,t) Market Similarity (i,t; j,t) Resource Similarity (i,t; j,t) Strategic Similarity (i,t; j,t)
Model 3a Benchmark: OLSa
Coefficient
Constant Resource similarity
Analysis of Dynamics of Resource Similarity.
s2mt s2j s2jt s2e R2 F REML log-likelihood
0.0272 0.0013 0.0279
0.0443
0.9231 9257.39 2446.00
(po0.1). Ordinary Least Squares (OLS) regression. b Random effects for focal and referent firm effects, and spatial autoregressive term. c Random effects for stable and time-varying effects for focal and referent firms. a
0.0019 0.0017
0.0740
(po0.05). (po0.01). (po0.001). y
0.0000
370.62
Dynamics of Competitive Repositioning
0.0009
s2m
429
430
JAVIER GIMENO ET AL.
high-performing (one standard deviation above the mean). However, firms did not appear to significantly change their strategic posture in response to performance differences. This result suggests that inertia in strategic posture may be higher than in other dimensions. The evidence provided partial support for the hypotheses, since we observe convergence in niche and resource positions toward better-performing proximate firms. Hypotheses 3a and 3b respectively stated that the relation between market and strategic similarity was divergent, since convergence in these two dimensions increased the intensity of rivalry. Thus, we proposed that firms were likely to decrease market and strategic similarity when the initial levels of market and strategic similarity were simultaneously high. Looking first at the effects on market similarity (hypothesis 3a), we found consistent strong support for the hypothesis that firms differentiated served markets relative to reference firms that were initially similar in both market and strategy dimensions. A firm would reduce market similarity by 7.5% in a year relative to a rival with low strategic similarity (one standard deviation below the mean), but 9.6% in a year relative to a rival with high strategic similarity (one standard deviation above the mean). On the other hand, we rejected hypothesis 3b, since strategic similarity did not systematically change for those firms that were initially similar in strategy and market position. Accordingly, we concluded that, while firms were trying to avoid excessive alignment with competitors in both market and strategy dimensions, they achieved their objective mainly by modifying served markets (exiting markets where strategy-similar competitors were positioned, entering markets in which incumbents were strategically different) rather than by changing their strategy postures. Again, this evidence contributes to the evidence that firms experience greater inertia for changes in strategic postures than for changes of market or resource positions. Finally, hypotheses 4a and 4b predicted that firms would converge to the strategic and resource positions of firms that were initially similar in these dimensions since these two dimensions are complementary. The results provided mixed support for the repositioning based on strategic similarity (hypothesis 4a), but strong support for that based on resource similarity (hypothesis 4b). In Table 3, the coefficient of the interaction between strategic similarity and resource similarity was significant in some specifications, but not in others. We concluded that the support for hypothesis 4a was mixed. On the other hand, the results provided consistent, strong support for convergent repositioning in terms of resource endowments (hypothesis 4b). Firms were generally more likely to imitate the resource endowments of other firms following a similar strategy, and the effect was stronger when
Dynamics of Competitive Repositioning
431
both strategy and resources were initially similar. Firms with high initial resource similarity (one standard deviation above the mean) would reduce resource similarity by 12.9% per year if the firms had low strategic similarity (one standard deviation below the mean), but by only 2% if the firms had high strategic similarity (one standard deviation above the mean). Again, it appears that firms achieved internal alignment of strategy-resource profiles by adapting their resource endowments rather than changing their strategy. This evidence was consistent with previous results of greater inertia associated with changes in strategic similarity.
DISCUSSION AND CONCLUSION Taking a multidimensional view of competitive positioning, this paper considers interfirm competition in three dimensions: output markets, resource endowments, and strategic postures. In so doing, the study contributes to both strategic management and ecology research. First, it offers a multidimensional conceptualization of competitive strategy, in direct contrast to previous one-dimensional considerations. Second, it highlights the multifaceted, complex interrelationship among the three dimensions of competitive positioning. Third, it advances a relative, pair-wise view for examining competitive repositioning and offers strategic insights at the firm-dyad level that are unavailable at the group or population level. This paper investigated the dynamics of firms’ competitive positioning relative to other industry incumbents. While support for the hypotheses was not complete, the results provided a rather consistent description of the factors that influence a firm’s changes in relative competitive position. Interestingly, our operational definition of strategic position in terms of market positions, strategic postures, and resource endowments provide interesting insights about the extent and type of strategic repositioning that takes place in the airline industry. As shown in Fig. 1, our theoretical framework used to develop hypotheses has the following characteristics: first, the three dimensions of strategy and resulting competitive positioning are subject to both external alignment and internal alignment pressures. Second, the external alignment is driven by two exogenous factors, rivalry and fitness, which affect competitive differentiation and mimetic search. In parallel to the legitimation of organizational form, mimetic search is based on the fitness of a given strategy dimension, which underlies the convergence of strategy dimensions among organizations.
432
JAVIER GIMENO ET AL. S ij (rij, fj)
+ –
+
–
Internal Alignment
R ij (rij, fj)
M ij (rij, fj)
Fig. 1. Sij, Rij, and Mij Refer to Strategic, Resource, and Market Similarities between the Focal Firm i and its Competitor j. r and f Refer to Rivalry between Two Firms and Fitness of a Firm’s Given Strategy Dimension. Arrows Refer to the Feedback Loop between Strategy Dimensions.
The internal alignment pressures generated two opposing patterns of interactions between strategy dimensions. As shown in the right-hand side of the figure, the outcome of internal alignment in the product market is consistent with the dynamic effects of niche overlap (Dobrev et al., 2001, 2003). This competitive differentiation may extend the resource-partitioning process in an interesting way. Given that strategic posture tended to be strongly inert, firms could differentiate their competitive positions by modifying their served markets or niche width. The end states of this process include (1) specialists of the same form operating in different markets and (2) specialists and generalists operating in the same markets. These predictions agree and extend the traditional version of the resource-partitioning process (Carroll, 1985). On the other hand, the internal alignment in the factor market precluded the resource-partitioning process. Rather, firms tended to mimic the resource niche positions of firms with similar strategic profiles, leading to crowding in the factor markets. Niche overlap in the technological space illustrates this effect clearly. In his study on the semiconductor industry, Stuart (1998) reports that firms with high niche overlap in the technological space have more learning and alliance opportunities than those with little overlap. Given the economic value of learning and alliance opportunities in the high-tech industry, a highly crowded position in the factor market should confer on a firm competitive advantage and offset the downside of niche overlap. To the extent that the benefits of pooling input resources may
Dynamics of Competitive Repositioning
433
help de-escalate rivalry, the internal alignment in the factor market may be consistent with the observation that the detrimental effects of niche overlap decline with industry age (Dobrev et al., 2001). The two opposing results of niche overlap in our framework have distinct implications for the legitimation process of organizational forms. An organization’s form is a set of social identity codes determined by outsiders (McKendrick, Jaffee, Carroll, & Khessina, 2003; Po´los, Hannan, & Carroll, 2002). Outsiders evaluate existing organizational forms and distribute their evaluations to transacting parties. Securities analysts in Zuckerman’s (1999) study are such an example. When such outsiders develop a favorable consensus on an organizational form, their opinions eventually induce individual firms to adopt that form, which leads their competitive positioning to be similar over time. Yet Zuckerman (1999) also acknowledges that once firms enter the zone of legitimacy, i.e., satisfy outsiders’ concerns, they tend to differentiate from one another in their product offerings. For example, new low-cost carriers such as JetBlue were able to gain legitimacy by imitating the strategic profile and resource endowments of successful low-cost carriers like Southwest, while avoiding market overlap with the established low-cost carriers. This suggests that, in this context, alignment on the resource and strategy space may be more beneficial for achieving legitimacy, while alignment in served markets may not have beneficial effects to counteract the negative effect of competitive intensity. Overall, the findings of our study suggest that a firm’s changes of relative position serve to align the strategy with the external environment (exploit opportunities by approaching better-performing firms that are sufficiently similar, avoid threats by differentiating from similar firms when rivalry is intense); likewise, they serve to align the dimensions of strategy with one another (convergent or divergent change along multiple dimensions). The results provide evidence of both external and internal alignment for market positions and resource endowments. Firms tended to approach those firms that were initially similar in these dimensions when those referent firms performed better, and tended to distance themselves if rivalry was intense. Served market and resource endowments are changed in ways that are internally consistent with strategy postures: firms mimicked market positions of market-similar firms that were strategically dissimilar, and imitated resource endowments of resource-similar firms that were strategically similar. The findings suggest that strategic and resource similarities tend to be more convergent dimensions, while strategic and market similarities are divergent. The findings are consistent with Baum and Haveman’s (1997) work on the Manhattan hotel industry, which highlighted the simultaneous impact of
434
JAVIER GIMENO ET AL.
agglomeration and differentiation tendencies within an industry. We found that convergent forces were more effective in creating configurations of strategic postures and resource endowments, while divergent forces pushed firms to develop novel combinations of market positions and strategic postures. Interestingly, we found only limited evidence of systematic external or internal alignment on strategic postures. It appears that inertial forces are strongest in this dimension of strategy, perhaps because strategic posture requires the effective alignment of multiple complementary activities within the value chain (Porter, 1996; Rivkin & Siggelkow, 2003). These findings in fact support our argument that niche and form do not converge in some markets. Rather, they suggest that firms may not change organizational form or strategic posture in a timely fashion, while they may modify their position in served markets and resources more flexibly. Market positions and resource endowments are generally easier to observe and copy, at least within the constraints imposed by a firm’s strategic profile. This should offer one boundary condition for the structural inertial theory (Hannan & Freeman, 1984). The model showed that most external and internal influences were active through the filter of local search and adaptation. For example, firms did not imitate the market position of all better-performing firms, only those that were initially proximate. Under conditions of intense rivalry, they did not differentiate from all firms, but mainly from those that already had high market overlap. The importance of local search is probably due to a variety of factors, including cognitive framing effects, facilitated vicarious learning under conditions of similarity, and reduced costs and risks of adaptation. The evidence of local search suggests that, in situations of radical change in the environment, competitive repositioning activities are likely to focus on localized, path-dependent optimization of the strategy, rather than a broader search for a more globally optimal configuration of strategy dimensions. The findings of this study are based on distinguishing three dimensions of strategy, i.e., actions taken in relation to the resource market, the output market, and internal alignment of activities. To the extent that the three dimensions of strategy are generally relevant in other contexts, they may provide some novel insights to existing ecological research. Taking the companion paper by Negro and Sorenson (2006, this volume) as an example, a strategy of vertical integration involves positioning decisions in all three dimensions. Integrated firms rely on different resource pools than nonintegrated firms. The configuration of activities of integrated producers also differs, since internal activities substitute for external sourcing. Vertical
Dynamics of Competitive Repositioning
435
integration may also influence the markets served, and therefore the niche width and overlap with other firms. In that respect, Negro and Sorenson provide some support for this view. The decision to integrate affected not only resources and strategic activities (which were not directly observed), but also the market niche width of integrated versus nonintegrated producers. Their findings show that alignment of these dimensions has implications for survival. In particular, differentiation in strategy and resource endowments associated with vertical integration may buffer firms against the competitive impact of niche overlap with nonintegrated producers. This is consistent with our argument that overlap in output markets may facilitate processes of strategic differentiation and resource partitioning. A paradoxical result relates to competition among integrated firms. Our model would predict that integrated firms (firms with similar strategic profiles and resources) would tend to partition the product markets when rivalry is high (or otherwise, face higher competitive pressures). Negro and Sorenson (2006) found that the effect of competitive overlap with integrated firms was actually not higher for integrated producers than for nonintegrated producers. Since convergence in the three dimensions should generally increase rivalry, the results may point out institutional conditions in the industry (perhaps their multimarket contact) that allow integrated producers to manage rivalry despite their high overlap in multiple dimensions. In any case, our model suggests that it would be interesting to map integrated and nonintegrated companies in the resource endowment and strategic profile dimensions, to better understand their competitive positioning. In summary, the paper develops the dynamics of pair-wise competitive repositioning and the factors that influence the directions of changes in positions relative to other industry incumbents. Our multidimensional conceptualization of relative competitive position in terms of the served market, strategy posture, and resource endowments provides a more comprehensive framework than previous work. Our use of pair-wise similarity measures to characterize a firm’s position in strategy space also deserves attention, since related work in strategy has tended to collapse this information into discrete groupings of strategically homogeneous firms. The paper provides a point of convergence between the strategy literature on strategic groups and generic strategies, and the ecology literature on niche-based competition. The results show that strategic change is possible but limited in scope. Despite high inertia, firms were able to modify their served markets and resource endowments in response to external and internal alignment pressures. It is interesting to note that, despite strategy’s more positive view about the viability and effectiveness of strategic change,
436
JAVIER GIMENO ET AL.
the dimension traditionally used in the strategy literature (strategic similarity) was the one most subject to inertial forces and least responsive to external or internal alignment. Future work could build on our conceptual and methodological approaches to further develop and refine our understanding of the microdynamics of business-level repositioning.
NOTES 1. It should be noted that the dimensions of strategy differ from niche dimensions. Researchers usually consider all the relevant dimensions of a given niche before analyzing niche overlap between two organizations. In that regard, niche dimensions are a ‘‘subcategory’’ of both dimensions of strategy, i.e., served markets and resource endowments. 2. Like any theorizing, our approach here is to endogenize the choice of individual managers by using simplified mechanisms, i.e., external and internal alignments. In that regard, it should be noted that strategy and ecology scholars alike assume ‘‘volition’’ or ‘‘purposefulness’’ by individuals when they describe the behavior of individual organizations. Yet they differ in explaining how individual volition is aggregated into the market, i.e., a selection mechanism, although such difference dissolves rapidly nowadays as many strategy scholars come to emphasize the social, nontechnical selective pressures (e.g., Baum & Dutton, 1996). 3. The legitimation process may limit competitive differentiation. This would generate one boundary condition for our theory other than the ‘‘intensive rivalry’’ that comes below: firms outside the zone of legitimacy, such as small, risky ventures, may not be subject to the competitive differentiation. 4. Unlike the legitimation process, the mimetic search underscores the identity not of an organizational form, but that of a specific organization, especially a betterperforming one. 5. Contestable markets were markets with very low costs of entry and exit, in which the threat of new entry ensured competitive conduct even in monopoly markets. 6. The number of firms in the sample varied during the period, from a peak of 37 in 1985 to a low of 23 in 1991. Entries tended to happen from 1979 to 1985 and after 1992, while exits dominated during 1985–1991. Only 13 firms remained for the entire 17-year period. 7. With 54 firms, there was a maximum of 2,862 possible ordered firm-pairs (54x53). The actual number is lower because some firms did not overlap in the same time periods. 8. During the period of observation, the 54 airlines in the sample used 83 different aircraft models according to the classification system developed by the DOT. However, since some of those models were close substitutes, the types of aircraft were consolidated in terms of operational similarity. We clustered the 83 different aircraft in terms of number and type of engines, number of available seats, cruising speed, and flying distance. Using the Ward’s minimum variance clustering criteria, the classification of aircraft was reduced to 26 more homogeneous types, which explained just over 99% of the overall variance in the above dimensions.
Dynamics of Competitive Repositioning
437
9. Tourist markets were defined as those in which the origin or destination city was Aspen, Atlantic City, Las Vegas, Reno, or any destinations in Florida, Hawaii, the Virgin Islands, Guam, American Samoa, the Mariana Islands, or Puerto Rico. 10. Clearly, the regression proposed here is underspecified, since there are unobserved characteristics of the market as well as market structure conditions that explain price levels. We decided not to control for unobserved heterogeneity by using fixed effects because part of the unobserved heterogeneity may be due to differences in competitive conditions, and therefore such variance should be captured in the construct of rivalry. 11. The dependent variables and lagged-dependent variables were standardized using the same mean and standard deviation to ensure that units were comparable across time.
REFERENCES Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage. Alexander, D. L. (1985). An empirical test of the mutual forbearance hypothesis: The case of bank holding companies. Southern Journal of Economics, 52, 122–140. Amit, R., & Shoemaker, P. J. H. (1993). Strategic assets and organizational rent. Strategic Management Journal, 14, 33–46. Bailey, E. E., Graham, D. R., & Kaplan, D. P. (1985). Deregulating the airlines. Cambridge, MA: MIT Press. Barnett, W. P., & Carroll, G. R. (1995). Modeling internal organizational change. Annual Review of Sociology, 21, 217–236. Barnett, W. P., & Hansen, M. (1996). The red queen in organizational evolution. Strategic Management Journal, 17, 139–157. Barney, J. B., & Hoskisson, R. B. (1990). Strategic groups: Untested assertions and research proposals. Managerial and Decision Economics, 11, 187–198. Baum, J. A. C., & Dutton, J. E. (Eds). (1996). The embeddedness of strategy: Advances in strategic management (Vol. 15). Greenwich, CT: JAI Press. Baum, J. A. C., & Haveman, H. A. (1997). Love thy neighbor? Differentiation and agglomeration in the Manhattan hotel industry, 1898–1990. Administrative Science Quarterly, 42, 304–338. Baum, J. A. C., & Korn, H. J. (1996). Competitive dynamics of interfirm rivalry. Academy of Management Journal, 39, 255–291. Baum, J. A. C., Li, S. X., & Usher, J. M. (2000). Making the next move: How experiential and vicarious learning shape the locations of chains’ acquisitions. Administrative Science Quarterly, 45, 766–801. Baum, J. A. C., & Mezias, S. J. (1992). Localized competition and organizational failure in the Manhattan hotel industry, 1898–1990. Administrative Science Quarterly, 37, 580–604. Baum, J. A. C., & Singh, J. V. (1994). Organizational niche overlap and the dynamics of organizational mortality. American Journal of Sociology, 100, 346–380. Baum, J. A. C., & Singh, J. V. (1996). Dynamics of organizational responses to competition. Social forces, 74, 1261–1297.
438
JAVIER GIMENO ET AL.
Baumol, W., Panzar, J. C., & Willig, R. (1982). Contestable markets and the theory of industrial structure. New York, NY: Harcourt Brace Jovanovich. Boone, C., Carroll, G. R., & van Witteloostuijin, A. (2004). Size, differentiation and the performance of Dutch daily newspapers. Industrial and Corporate Change, 13, 117–148. Bothner, M. S. (2003). Competition and social influence: The diffusion of the sixth-generation processor in the global computer industry. American Journal of Sociology, 108, 1175–1210. Boyd, R., & Richerson, P. J. (1985). Culture and the evolutionary process. Chicago: University of Chicago Press. Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in populations of organizations. American Journal of Sociology, 90, 1262–1283. Carroll, G. R., & Swaminathan, A. (1992). The organizational ecology of strategic groups in the American brewing industry from 1975 to 1988. Industrial and Corporate Change, 1, 65–97. Carroll, G. R., & Swaminathan, A. (2000). Why the microbrewery movement? Organizational dynamics of resource partitioning in the American brewing industry after prohibition. American Journal of Sociology, 106, 715–762. Caves, R. E., & Porter, M. E. (1977). From entry barriers to mobility barriers. Quarterly Journal of Economics, 91, 241–261. Chen, M-J. (1996). Competitor analysis and inter-firm rivalry: Toward a theoretical integration. Academy of Management Review, 21, 100–134. Chen, M-J., & MacMillan, I. C. (1992). Nonresponse and delayed response to competitive moves: The roles of competitor dependence and action irreversibility. Academy of Management Journal, 35, 539–570. Chen, M.-J., Su, K. H. & Tsai, W. (2006). Competitive tension: The awareness-motivationcapability perspective. Academy of Management Journal, forthcoming. Child, J. (1972). Organizational structure, environment, and performance: The role of strategic choice. Sociology, 6, 1–22. Cliff, A. D., & Ord, J. K. (1981). Spatial processes: Models and applications. London: Pion Limited. Cool, K., & Schendel, D. (1987). Strategic group formation and performance: The case of the U.S. pharmaceutical industry 1963–1982. Management Science, 33, 1102–1124. Cyert, R. A., & March, J. G. (1963). A behavioral theory of the firm. Englewood Cliffs, NJ: Prentice-Hall. D’Aspremont, C., Gabszewicz, J. J., & Thisse, J.-F. (1979). On Hotelling’s stability in vompetition. Econometrica, 47, 1045–1050. Dierickx, I., & Cool, K. (1989). Asset stock accumulation and sustainability of competitive advantage. Management Science, 35, 1504–1511. Dobrev, S. D., Kim, T., & Hannan, M. T. (2001). Dynamics of niche width and resource partitioning. American Journal of Sociology, 5, 1299–1337. Dobrev, S. D., Kim, T., & Hannan, M. T. (2003). Shifting gears, shifting niches: Organizational inertia and change in the evolution of U.S. automobile industry, 1885–1981. Organization Science, 14, 264–282. Doreian, P. (1981). On the estimation of linear models with spatially distributed data. In: S. Leinhardt (Ed.), Sociological methodology (pp. 156–192). San Francisco: Jossey-Bass. Edwards, C. D. (1955). Conglomerate bigness as a source of power. In: Business Concentration and Price Policy (pp. 331–352). A Conference of the Universities-NBER, Princeton, NJ: Princeton University Press.
Dynamics of Competitive Repositioning
439
Feinberg, R. M. (1984). Mutual forbearance as an extension of oligopoly theory. Journal of Economics and Business, 36, 243–249. Feinberg, R. M. (1985). ‘Sales-at-risk’: A test of the mutual forbearance theory of conglomerate behavior. Journal of Business, 58, 225–241. Festinger, L. (1954). A theory of social comparison processes. Human relations, 7, 117–140. Fiegenbaum, A., & Thomas, H. (1990). Strategic groups and performance: The U.S. insurance industry, 1970–1984. Strategic Management Journal, 11, 197–215. Fiegenbaum, A., & Thomas, H. (1995). Strategic groups as reference groups: Theory, modeling and empirical examination of industry and competitive strategy. Strategic Management Journal, 16, 461–476. Gimeno, J. (1999). Reciprocal threats in multimarket rivalry: Staking out ‘spheres of influence’ in the U.S. airline industry. Strategic Management Journal, 20, 101–128. Gimeno, J., & Woo, C. Y. (1996). Hypercompetition in a multimarket environment: The role of strategic similarity and multimarket contact in competitive de-escalation. Organization Science, 7, 322–340. Graham, D. R., Kaplan, D. P., & Sibley, D. S. (1983). Efficiency and competition in the airline industry. Bell Journal of Economics, 14, 118–138. Greene, W. H. (1997). Econometric analysis. Upper Saddler River, NJ: Prentice-Hall. Greve, H. R. (1995). Jumping ship: The diffusion of strategy abandonment. Administrative Science Quarterly, 40, 444–473. Greve, H. R. (1996). Patterns of competition: The diffusion of a market position in radio broadcasting. Administrative Science Quarterly, 41, 29–60. Hannan, M. T., & Carroll, G. R. (1992). Dynamics of organizational populations: Density, legitimation, and competition. NY: Oxford University Press. Hannan, M. T., & Freeman, J. (1984). Structural inertia and organizational change. American Sociological Review, 49, 149–164. Hannan, M. T., & Freeman, J. (1986). Where do organizational forms come from? Sociological Forum, 1, 50–72. Hannan, M. T., & Freeman, J. (1989). Organizational ecology. Cambridge, MA: Harvard University Press. Hannan, M. T., & Freeman, J. H. (1977). The population ecology of organizations. American Journal of Sociology, 82, 929–964. Hatten, K. J., & Hatten, M. L. (1987). Strategic groups, asymmetrical mobility barriers and contestability. Strategic Management Journal, 8, 329–342. Hatten, K. J., & Schendel, D. E. (1977). Heterogeneity within an industry. Journal of Industrial Economics, 26, 592–610. Haunschild, P. R., & Miner, A. S. (1997). Modes of interorganizational imitation: The effects of outcome salience and uncertainty. Administrative Science Quarterly, 42, 472–500. Haveman, H. A. (1992). Between a rock and a hard place: Organizational change and performance under conditions of fundamental environmental transformation. Administrative Science Quarterly, 37, 48–75. Haveman, H. A., & Rao, H. (1997). Structuring a theory of moral sentiments: Institutional and organizational coevolution in the early thrift industry. American Journal of Sociology, 102, 1606–1651. Hofer, C. W., & Schendel, D. E. (1978). Strategic formulation: Analytical concepts. St. Paul, MN: West Publishing.
440
JAVIER GIMENO ET AL.
Hurdle, G. J., Johnson, R. L., Joskow, A. S., Werden, G. J., & Williams, M. A. (1989). Concentration, potential entry, and performance in the airline industry. Journal of Industrial Economics, 38, 119–139. Jovanovic, B. (2001). Fitness and age: Review of Carroll and Hannan’s demography of corporation and industries. Journal of Economic Literature, 34, 105–119. Kelly, D., & Amburgey, T. L. (1991). Organizational inertia and momentum: A dynamic model of strategic change. Academy of Management Journal, 34, 591–612. Ketchen, D. J., Jr., Thomas, J. B., & Snow, C. C. (1993). Organizational configurations and performance: A comparison of theoretical approaches. Academy of Management Journal, 36, 1278–1313. Korn, H. J., & Baum, J. A. C. (1999). Chance, imitative, and strategic antecedents to multimarket contact. Academy of Management Journal, 42, 171–193. Leifer, E. M. (1985). Markets as mechanisms: Using a role structure. Social Forces, 64, 442–472. Levine, M. E. (1987). Airline competition in deregulated markets: Theory, firm strategy, and public policy. Yale Journal of Regulation, 4, 393–494. Lincoln, J. R. (1984). Analyzing relations in dyads: Problems, models, and an application to interorganizational research. Sociological Methods & Research, 13, 45–76. McGee, J., & Thomas, H. (1986). Strategic groups: Theory, research and taxonomy. Strategic Management Journal, 7, 141–160. McGrath, R. G., Chen, M.-J., & MacMillan, I. C. (1998). Multimarket maneuvering in uncertain spheres of influence: Resource diversion strategies. Academy of Management Review, 23, 724–740. McKendrick, D., Jaffee, J., Carroll, G., & Khessina, O. (2003). In the bud? Analysis of disk array producers as a (possibly) emerge organizational form. Administrative Science Quarterly, 48, 60–94. McPherson, J. M. (1983). An ecology of affiliation. American Sociological Review, 48, 519–532. McPherson, J. M., & Rotolo, T. (1996). Testing a dynamic model of social composition: Diversity and change in voluntary groups. American Sociological Review, 61, 179–202. Mehra, A. (1996). Resource and market-based determinants of performance in the U.S. banking industry. Strategic Management Journal, 17, 307–322. Miller, D., & Chen, M. J. (1994). Sources and consequences of competitive inertia: A study of the U.S. airline industry. Administrative Science Quarterly, 39, 1–23. Negro, G., & Sorenson, O. (2006). The competitive dynamics of vertical integration: Evidence from U.S. motion picture producers, 1912–1970. In: J. A. C. Baum, S. D. Dobrev, & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 363–398). Oxford, UK: JAI. Peteraf, M. A., & Bergen, M. E. (2003). Scanning dynamic competitive landscapes: A marketbased and resource-based framework. Strategic Management Journal, 24, 1027–1041. Podolny, J., Stuart, T. E., & Hannan, M. T. (1996). Networks, knowledge, and niches: Competition in the worldwide semiconductor industry, 1984–1991. American Journal of Sociology, 102, 659–689. Po´los, L., Hannan, M. T., & Carroll, G. R. (2002). Foundations of a theory of social form. Industrial and Corporate Change, 11, 85–115. Porac, J. F., Thomas, H., Wilson, F., Paton, D., & Kanfer, A. (1995). Rivalry and the industry model of Scottish knitwear producers. Administrative Science Quarterly, 40, 203–227. Porter, M. E. (1980). Competitive strategy. New York: The Free Press.
Dynamics of Competitive Repositioning
441
Porter, M. E. (1985). Competitive advantage: Creating and sustaining superior performance. New York: The Free Press. Porter, M. E. (1996). What is strategy? Harvard Business Review, 74, 61–80. Rivkin, J. W. (2000). Imitation of complex strategy. Management Science, 46, 824–844. Rivkin, J. W., & Siggelkow, N. (2003). Balancing search and stability: Interdependencies among elements of organizational design. Management Science, 49, 290–311. Siggelkow, N. (2001). Change in the presence of fit: The rise, the fall, and the renaissance of Liz Claiborne. Academy of Management Journal, 44, 838–857. Sohn, M. (2000). Distance and cosine measures of niche overlap. Social Networks, 23, 141–165. Strassmann, D. L. (1990). Potential competition in the deregulated airlines. Review of Economics and Statistics, 72, 696–702. Stuart, T. E. (1998). Network positions and propensities to collaborate: An investigation of strategic alliance formation in a high-technology industry. Administrative Science Quarterly, 43, 668–698. Stuart, T. E., & Podolny, J. M. (1996). Local search and the evolution of technological capabilities. Strategic Management Journal, 17, 21–38. Teece, D. J., Pisano, G., & Shuen, A. (1997). Dynamic capabilities and strategic management. Strategic Management Journal, 18, 509–533. Wernerfelt, B. (1984). A resource-based view of the firm. Strategic Management Journal, 5, 171–180. White, H. C. (1981). Where do markets come from? American Journal of Sociology, 87, 517–547. Zuckerman, E. W. (1999). The categorical imperative: Securities analysts and the illegitimacy discount. American Journal of Sociology, 104, 1398–1438.
This page intentionally left blank
442
PART VII: COOPERATIVE STRATEGY
443
444
FIGHTING A COMMON FOE: ENMITY, IDENTITY AND COLLECTIVE STRATEGY Jo-Ellen Pozner and Hayagreeva Rao ABSTRACT In this paper, we explore the conditions under which organizations that compete in both market and non-market domains might engage in collective strategy. We study low-power FM radio activists in the U.S., who employed a collective strategy both within and across geographic communities to gain the right to broadcast in low-power broadcast spectra. By comparing and contrasting two stages of the micro-radio movement, we argue that, under certain conditions, for collective strategy to be viable, organizations competing on the dimensions of both ideology and resources must recognize themselves as members of an identity group, based on their common struggle against a stronger, more salient enemy. We highlight the role of collective strategies in the processes of organizational ecology, and discuss the generalizability of our argument.
INTRODUCTION A productive area of interchange between the ecology and strategy literatures is in the domain of interorganizational cooperation. For the most part, Ecology and Strategy Advances in Strategic Management, Volume 23, 445–479 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23014-6
445
446
JO-ELLEN POZNER AND HAYAGREEVA RAO
strategy scholars have focused attention on cooperation through dyadic and multilateral arrangements in the market domain, whereas organization ecologists have documented mutualism between populations. Neither group of scholars, however, has devoted sufficient attention to collective strategies in non-market domains. Although scholars on both sides highlight the role of resource competition as a trigger of cooperation, for the most part, they have glossed over the role of identity as a mobilizer of cooperation at the collective level. Collective strategy has been defined as ‘‘the joint formulation of policy and implementation of action by the members of interorganizational collectivities’’ (Astley, 1984, p. 526). Although this definition suggests that collaborators must identify with other organizations in the collective to participate in a joint strategy, the question of identity has not been fully explored by researchers of interorganizational coordination, which focuses primarily on linkages among interdependent organizations and the environment in which they operate. This literature pays particular attention to organizational struggles to control environmental uncertainty and to secure stocks of scarce resources, but does not explore the link between identity and resource competition. In this essay, we propose that group identification – self-cognizance as an organizational community – within organizational communities that compete with salient rivals on the dimensions of both resources and ideology fosters collective strategies of cooperation, particularly in non-market domains. Our analysis focuses on a specific type of market structure: one in which production resources are homogeneous and sharply constrained, although consumption resources are heterogeneous, and where there are economies of scale for generalists, who dominate the center of the resource space. In such environments, according to resource partitioning theory (Carroll & Hannan, 2000), specialist organizations are likely to appear to serve peripheral resource spaces not served by generalists. Although collective strategy – explicit coordination of interorganizational action, or direct interaction among organizational collectives (Astley & Fombrun, 1983; Oliver, 1988) – is likely to promote the legitimacy and survival of emerging specialists, such cooperation is difficult to enact; this is particularly true of organizational communities marked by commensalism, or competition, on both ideological and resource dimensions, as well as the struggle for legitimacy of the specialist organizational form. If, however, the nascent specialist community recognizes that its most salient foes in terms of identity, resources, and legitimacy are not other specialists, but rather generalist producers, it will develop an identity in opposition to the larger enemy. This
Fighting a Common Foe: Enmity, Identity and Collective Strategy
447
group identification will, in turn, facilitate the mobilization of collective action, and consequently organizational legitimacy, which should, in turn, increase organizational founding and survival rates. To illustrate our argument, we present the case of low-power FM (LPFM) radio activists (also called micro-radio) in the United States, who successfully deployed a collective strategy to gain the right to broadcast on the commercial radio spectrum. The two distinct phases in the history of the micro-radio movement illustrate the factors critical to the success of cooperative strategy. After an initially rocky effort, in which federal regulators were targeted as the enemy of low-power radio, disparate groups of micro-radio activists, who proposed competing ideologies and who competed with each other indirectly for resource space on the broadcast spectrum, joined together. Once these erstwhile foes developed a collective identity, by framing commercial radio broadcasters as their common enemy, against which all competed both ideologically and for space on the broadcast spectrum, they were able to set aside their differences and work together. That is, when they recognized that they were all more distant from commercial broadcasters in terms of ideology and resource competition than from each other, the common fight became their most salient mission, resulting in a strong group identity. This identity, in turn, mobilized a collective strategy that enabled micro-radio activists to hasten the process of resource partitioning in the radio broadcast industry. Moreover, the development of a strong group identity enabled micro-radio activists to deploy a cooperative ideology that was palatable to complementary organizations, collaboration with which was crucial to legitimating the new organizational form. Thus, it was the struggle over ideology and resource control against a salient rival that facilitated the development of a strong group identification, which in turn created an opportunity structure wherein successful strategic cooperation was possible. This can be contrasted with the model presented in this volume by Shipilov, Rowley, and Aharonson (2006), which also deals with the choice of partner for cooperation, although in the context of formal, cooperative ties among potential competitors, and among organizations not specifically engaged in the fight for legitimacy. We argue that, when organizations recognize themselves as members of a group, their attention is directed away from within-group and toward between-group competition; this, in turn, facilitates within-group cooperation. Thus, we propose, when spurred by the threat of competition for critical resources from a rival, group identification can facilitate successful collective strategy. We draw on a case study to produce basic propositions regarding the emergence and efficacy of collective strategies, highlighting their role in the fundamental processes of organizational ecology. Finally, we discuss the
448
JO-ELLEN POZNER AND HAYAGREEVA RAO
Necessary conditions: Resource partitioning processes
• Competition for market and non-market resources • Constrained /regulated market resources
Emergence of new, specialist organizational form
Identification of generalists as salient enemy
Fig. 1.
Competition for resources and ideology among specialists
Struggle for legitimacy
Within-group identification among specialists
Collective strategy among specialists
Legitimation and proliferation of specialist form
Proposed Model of Enmity, Identity and Collective Strategy.
conditions that are necessary and sufficient to generate this mechanism, and discuss the generalizability of this argument to industries marked by the emergence of a new organizational form, where the new form competes for both market and non-market resources, where market resources are constrained or regulated, and where a salient enemy can be identified. A graphic representation of this model is presented in Fig. 1. We begin our discussion with a history of the micro-radio movement in the United States, focusing on the forces that promoted the development of collective identity, then collective strategy, and its consequences for ecological processes. Drawing on data collected by Greve, Pozner, and Rao (2006), we contrast the two distinct phases in the history of the micro-radio movement.
COLLECTIVE STRATEGY AMONG MICRO-RADIO ACTIVISTS Early History The struggle for control over the radio airwaves dates back almost to Marconi’s first trans-Atlantic broadcast in 1901. The sinking of the Titanic in
Fighting a Common Foe: Enmity, Identity and Collective Strategy
449
1912 spurred the United States government to regulate its airwaves by raising concerns that airwave congestion might block critical transmissions in times of crisis. Consequently, the Radio Act of 1912 required broadcasters to register with the U.S. Department of Commerce, which assigned licenses based on the worthiness of applications. This and subsequent regulations did not prevent unlicensed amateur radio enthusiasts from continuing to broadcast (Anderson, 2004b). Because early thinking surrounding radio was more closely related to a metaphor of public transportation, for which tolls could be collected for discrete use, than to that of a transferable, private good, it was in the 1920s, when radio producers began to broadcast regularly in an effort to boost radio receiver sales, that broadcasters became concerned with the activities of amateurs (Leblebici, Salancik, Copay, & King, 1991). Beginning in the early 1920s, the Department of Commerce attempted to remove amateurs from the airwaves, thus professionalizing radio broadcasting while leaving educational institutions – primarily colleges and universities – as the only legitimate, non-commercial broadcasters (Streeter, 1996). The idea of the radio spectrum as a scarce resource also emerged in the 1920s. Then Secretary of Commerce Herbert Hoover froze license applications in 1922 in response to complaints about interference from competing stations. Through the freeze, Hoover developed the doctrine of spectrum scarcity, which holds that access to the airwaves is limited by the carrying capacity of the band allocated to commercial broadcasting (Anderson, 2004b). The scarcity created by this doctrine made licenses more valuable, and led to their trade on the open market, setting the stage for competition for bandwidth as the focus of both domain defense and subsequent strategic cooperation. Changes in broadcasting technology and economics changed the radio landscape in the 1940s and 1950s, whereafter radio became a much more local, specialized medium (Leblebici et al., 1991; Sterling & Kittross, 1978). The move toward localism reinforced the trend begun by the Communications Act of 1934, which limited ownership to two stations per market, with a cap of 20 stations nationwide. In 1948, the Federal Communications Commission (FCC) created Class D radio stations, a low-power radio service that enabled educational institutions to train students in broadcasting on a dedicated band of the FM dial; this service further contributed to public participation in licensed radio and sanctioned local radio broadcasting (Anderson, 2004a). Nearly two decades later, in 1967, the Public Broadcasting Act established a national public radio service and created the Corporation for Public Broadcasting (CPB), a private, non-profit corporation, to ensure universal access to high-quality,
450
JO-ELLEN POZNER AND HAYAGREEVA RAO
non-commercial programming (CPB Board of Directors, 1999). The CPB redefined the mission of public and educational radio from that of training broadcasters to that of educating listeners, and beginning in 1972, lobbied the FCC to eliminate Class D stations and to standardize educational radio along the model of National Public Radio (NPR); NPR, not coincidentally, was directly funded by the CPB until 1987 (Dvorkin, 2000). The CPB argued that the relatively low quality and irregular schedules of Class D stations was an inefficient use of radio spectrum space (Anderson, 2004a). Choosing to serve a wider audience rather than to promote diverse voices, the FCC eliminated Class D licenses in 1978, and required educational institutions either to upgrade or to move to the commercial band of the FM spectrum (Ruggiero v. FCC, 2002). The educational band was then given to so-called translator or booster stations, which are dedicated to relaying the signal – and thus extending the broadcast radius – of their commercial parents (Federal Communications Commission: Experimental Radio, Auxiliary, Special Broadcast and Other Program Distributional Services, 2002).
Pirates on the Airwaves: 1979–1995 Unlicensed broadcasters began invading the FM dial following the elimination of the Class D radio service in 1978. Pirate radio – unlicensed, illegitimate broadcasts, often run by individuals using mobile transmitters – was the only niche left in which enthusiasts could exercise self-expression. Like earlier hobbyists, who felt that because they did not cause interference with licensed broadcasters, they were not the intended targets of federal regulation, pirates ranging from the teenagers playing music in their bedrooms to civil rights groups broadcasting radical political messages in the 1960s and 1970s flaunted federal licensing rules (Anderson, 2004b). Because radio piracy emerged as a means of illegitimate, often individual, self-expression, it was unlikely to encourage the strong group identification capable of sustaining coordinated strategic action. Not only were pirates’ individual identities incompatible with cooperative strategy, so were their ideologies. The pirates who became active after 1978 lamented the lack of local responsiveness and diversity of voice, ownership and employment in radio broadcasting, emphasizing the need for diversity and non-conformity. Pioneering micro-radio broadcasters such as Walter Dunn in Fresno, California, and Mbanna Kantako in Springfield, Illinois, broadcast their own music, publicity for African–American businesses, and ‘‘militant talk’’ (Sakolsky, 1992), such as raising awareness about police
Fighting a Common Foe: Enmity, Identity and Collective Strategy
451
brutality in minority communities (Anderson, 2004a). Organizations such as Americans for Radio Diversity (ARD) and Radio Mutiny aimed at promoting more interesting and local programming choices. These organizations lamented that ‘‘[m]ainstream media offers no place for independent thought or dissenting opinions. The cultural life of our society is becoming terrifyingly blurred with our commercial life’’ (Radio Mutiny, 2004). Radio Mutiny, in particular, objected to the trend among NPR and other legitimate radio outlets to shy away from dissenting opinions and controversial points of view; responsibility for this conservative turn was ascribed to the U.S. Congress, a significant financial supporter of public radio, and the FCC (Rosenblum, 1997). Thus there was no unifying, strong identity among radio pirates, each of which advocated and represented its own idiosyncratic identity and ideology. More specifically, not only was there no collective identity, but there was no motivation to cooperate based on ideology. The ideology of the micro-radio movement in its pirate era was exemplified by Stephen Dunifer, founder of the pirate station Free Radio Berkeley, a 50-watt station that broadcast round the clock, originally from Dunifer’s own apartment. Dunifer’s piracy was motivated not only by his perception of bias in commercial media, but also by sympathy with other unlicensed broadcasters who had been shut down by the FCC, which he believed to be illegitimate. Dunifer argued, instead, that radio was a medium of free speech and a forum for diverse voices (Dunifer, 2004). Through Free Radio Berkeley, Dunifer hoped to get the public involved in piracy, creating an ‘‘ungovernable situation’’ for the FCC and forcing it to legalize lowpower, non-commercial broadcasting (Dunifer, 2004). The key to this strategy was open civil disobedience aimed at drawing the FCC into a public justification of its licensing policies (Coopman, 1997a). Although public disobedience on a grand scale failed to materialize, Dunifer was able to draw attention to the pirate radio movement through a highly publicized court case. After the FCC attempted to close Free Radio Berkeley months after its founding, a pro-micro-radio legal team helped Dunifer argue that FCC licensing policy and the ban on low-power radio both violated the First Amendment, and discriminated against 95% of the U.S. population, as the more powerful the station, the more expensive it is to operate (Coopman, 1997a). Critical to this argument was the fact that, although stand-alone stations under 100 watts were illegal, the FCC allowed low-power transmitters to rebroadcast the signals of their parent stations, though forbidden from broadcasting local programming (Federal Communications Commission: Experimental Radio, Auxiliary, Special Broadcast and Other Program Distributional Services, 2002). Despite a decision in
452
JO-ELLEN POZNER AND HAYAGREEVA RAO
favor of the FCC in 1998, the National Association of Broadcasters (NAB), the industry lobbying group, felt sufficiently threatened that it filed a comment justifying the elimination of Class D licenses through spectrum efficiency (Edmonson, 2000). News of the Dunifer case was disseminated on the internet, helping to mobilize a wave of pirate activity already underway in the late 1990s (Anderson, 2004a, 2004b). Advances in technology and communications media also encouraged the development of a micro-radio movement by improving the quality and reducing the price of radio transmitters, making pirate broadcasting relatively easy and inexpensive. The internet also enabled a degree of coordination among pirates, who shared information about FCC enforcement and replaced each others’ equipment when confiscated in FCC raids (Coopman, 1997b). This diverse, loosely organized community of pirates also began to coordinate action, albeit loosely, by launching demonstrations with hand-carried radio transmitters in front the headquarters of both the FCC and the NAB in Washington, DC, in October 1998 (Flugennock, 1998). Those involved describe this phase of the movement as extremely informal, without a unifying identity save that of individuals ‘‘breaking the same law’’ (triDish, 2005). Thus the primary focus of the pirates’ action was still individual in nature, rather than collective, although the roots of coordinated action can be traced to this time period. Broadly speaking, the pirates’ ideology focused on the elimination of Class D service, and FCC licensing policy more generally, which limited freedom of speech while privileging certain groups over others. This loose movement, therefore, entailed an identity as critique strategy (Bernstein, 1997), whereby the pirates strove to distance themselves from mainstream society, creating an identity in contrast to the elitism and discrimination they perceived within commercial broadcasting. It did not, however, produce a positive identity, partially due to the strong ideological divides among pirates, some of whom favored the creation of a commercial lowpower service, a non-commercial service, or a completely local, grass-roots service, while still others were completely absorbed with their individual ideologies and uninterested in media reform (Schellhardt, 2005; triDish, 2005). What common rhetoric there was among pirates centered mainly on contrasting their own liberal views with those heard on the radio, while also framing the FCC as the conservative force responsible for the lack of liberal representation. The FCC thus became the primary focus of the pirates enmity and their goal: to subvert the FCC’s legitimacy and authority by claiming voice through anarchy and civil disobedience. Thus not only did the pirates embody diverse identities, they also failed to find common
Fighting a Common Foe: Enmity, Identity and Collective Strategy
453
ground in their struggle for resources, or bandwidth. By framing the U.S. government as the enemy against which to act, the pirates framed their contest not as competition for resources or even for legitimacy of a new organizational form, but rather as a debate over the legitimacy of regulatory authority, a very different dispute. These failures account, in part, for the failure of the pirate movement to gain momentum and achieve its goals. The biggest success in the pirates’ unorganized campaign to create chaos on the airwaves, which consisted primarily of individual action, was the development of a countermovement. The FCC increased its enforcement efforts against unlicensed broadcasters in response to increased piracy, and the NAB mobilized an effort to suppress the movement. Although initially sporadic, enforcement activity rose as the NAB increased pressure on the FCC to shut down pirate operations as they proliferated (Spangler, 1997). At the same time, the NAB, CPB, and NPR – all of which had strong ties to the FCC – mobilized in opposition to the pirates. As incumbent broadcasters have done since the 1920s, they invoked resource-based arguments against low-power broadcasting, positing the radio spectrum as a scarce resource to which not all comers may have access; adding voices to the airwaves would crowd the radio dial, cause interference to existing signals and harm existing stations’ financial well-being (Tymon, 1998). Fundamentally, the piracy phase of the micro-radio movement failed because its ideology was inconsistent with collective strategy. Pirate broadcasters saw themselves as different, as crusaders, and as exceptionally local activists – that is, activists for their own individual causes alone. Their identity was not one of a strategic group or an organizational community, but rather of outlaws working independently, often with sharply different and competing ideologies. Moreover, the enemy against which they directed their activity was, essentially, the FCC, a bureaucratic organization that presented only an enemy, not a true competitor for resources. Thus, the pirates never coalesced with a strategic group identity. Moreover, they were unable to generate an ideology towards which other constituencies, those with the resources to promote their goals, could sympathize. The pirates’ master strategy – civil disobedience and resistance – did not resonate with the general public, nor did the extremely liberal and often radical viewpoints voiced by most pirates. Pirates’ self-definition in sharp contrast to dominant popular culture alienated and excluded many potential supporters, and precluded the development of a community identity capable of mobilizing cooperative action. Therefore, despite the consistency of its message and the dedication of its members, the pirate movement was ultimately unsuccessful in achieving its goal of legalizing low-power radio.
454
JO-ELLEN POZNER AND HAYAGREEVA RAO
Micro-Radio and Cooperative Strategy: 1996–2004 Ironically, it was the countermovement spurred by the pirate movement that helped micro-radio activists to develop the strong group identity necessary for successful cooperative strategy. Unswayed by pirates’ attempts to force reform, and under heavy lobbying by the NAB and others, the FCC and Congress encouraged consolidation and commercialization of radio ownership through the mid-1990s. The Telecommunications Act of 1996 facilitated consolidation by increasing the single-market ownership cap from two to eight stations and eliminating the national ownership cap. Subsequent to this change, the first wholesale telecommunications regulation overhaul since 1934, came a wave of consolidation within all broadcast media. Between 1996 and 1998, approximately 50% of U.S. radio stations changed hands, with the most significant losses coming from independent stations (Smith & Rosenfeld, 1999) and minority ownership of radio stations dropped 9% within two years of deregulation (DeBarros, 1998). By the end of 2003, Clear Channel Communications owned more than 1,225 stations in 230 U.S. cities, or 11% of all stations, putting it into the homes of 70% of the population (Sharlet, 2003). This reform allowed national station owners to reduce costs by consolidating advertising sales replicating set formats nationwide, replacing local personalities with syndicated and computerautomated programming, and eliminating local news departments (Fisher, 1998). By the late 1990s, one could listen to the same broadcast on a Clear Channel station in virtually any market in the U.S., but could hardly find any truly local programming. Thus although the number of stations did not decline, the number of listening options was limited through consolidation, fueling more activity by radio reform activists. The Telecommunications Act of 1996 was a turning point in the microradio movement. As the spate of media mergers and acquisitions led to increasingly centralized ownership, radio listeners and FCC regulators lamented the homogenization of programming, lack of diversity, and vanishing local coverage on the FM airwaves. As the media giants became ever larger, and as they lobbied against micro-radio ever harder, they presented a salient, inimical out-group competing for a critical resource – bandwidth – with the micro-radio community. Bandwidth was a resource that engendered competition among micro-radio activists, particularly those operating in the same communities, as was later demonstrated by the number of mutually exclusive, competing applications for available bandwidth by LPFM applicants. Nevertheless, this competition was less critical than that against incumbents already in possession of bandwidth. In fact, the incumbents
Fighting a Common Foe: Enmity, Identity and Collective Strategy
455
themselves may have made resource competition even more salient by emphasizing the need to preserve signal integrity by keeping others’ off of the airwaves and insisting on prosecution of pirates. As incumbents’ lobbying efforts became more vocal, making the competition for resources more intense, the ideologies of what had been a fragmented group of activists and organizations converged towards fighting this strong, unified enemy. It was against this backdrop that a new group identity, and ultimately a more successful micro-radio movement, emerged. The formation of a group identity was facilitated by the professionalization of pirate radio that began in the early 1990s, due primarily to the incorporation of new interest groups into the micro-radio movement. An important component of this change was the participation of radio programmers and engineers who had been fired from corporate radio due to consolidation. These displaced radio professionals, having experienced the effects of homogenization of commercial radio directly, became focused on serving the interests of local communities through broadcasting, rather than advocating specific and idiosyncratic ideologies (Hall, 2005). The involvement of radio professionals also attracted the participation of local opinion leaders in communities across the country, who were less interested in media reform than they were in community-level issues, with a strong anticommercial philosophy (triDish, 2005). At the same time, media reform groups such as the Amherst Alliance began to take up the cause of lowpower radio, with the goal of bringing new voices, although not exclusively local voices, to the airwaves. The media reform activists advocated for change from within the establishment, and had less of an ideological stance than other micro-radio activists, though they were for the most part in favor of a commercial low-power radio service (Schellhardt, 2005). Eventually, two more groups joined forces with the micro-radio movement: Evangelical Christian groups, such as the Christian Community Broadcasters, who sought legitimate outlets for their ideological messages; and pro-commercial low-power radio activists, who are interested in more professional broadcasting, but at a lower cost of entry onto the airwaves (Schellhardt, 2005). It is clear that these broad categories of micro-radio activists represented highly divergent viewpoints, some of which are directly in competition with one another. One activist describes the differences among the various constituencies as strong barriers to collective action, with greater we-feeling within each group (Schellhardt, 2005). Yet, the same activist praises these groups for their ability to put aside their differences and work together on the fundamental issue of establishing a low-power radio service. Part of the credit for establishing a super-ordinate identity as micro-radio activists among these
456
JO-ELLEN POZNER AND HAYAGREEVA RAO
diverse groups must go to information technology, through which collective action is more easily coordinated. Perhaps more important to creating a strong group identity, however, are the protests and conferences organized by influential groups such as the Prometheus Project and the Amherst Alliance, which gave participants a feeling of commitment to a radio-reform movement (Hall, 2005; Reese, 2005; Schellhardt, 2005; triDish, 2005). In contrast to the pirates, who operated outside the law, the new activists identified themselves as a community, or strategic group, by forming public action groups and exploiting existing channels of protest (Reece, 1998), and by consciously overcoming ideological differences to engage in collective action. Critical to the development of this strong community identity and a collective strategy was the recognition of the depth of the ideological schism between commercial broadcasters and the various micro-radio groups relative to that among the micro-radio activists. After all, the NAB was actively engaged in efforts to delegitimate micro-radio, whereas the various groups primarily squabbled over the details of the implementation of a low-power service. Most importantly, perhaps, the geographical dispersion of microradio activists meant that they were infrequently in direct competition with one another for bandwidth, whereas all were constantly engaged in direct competition for spectrum space with commercial radio stations at the local level. As one advocate argued: ‘‘The fact that the NAB and its members prevent non-commercial spaces from existing means that the airwaves are preserved for those whose sole motive is money’’ (Clarke, 2004). By focusing on the commonality of the struggle against commercial radio, rather than the relatively minor differences of opinion and indirect resource competition among reformist groups, micro-radio activists found the will to consciously develop and deploy a collective strategy. The framing of the micro-radio movement’s identity as anti-commercial radio, with its monopolistic grip on the airwaves, rather than as an expression of multiple identities and agendas, mobilized activists to cooperate both within their own groups and with like-minded communities. Activists carefully painted the increasingly homogenized and centralized commercial radio industry as the enemy, creating a salient out-group against which many different identity groups could mobilize, while coalescing as a unified strategic group with a strong identity. The movement then easily resonated with a nationwide movement celebrating authenticity, diversity, and community, giving it access to the resources of grass-roots organizations engaged in similar activity, ranging from ‘‘the Green Party, the United States Catholic Conference, the Library Association of America, the ACLU, the Council of Calvin Christian Reformed Church, Native American tribes and the United Church
Fighting a Common Foe: Enmity, Identity and Collective Strategy
457
of Christ; celebrities like the Indigo Girls, Bonnie Raitt, and Kurt Vonnegut; and the cities of Detroit, Seattle, Ann Arbor, Michigan, and Santa Monica, Berkeley and Richmond, California, among others’’ (Boehlert, 2003). Thus, whereas the pirates’ ideology and anti-regulatory stance had alienated community groups, in targeting commercial radio, the moderate frame induced by the later micro-radio activists was sufficient to attract the participation of groups that might have previously been considered ideological rivals. In addition, by painting the radio chains as the enemy, the micro-radio movement was able to attract powerful allies and take advantage of a growing political opportunity structure. First, by framing the commercial radio interests as their enemy, the community micro-radio activists were able to work legitimately with the FCC to overcome the problems caused by consolidation. The movement also resonated with William Kennard, who became FCC Chairman in 1997. Concerned with increasing media consolidation on following the Telecommunications Act of 1996, which he claimed was changing the radio industry ‘‘from one of independently owned operators to something akin to a chain store’’ (DeBarros, 1998), Kennard viewed micro-radio as a legitimate balance to the corporate presence on and homogenization of the airwaves (Boehlert, 2003). It has also been suggested that Kennard, the FCC’s first African-American chairman, as well as a democrat working with a republican-dominated Congress, both valued diversity for its own sake and saw LPFM as a means to give minorities and underserved constituencies access to broadcasting (Mayer, 1998). Thus changes in leadership and philosophy within the regulatory agency itself created a political opportunity structure, and simultaneously increased the salience of commercial radio interests as an enemy, contributing to the success of the micro-radio movement on two levels. The mature micro-radio movement clearly differed from pirate radio in several respects. Pirate radio espoused a radical, individualistic ideology that inhibited the formation of a strong identity – it celebrated its differences with the FCC – its proclaimed enemy, from which it stole bandwidth like the Robin Hood of the airwaves – and posed pirate radio as a radical alternative to commercial radio dominated by a few corporations. By contrast, later micro-radio activists presented a moderate political ideology, which facilitated the formation of a strong identity. Viewing the FCC as an ally, in contrast, allowed micro-radio activists to use existing channels of political protest. Focusing on commercial radio – a salient out-group with which to engage in legitimate ideological and resource competition – enabled the micro-radio activists to create a strong strategic group identity that both unified them and drew in other, supporting groups. Unlike the pirates,
458
JO-ELLEN POZNER AND HAYAGREEVA RAO
whose radical agenda created a narrow identity that excluded many potential supporters to secure a political outcome, this community’s identity was based on non-controversial, inclusive themes such as using radio to promote community development and identification, local dialogue, education, experimentation and diversity, and a new class of radio entrepreneurs. Moreover, whereas the pirates had no central organization through which to operate in the political realm, and in fact preferred a decentralized movement in the political and market realms, the later micro-radio activists’ unifying identity enabled them to deploy a collective strategy both by building organizations for the purpose of coordinating action in the political ream, and by teaming with grass-roots entrepreneurial efforts in local communities in the market realm. Just as radical flank effects arguments would predict (Haines, 1997), the activities of the pirates actually enhanced the success of community radio. First, although Dunifer’s campaign to create chaos on the airwaves failed to produce numbers sufficient to force the FCC’s hand, the cost of FCC enforcement activity, and growing pro-diversity sentiment within and outside the agency made it more amenable to moderate rhetoric in radio licensing reform, opening the opportunity structure for the later-stage activists. Similarly, the vigilance with which the NAB, CPB, and NPR fought against micro-radio, and their lobbying of the ever less-receptive FCC, created a counter-movement that threatened the micro-radio movement to the point that a strong group identity emerged. Moreover, in arguing his case against the FCC, Dunifer developed compelling legal arguments in defense of micro-radio, which were appropriated by the later micro-radio activists to develop legitimate and compelling proposals for regulatory reform (Opel, 2004). Finally, the later-stage activists gained exposure and resonance through FCC enforcement efforts against pirates; external identification of micro-radio as a movement enacting a concerted, collective strategy, through academia and local print media, solidified the movement’s identity and facilitated cooperation.
The Rebirth of Micro-Radio Through a collective strategy, micro-radio activists achieved their goals of regulatory change and the establishment of a new, low-power radio service. In July 1997, micro-radio activists filed a petition with the FCC proposing that the FCC dedicate one channel on both the AM and FM bands nationwide for local micro-stations, with the goal of fostering community
Fighting a Common Foe: Enmity, Identity and Collective Strategy
459
development and identification, developing local dialogue, and providing an outlet for experimentation and diversity, particularly in areas underserved by commercial, high-powered media (Leggett, Leggett, & Schellhardt, 1997). Several months later, an experienced broadcast engineer who had unsuccessfully applied several times for a full-power broadcast license, filed a similar petition encouraging the licensing of low-power stations to create a new class of radio entrepreneurs, to minimize the problem of unlicensed radio broadcasting without causing harm to licensed stations, and to provide more diverse and locally responsive programming (Skinner, 1998). The latter proposal also opened the resource space by demonstrating the feasibility of allowing more low-power broadcasters on the air without causing interference to existing stations. The FCC made both proposals available for public comment in 1998, drawing record numbers of comments both in favor of and in opposition to micro-radio. Despite pressure from the NAB, the FCC issued a Notice of Proposed Rulemaking in the Creation of a Low Power Radio Service on January 28, 1999. This document, and the report and order issued in January 2000, adopted elements of both petitions while making concessions to the opponents of LPFM (Federal Communications Commission, 2000, 1999). The FCC authorized a LPFM radio service in January 2000. LPFM stations are authorized for non-commercial, educational, and community-based broadcasting only. Their operations are limited to an effective radiated power (ERP) of 100 W (0.1 kW) or less and maximum facilities of 100 W ERP at 30 m (100 feet) antenna height above average terrain (HAAT), giving them a broadcast radius of approximately 3.5 m (Federal Communications Commission, 2000). The FCC decision induced the NAB and NPR to lobby Congress, which passed the Radio Broadcasting Preservation Act (RBPA) in December 2000 as a rider on a spending bill – the first time Congress had ever become involved with engineering standards (Prometheus Project, 2000). As we have argued, the NAB and NPR’s efforts to derail LPFM actually strengthened the resolve of FCC chief Kennard to support micro-radio (Kennard, 2000a, 2000b). Although the fight over the LPFM service continued through 2005, opinion regarding media consolidation within Congress has latterly shifted toward the micro-radio activists (Mediageek, 2004), and the service has been adopted and expanded. Figs. 1 and 2 illustrate the degree to which the LPFM service has expanded since its inception, particularly as the growth in the number of full-power commercial stations has slowed (Greve et al., 2005). Greve et al. (2006) have shown that the rise of LPFM stations has even contributed to the loss of market share of incumbent, commercial radio
460
JO-ELLEN POZNER AND HAYAGREEVA RAO
1200
1000
800
600
400
200
Fig. 2.
4/2005
1/2005
10/2004
7/2004
4/2004
1/2004
10/2003
7/2003
4/2003
1/2003
10/2002
7/2002
4/2002
1/2002
10/2001
7/2001
4/2001
0
Number of Licensed LPFM Stations, April 2001–April 2005.
stations (see Figs. 3 and 4) (Greve et al., 2006). Thus, it is clear that the collective action undertaken by micro-radio activists achieved its aims both in establishing legitimacy for the new organizational form and in carving out resources to the extent of eroding the generalists’ resource base.
COLLECTIVE STRATEGY Development of Collective Strategy The example of U.S. micro-radio activists demonstrates our proposed model of enmity, identity, and collective strategy, as well as the role of collective strategy in the processes of legitimation and resource partitioning. When emerging specialist organizations with competing ideologies seek legitimacy and access to scarce production resources, collective identity will emerge through the recognition of the commonality of specialists’ enmity with generalists. This identification in opposition to generalists will help mobilize collective strategy, thus spurring the legitimation and proliferation of
461
1200
3200
1000
3000
800
2800
600
2600
400
2400
200
2200
LPFM stations
Commercial radio stations
Licensed LPFM stations
Fighting a Common Foe: Enmity, Identity and Collective Strategy
Commercial stations
0
2000 2001
Fig. 3.
2002
2003
2004
Number of Licensed LPFM Stations and Commercial Radio Stations, 2001–2005.
0.035
0.09 0.08
0.0345 0.034
0.06 0.05
0.0335 0.04 0.033
Diaries
Concentration
0.07
0.03 0.02
0.0325
0.01 0.032 0
1
2
3
4
5
6
7
8
9
0 10
Herfindahl Diaries
Fig. 4.
Effect of LPFM Stations on LPFM Audience and Chain Concentration (Greve et al., 2006).
462
JO-ELLEN POZNER AND HAYAGREEVA RAO
specialists proposed by resource partitioning theory (Carroll, 1985; Carroll & Hannan, 2000). This model explains why the pirates, in framing their enemy as the FCC, were unsuccessful in their struggle for legitimacy and resource space; their struggle to delegitimate the FCC, rather than to legitimate their own identity, and their failure to recognize the primacy of their direct struggle for resources with chain-owned commercial radio prevented them from developing a group identity and working in concert, resulting instead in internal competition for ideology and resources (see Fig. 1). Moreover, the pirates’ idiosyncratic, anti-establishment identities failed to resonate with outside groups, which might have legitimated them as an organizational form. In stark contrast, the later micro-radio activists’ cognizance of their common foe helped forge a super-ordinate identity, which in turn enabled them to overcome intra-group competition over ideology and resources, and work together to win legitimacy and resource space. The critical interaction of group identification, identity, and resource competition has not been explored in either the organizational strategy or the organizational ecology literatures. The collective strategy literature developed partially in response to population ecologists’ focus on the effects of the resource environment on individual organizations. Astley and Fombrun (1983) encouraged ecologists to consider strategic choice at the collective level as a viable means of exerting control over the environment by absorbing variation and uncertainty, making groups of organizations less susceptible to variation and selection pressures than are individual firms. Cooperation among special interest groups is thus a means to construct a negotiated environment (Cyert & March, 1963), whereby collectives of interdependent organizations occupying the same resource space play an agentic role, mobilizing resources and formulating concerted strategic action to take control over their collective fate. The classic explanation for collective action thus centers around resource scarcity: organizations work together to control scarce resources efficiently and economically, to minimize susceptibility to outside forces – such as powerful competitors and exogenous market change – and thus to promote organizational survival (Aldrich, 1979; Astley & Fombrun, 1983; Bresser & Harl, 1986; Oliver, 1990; Pennings, 1981; Pfeffer & Nowak, 1976; Pfeffer & Salancik, 1978; Thompson, 1967; Van de Ven, 1976; Van de Ven & Walker, 1984; Whetten & Leung, 1979). A number of ecological studies have documented mutualism between populations of organizations and have emphasized economic and technological interdependencies (Barnett, 1990; Carroll & Swaminathan, 1992; Hunt & Aldrich, 1998; Shipilov et al., 2006), to the exclusion of political and ideological ones (Carroll, Delacroix, & Goodstein, 1990),
Fighting a Common Foe: Enmity, Identity and Collective Strategy
463
although more recent ecological work does address identity and ideology (Barnett & Woywode, 2004; Carroll & Swaminathan, 2000; Dobrev, 2000, 2001; Minkoff, 2002; Ruef, 2000). Although many authors suggest that participation in collective strategy creates a sense of solidarity among organizational actors, the question of group identity has not been fully investigated. Rather, it is interdependence based on competition for relatively scarce resources that is assumed to spur organizations to coordinate their actions so as to minimize environmental uncertainty and variation (Aldrich, 1979; Astley & Fombrun, 1983; Bresser & Harl, 1986; Oliver, 1990; Pennings, 1981; Thompson, 1967). Moreover, although several scholars have investigated cooperative strategies in the non-market domain (Galaskiewicz, 1985; Hardy & Phillips, 1998; Oliver, 1990; Phillips, Lawrence, & Hardy, 2000), the cooperative strategy literature generally privileges the market domain, where organizational survival and resource accumulation – rather than identity, ideology and legitimacy – are the primary measures of success.
Resource Partitioning and Collective Strategy Collective strategy and organizational ecology largely fail to account for potential coordination under conditions such as those we see in the case of micro-radio activism, and more generally, for organizational environments experiencing resource partitioning. Resource partitioning describes the process through which generalists consolidate and gain market share over time, leaving only peripheral resources underserved; eventually, specialist organizations arise to serve those niche markets, leaving a stratified market in which large generalists and small specialists coexist (Carroll, 1985; Carroll, Dobrev, & Swaminathan, 2002; Carroll & Hannan, 2000). Resource partitioning occurs primarily in industries marked by scale economies, heterogeneous resources, and limits on the range and adaptability of both specialists and generalists. Our case study illustrates the process of resource partitioning in the U.S. radio industry, which began in the 1990s with the consolidation of ownership, facilitated by the Telecommunications Act of 1996, and the consequent emergence of larger, generalist radio station owners. These generalists were able to achieve economies of scale by consolidating advertising sales and by eliminating local radio personalities, broadcasting instead standard programming that could be produced in a central location. Such consolidation and homogenization of radio programming left many listening audiences under-served, as was evidenced by
464
JO-ELLEN POZNER AND HAYAGREEVA RAO
the emergence of a movement in support of bringing diversity and differences of opinion, as well as local programming, to the airwaves. Clearly, those generalist broadcasters could not serve small constituencies given their business model, nor could micro-radio stations serve the larger, more mainstream audiences, limiting the adaptability of both organizational forms. Finally, it is important to note that, although audience resources are heterogeneous, production resources in the radio industry – most importantly, bandwidth – are both homogeneous and severely constrained, making this somewhat a special case of resource partitioning. The proliferation of specialist organizations in the resource partitioning process can only occur if specialists accomplish two goals: achieve legitimacy for the new organizational form, and gain access to critical resources. These conditions imply that specialists must compete with each other, as well as with incumbent generalists, on the dimensions of both ideology and resources. Inasmuch as organizations reduce uncertainty and deal with bounded rationality by dividing their environment into parts, or cognitive groups, thus limiting their attention to smaller ‘‘neighborhoods of action’’ (Levinthal & March, 1993; Peteraf & Shanley, 1997), the focus of this competition is likely to be other specialist organizations, who may be closer on both resource and ideological dimensions to one another than they are to the generalists. Moreover, the need to legitimate the specialist form tends to channel attention toward competing specialist definitions, rather than toward the established incumbent, whose legitimacy is already secure. This two-dimensional competition necessitated by the struggle for establishing legitimacy among new organizational forms tends to dominate interorganizational relations, particularly among specialists. At the same time, when production resources are both homogeneous and sharply constrained, as is the case for radio bandwidth, coordination among specialists is needed to dislodge established generalists from their privileged position. Both the intensity and the ideological identity dimensions of inter-specialist competition imply that cooperative strategy – the emergent phenomenon that is ‘‘the unintended result of aggregated repetitive patterns of pair wise interorganizational activity’’ (Dollinger, 1990, p. 267) – is highly unlikely to emerge spontaneously. Thus collective strategy, resulting from a more conscious decision to coordinate action, must emerge if specialists are to win both access to critical resources and legitimacy. Proposition 1. Given scarce and/or regulated production resources, representatives of a new organizational form must engage in collective strategy to gain access to resources and to win legitimacy.
Fighting a Common Foe: Enmity, Identity and Collective Strategy
465
Identity and Collective Strategy The enacting of collective strategy, however, is challenging under the circumstances described by resource partitioning theory, as illustrated in our case study. In the case of micro-radio, the scarce, critical resource in question – radio bandwidth – is regulated by the state, and the costs associated with licensing a full-power station represent a significant barrier to entry. The structure of the radio industry in the United States thus required concerted lobbying among specialists to succeed (Murray, 1978; Reger, Duhaime, & Stimpert, 1992). Such lobbying was particularly critical to the success of the micro-radio movement, given the aggressive response of incumbent broadcasters, brought about due to the scarcity of bandwidth available to both full-power, commercial, and low-power radio stations. Emerging specialist populations, however, do not generally meet the characteristics of groups that are generally successful in organizing collective efforts in lobbying for regulation (Maijoor & VanWitteloostuijn, 1996). Although specialists are small organizations, which are more likely to lobby than groups of large organizations (Olson, 1965) because of the relative ease of organizing collective action (Lindahl, 1987), they are in direct competition, and therefore less likely to cooperate. The expected benefits to each specialist are high, promoting lobbying (Downs, 1957), yet the potential costs to incumbents are potentially higher, engendering counter-lobbying (as we saw by the NAB, NPR, and other incumbent groups), reducing the incentive for collective strategy. Similarly, the ease of identifying both winners and losers (Maijoor & VanWitteloostuijn, 1996) makes joint lobbying efforts less likely. Finally, without representative organizations capable of restricting benefits of lobbying to those actively involved, organizing collective lobbying for regulation is exceedingly difficult (Maijoor & VanWitteloostuijn, 1996). To achieve collective action and lobbying for regulation, therefore, a force that unifies specialists by overcoming the competitive tendencies among them is essential; we argue that this force is group identification. Proposition 2. Representatives of a new organizational form will engage in collective strategy only if they identify themselves as members of a common group. How can a group identity sufficient to encourage collective action among a population of organizations engaged in competition for both resources and ideology? Whereas the processes and consequences of social identification at the individual and group levels are well studied in the social
466
JO-ELLEN POZNER AND HAYAGREEVA RAO
psychological and organizational behavior literature (Ashforth & Mael, 1989; Messick & Mackie, 1989; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987), the corresponding cognitive processes at the organizational level are somewhat less understood. There is ample evidence that managers view their industries as composed of groups of organizations (Baum & Lant, 2003; Fombrun, 1986; Fombrun & Zajac, 1987; Gripsrud & Gronhaug, 1985; Porac & Thomas, 1990, 1994; Porac, Thomas, & Badenfuller, 1989; Reger & Huff, 1993), and that organizations share social identities based on salient group characteristics (Dutton & Dukerich, 1991; Dutton, Dukerich, & Harquail, 1994; Fiol & Huff, 1992), yet little is known about the emergence of such cognitive groupings. At the individual level, actors choose to associate with organizations with which they identify (Blau, 1994; Douglas, 1986), engendering positive emotion from the moment of accession. Once the relationship is established, it generates a social category, leading to a sense of identification with the group. The sense of identifying with the group leads members to favor each other over non-members, and to feel greater positive emotions towards in-group members than out-group members (Brewer, 1979; Brewer, 1996; Tajfel & Turner, 1986; Turner et al., 1987). Similarly, social identification at the organizational level requires that actors define themselves with the group and align their values relative to the characteristics and actions of the group, increasing association and alignment, and thus attachment (Peteraf & Shanley, 1997). Thus it is not group membership per se, but rather the alignment of values, commitment to the group, and acceptance of a common code of behavior that accompany group identification that makes certain organizational populations both receptive to the idea of cooperative strategy and capable of executing such collective action, be it intended or emergent. Nevertheless, it is self-cognizance of group membership that allows emergent cooperative strategies to become more deliberate, and therefore more widespread (Dollinger, 1990). In surveying one’s organizational neighborhood, patterns emerge that tend to promote group identification and interorganizational cooperative strategizing. Group identification emerges when organizations recognize that they are interdependent in such a way that markets and competition cannot address (Astley & Fombrun, 1983; Dollinger, 1990; Doz & Baburoglu, 2000; Phillips, Lawrence, & Hardy, 2000). Often, this recognition results from ideological affinity or similarity, which is found to promote cooperative behavior, so long as the key resources upon which the organizations in question depend are sufficiently different to avert within-group competition (Ingram & Roberts, 2000; Simons & Ingram, 2004). In fact,
Fighting a Common Foe: Enmity, Identity and Collective Strategy
467
some degree of similarity may be a prerequisite for of cooperative strategy to emerge spontaneously within an organization population. There is evidence from the social movement literature that diversity increases the costs of collective action by inhibiting the formation of coalitions (Ingram & Rao, 2004; Schneiberg & Bartley, 2001), and, by extension, preventing the development of a unifying collective identity. Moreover, identification and cooperation are facilitated by ideological similarity and resource distinctiveness. Organizational populations with ideological affinity sympathize with one another, and tend to develop mutualistic relationships (Ingram & Simons, 2000). In contrast, Simons and Ingram (2004) find both negative relationships and increased competition among organizational populations with competing ideologies, just as ideologically similar organizations compete when they share resource profiles. Thus strategic group identification may not emerge when organizations are too ideologically distinct, nor when they compete directly for resources; rather, it emerges most easily within populations of ideologically similar but resource-distinct organizations. If it is true, as we saw in the case of pirate broadcasters, that organizations competing on both ideology and resource dimensions are not likely to develop an overarching group identity spontaneously, how can we account for the emergence of a common identity? In fact, active recognition of strategic group membership may be predicated only on the presence of a contrasting group that provides an external threat to critical resources (Doz & Baburoglu, 2000) or a threatening ideology (Simons & Ingram, 2004). This is particularly true of domains and organizational communities that have previously been marked by open competition, or among more heterogeneous organizational populations, where a new type of relationship must be negotiated to accomplish the identification required for cooperation (Phillips et al., 2000). Just as social identity theory holds that the presence of a rival group or ‘‘out-group’’ leads to increased cohesion and cooperation within a group, increasing the likelihood of collective action (Brewer & Kramer, 1986; Kramer & Brewer, 1984; Sherif, Harvey, White, Hood, & Sherif, 1961), so, too, is interorganizational cooperation encouraged by the presence of a salient out-group. For example, Ingram and Inman (1996) find that the presence of rivalry among similar strategic groups competing for the same resources – in their study, between hotels in Niagara Falls, New York, and Niagara Falls, Ontario – can promote within-group cohesion and, consequently, coordinated strategy. Likewise, in their study of chain stores, Ingram and Rao (2004) find that the emergence of an antagonistic social movement can spur an organizational community to identify as a group and
468
JO-ELLEN POZNER AND HAYAGREEVA RAO
take collective action through counter-mobilization (McAdam, Tarrow, & Tilly, 2001). In addition, the presence of a rival group has been shown to encourage cooperation by reducing free-riding (Bornstein, Erev, & Rosen, 1990; Erev, Bornstein, & Galili, 1993), thus increasing the efficacy of cooperative strategy. Thus, we argue, the presence of a common, salient, and threatening outgroup, and the common identification of that out-group, is sufficient to unite even ideologically different organizational populations, even when only certain subgroups are engaged in direct competition for resources. This is certainly the case presented by our analysis of the micro-radio movement in the United States, and is supported by other examples. In February 2005, the National Sugar Association, aided by the Center for Science in the Public Interest, a frequent critic of the National Sugar Association, and the National Grange, an agricultural and rural advocacy group, filed suit against McNeil Nutritionals, the division of Johnson & Johnson that markets Splenda, the most popular artificial sweetener in the U.S. Also joining in the legal effort were Merisant Worldwide, manufacturer of the artificial sweetener Equal, and other artificial sweetener manufacturers. The lawsuit was aimed at stopping McNeil Nutritionals from producing advertising that suggests that Splenda is natural because it is made from sugar, whereas its competitors claim that it is ‘‘a highly processed chemical compound’’ (Burros, 2005). The unusual group of complainants, all of whom were either competitors or ideological adversaries, banded together to thwart the growth of a dangerous newcomer to the artificial sweetener market. Introduced in 2000, Splenda’s share of the tabletop sweetener market rose from 37.3% in 2003 to 48.5% in 2004, while Equal’s share dropped by almost 4.5% (Warner, 2004). In this case, the presence of a successful competitor, which had created its own out-group through its differentiating advertising, was sufficient to bring together groups with competing ideologies that did not compete for resources, as well as organizations in direct competition with one another, indicating that sufficiently strong threats may help organizational populations overcome their differences to enact collective strategies. Drawing on the examples of both micro-radio and Splenda, it seems that a critical element to the successful framing of a common enemy in engendering group identification and collective strategy is the recognition of differences in relative intensity of competition for both resources and ideology. Identification of commercial broadcasters as the grand foe was possible only when the different factions of micro-radio activists recognized that their ideologies, although different inasmuch as they advocated for different features of a low-power radio service (totally non-commercial versus
Fighting a Common Foe: Enmity, Identity and Collective Strategy
469
commercial, e.g.), were still much closer to each other than any were to the ideology of commercial, chain-owned radio, which sought to delegitimate the low-power form and validity of the community radio concept completely. Moreover, acknowledgement that the local nature of competition for spectrum space made micro-radio activists only indirect competitors in the vast majority of cases (although a number of competing applications for LPFM broadcast frequencies were filed with the FCC), whereas commercial, chain-owned radio was a direct competitor for bandwidth in each location made intra-group competition much less salient than inter-group competition. This, in turn, promoted the feeling of identification as a micro-radio activist, and promoted collective strategy. Proposition 3. Representatives of a new organizational form will identify themselves as members of a common group only if they recognize the presence of a salient, shared rival. It is also important to note that the inclusive, community-oriented ideologies presented by the majority of micro-radio activists, in sharp contrast to earlier pirates, facilitated the development of a group identity on the part of external stakeholders, as well, with significant consequences for the legitimacy of the form and the success of collective strategy. Polos, Hannan, and Carroll (2002) pose organizational form as a type of socially coded identity, which provides organizations with assumptions about acceptable and expected behavior. This literature is focused on the social code – implying both categorical and penal codes – that govern organizational behavior based on organizational structure and niche, defining identity space as equivalent to resource space (Baron, 2004; Hannan, 2005; Hannan, Carroll, & Polos, 2003). Similarly, resource partitioning theory asserts that the legitimating effects of specialist organizational identities depend on the normative status of the specialist form, as well as its social visibility (Carroll & Hannan, 2000). Therefore, by reaching out to existing community groups, and by engendering a strong defensive reaction on the part of entrenched radio broadcasters, micro-radio activists promoted a certain understanding of the low-power organizational form, which created an identity for the overall movement. With a coherent external identity, it was easier to overcome intra-group competitive forces and to develop a coherent internal identity, which both enabled pro-social behavior and constrained legitimate action to exclude the relatively anarchistic behavior of earlier pirates. Thus, cooperation with outside groups may help in the development of a group identity that facilitates collective strategy among erstwhile competitors.
470
JO-ELLEN POZNER AND HAYAGREEVA RAO
CONCLUSION This essay was motivated by two gaps in the literature on collective strategy. Although strategy scholars have catalogued the types of collective strategy and the characteristics of industries in which they emerge, they have not sufficiently addressed the need for identification with a group on which collective strategy is often predicated. Similarly, whereas organizational ecologists have begun to address the question of identity, their perspective focuses more on the organizational structure defining forms and their behavioral and penal codes than the consequences of identifying with a group. We argue that group identification is essential for certain organizational communities to engage in successful, sustainable collective strategy. Specifically, we assert that emerging specialist forms struggling for legitimacy, particularly in industries undergoing the process of resource partitioning, will only enact collective strategies when they identify themselves as a group. Moreover, we contend that this identification is likely to emerge, particularly among groups of organizations competing on the bases of both resources and ideology, only under certain necessary and sufficient conditions. First, the presence of an external threat or a salient out-group encourages the development of a strong group identity that structures the opportunity for cooperation by calling attention to the commonalities among the emerging organizations; the salience of such a rival is absolutely imperative in inducing cooperation among groups that might otherwise be ideological foes. Second, competition with that out-group for critical resources provides the motivation to mobilize organizational communities to implement collective strategies. As the case of micro-radio demonstrates, both conditions must be met for cooperative strategy to be enacted. In the first, pirate phase of micro-radio activism, the presence of an enemy – the FCC – was insufficient to mobilize consistent, effective cooperation. It was only when micro-radio activists identified commercial radio, a group with which they were in direct competition for airspace and which was actively seeking to delegitimate the new organizational form, as its primary threat that it was able to develop a strong group identity capable of engaging in sustained, coordinated activity. Moreover, identification as a strategic group facilitated the development of an ideology among micro-radio activists capable of attracting ideological affinity groups, the community organizations that worked with the radio enthusiasts, and that themselves framed commercial radio as their enemy, to bring about regulatory change.
Fighting a Common Foe: Enmity, Identity and Collective Strategy
471
This case highlights an important gap in theory on organizational ecology. Although resource partitioning theory argues that new organizational forms emerge to service underserved market segments when market share becomes concentrated in the hands of generalists (Carroll, 1985; Carroll et al., 2002; Carroll & Hannan, 2000; Carroll & Swaminathan, 2000), it generally neglects the mechanisms through which those specialists gain legitimacy. Yet, in industries that are dominated by entrenched incumbents, particularly when subject to government regulation and marked by scarce resources, barriers to entry are high, and may be broken only through collective action. The case of micro-radio activists, however, suggests that group identification through the recognition that incumbents are more salient rivals for legitimacy and resources than are other new entrants can spur collective strategy. Through coordinated action, new organizations may overcome strategic barriers to lobbying, thus gaining legitimacy and numbers necessary to compete effectively against incumbents. We also add to the theory of resource partitioning by specifying conditions, the under which specialists will emerge when production resources are homogeneous, constrained, and regulated. Generally, resource partitioning theory assumes heterogeneous resources, primarily as relates to dimensions of taste and consumption (Carroll, 1985; Carroll et al., 2002; Carroll & Hannan, 2000). In this essay, we study an industry in which production resources – in the case of micro-radio, radio bandwidth – are completely homogeneous and sharply limited; thus, even if new organizations appeal to specialized groups of consumers, they must fight for constrained production resources. This requirement makes collective action, in the form of lobbying for access to scarce resources, even more important. Given the significant barriers to joint action in such industries (e.g., Maijoor & VanWitteloostuijn, 1996), group identification is critical to the enactment of collective strategy. This argument applies not only to the emergence of specialists through the process of resource partitioning, but also the emergence of new organizational forms more generally. Similar to McKendrick, Jaffee, Carroll, and Khessina’s (2003) assertion that perceptually focused identities facilitate the emergence of new organizational forms, we argue that group identification more generally enables the establishment and legitimation of a new form. It is not the fact of group identification itself, but rather the ability to overcome rivalry to enable collective action that enables this legitimation. As demonstrated in the contrast between radio pirates and later micro-radio activists, without forging a common identity that is capable of reaching out to others with complementary ideologies and goals, it is much more difficult for emerging organizational forms to establish their legitimacy.
472
JO-ELLEN POZNER AND HAYAGREEVA RAO
Moreover, although this model of collective strategy applies specifically to the entry of new sub-forms within existing industries, it can be generalized to the emergence of entirely new organizational forms or industries. When new forms emerge within existing industries, incumbents represent a salient rival on both ideological and resource dimensions, providing a natural focus for group identification. This does not rule out the possibility, however, of organizational communities in new industries defining a common identity that would enable them to overcome internal competition for legitimacy and resources and enact coordinated action. Under such circumstances, such identification is more likely to take shape implicitly, through repeated interaction, than it is to emerge through a conscious decision to work together. Thus, our argument will apply to any industry, new or extant, in which a new organizational form is emerging and in which a salient enemy can be identified. For our argument to hold, it is necessary for the new organizational form to be engaged in a fight for both market and non-market resources, and for the market resources to be sharply constrained or regulated, such that collective action is required to gain access. Without the struggle for both legitimacy and resources, and without the constraints to acquiring those resources, collective action among commensal organizations is likely to be less critical to organizational survival, making group identification less necessary. In the chapter by Shipilov et al. (2006), for example, wherein potential partners compete only in market domains, this type of identification is clearly not necessary, and resource, size, and status complementarity are more important predictors of interorganizational cooperation. Finally, this collective action can only emerge when there exists a salient out-group competing for both market and non-market resources capable of generating group identification. An important limitation to the generalizability of our argument is the character of the resource space and the market in question. For example, in the United States, the broadcast media industry is fairly conservative, and its regulation is highly politicized, further constraining the legitimacy of new entrants. In other countries, however, radio may be governed by a more liberal regime, or one more sensitive to outside viewpoints, lessening the need for specialist organizations to struggle to obtain legitimacy. Moreover, the size of the resource space may be consequential. Generalists in a market as large as the United States have more to lose, and thus may be more difficult to challenge, than they are in smaller markets. In smaller markets, therefore, the competition for resources and legitimacy may be less intense, so that collective strategies are not required to ensure the survival of new organizational forms.
Fighting a Common Foe: Enmity, Identity and Collective Strategy
473
It is important to note that we are not making a case for the superiority of collective strategy over cooperative strategy. Rather, we are arguing that cooperative strategies may not emerge naturally among commensal communities. In these cases, the development of a collective identity promotes the development of explicit collective strategy, which is likely to be more effective than cooperative strategy because of its explicit nature. Thus, we are not making a universal assertion, but rather one specific to the development of collective identity in fighting a common foe. Given these limitations, our argument should be seen as a complement to the essay by Shipilov et al. (2006) in this volume. Their chapter focuses on the choice of strategic partner when organizations are not involved in direct competition, and when they possess complementary market resources, whereas our area of inquiry relates to organizations in direct competition for market and non-market resources. In both cases, membership in a group or network is critical to the development of collective strategy among specialist organizations, although identification with such a group is not necessarily a precondition for cooperation in Shipilov et al.’s model. Future research might extend both models by testing the relationships each hypothesizes under the market conditions specified by the other. Finally, our essay points to a future direction for research on cooperative strategy: the sustainability of collective action. That is, for how long can would-be competitors work together to compete against a common threat? Can identification with an organizational community, which is based on the presence of a salient out-group, survive the presence of that threat? One is tempted to argue that, when identification is based on shared group characteristics – particularly ideology – it should persist and encourage future collaborative efforts. Although such positive identities might persist, even absent institutionalization through the development of communal organizations, the opposite is likely to be true of identities forged only of common enmity, as is the case with the complainants in the anti-Splenda lawsuits. Such cooperation is not borne of the desire to reduce competition in general, but rather to weaken the competitive position of one particularly strong rival. Future research should test this proposition to further discover the limits to collective strategy.
REFERENCES Aldrich, H. E. (1979). Organizations and environments. Englewood Cliffs, NJ: Prentice-Hall. Anderson, J. (2004a). A can of worms: Pirate radio as public intransigence on the public airwaves. Unpublished Master’s, University of Wisconsin, Madison, Madison, Wisconsin.
474
JO-ELLEN POZNER AND HAYAGREEVA RAO
Anderson, J. (2004b). The history of LPFM. Retrieved June 3, 2004, from http://www.diymedia. net/feature/fhistlpfm.htm Ashforth, B. E., & Mael, F. (1989). Social identity theory and the organization. Academy of Management Review, 14(1), 20. Astley, W. G. (1984). Toward an appreciation of collective strategy. Academy of Management Review, 9(3), 526–535. Astley, W. G., & Fombrun, C. J. (1983). Collective strategy – social ecology of organizational environments. Academy of Management Review, 8(4), 576–587. Barnett, W. P. (1990). The organizational ecology of a technological system. Administrative Science Quarterly, 35(1), 31–60. Barnett, W. P., & Woywode, M. (2004). From red Vienna to the Anschluss: Ideological competition among Viennese newspapers during the rise of national socialism. American Journal of Sociology, 109(6), 1452–1499. Baron, J. N. (2004). Employing identities in organizational ecology. Industrial and Corporate Change, 13(1), 3–32. Baum, J. A. C., & Lant, T. K. (2003). Hits and misses: Managers’ (mis)categorization of competitors in the Manhattan hotel industry. Geography and Strategy, 20, 119–156. Bernstein, M. (1997). Celebration and suppression: The strategic uses of identity by the lesbian and gay movement. American Journal of Sociology, 103(3), 531–565. Blau, P. M. (1994). Structural contexts of opportunities. Chicago: University of Chicago. Boehlert, E. (2003). New life for little radio? Opposition to big media could invigorate low-power FM radio. http://dir.salon.com/story/news/feature/2003/07/23/lpfm/index_np.html in Salon.com: Salon.com. Bornstein, G., Erev, I., & Rosen, O. (1990). Intergroup competition as a structural solution to social dilemmas. Social Behaviour, 5(4), 247–260. Bresser, R. K., & Harl, J. E. (1986). Collective strategy – vice or virtue. Academy of Management Review, 11(2), 408–427. Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A cognitivemotivational analysis. Psychological Bulletin, 86(2), 307–324. Brewer, M. B. (1996). In-group favoritism: The subtle side of intergroup discrimination. In: D. M. Messic, & A. Tenbrunsel (Eds), Codes of conduct: Behavioral research and business ethics (pp. 160–171). New York: Russel Sage Fund. Brewer, M. B., & Kramer, R. M. (1986). Choice behavior in social dilemmas: Effects of social identity, group size, and decision framing. Journal of Personality & Social Psychology, 50(3), 543–549. Burros, M. (2005). Splenda’s ‘sugar’ claim unites odd couple of nutrition wars. New York Times, February 15. Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in populations of organizations. American Journal of Sociology, 90(6), 1262–1283. Carroll, G. R., Delacroix, J., & Goodstein, J. (1990). The political environment of organizations: An ecological view. In: B. M. Staw, & L. L. Cummings (Eds), The evolution and adaptation of organizations (pp. 67–100). Greenwich, CT: JAI. Carroll, G. R., Dobrev, S., & Swaminathan, A. (2002). Organizational processes of resource partitioning. Research in Organizational Behavior, 24, 1–40. Carroll, G. R., & Hannan, M. T. (2000). The demography of corporations and industries. Princeton, NJ: Princeton University Press.
Fighting a Common Foe: Enmity, Identity and Collective Strategy
475
Carroll, G. R., & Swaminathan, A. (1992). The organizational ecology of strategic groups in the American Beer Brewing Industry from 1975 to 1990. Industrial and Corporate Change, 1, 65–97. Carroll, G. R., & Swaminathan, A. (2000). Why the microbrewery movement? Organizational dynamics of resource partitioning in the US brewing industry. American Journal of Sociology, 106(3), 715–762. Clarke, B. (2004). The transistor triangle: NAB, congress and FCC collaborate to keep airwaves a corporate wasteland. From http://www.media-alliance.org/article. php?story=20040514121048391 Coopman, T. (1997a). Free radio vs. the FCC: A case study of micro broadcasting. Paper presented at the National Communication Association, Mass Communication Division, Chicago, IL. Coopman, T. (1997b). Pirates to micro broadcasters: The rise of the micro radio movement. Paper presented at the Western States Communication Association, Monterey, California. CPB Board of Directors. (1999). Mission statement. Retrieved June 8, 2004, from http:// www.cpb.org/about/corp/mission.html Cyert, R. M., & March, J. G. (1963). A behavioral theory of the firm. Englewood Cliffs, NJ: Prentice-Hall. DeBarros, A. (1998). Consolidation changes face of radio. USA Today, July 7. Dobrev, S. D. (2000). Decreasing concentration and reversibility of the resource partitioning process: Supply shortages and deregulation in the Bulgarian newspaper industry, 1987– 1992. Organization Studies, 21(2), 383–404. Dobrev, S. D. (2001). Revisiting organizational legitimation: Cognitive diffusion and sociopolitical factors in the evolution of Bulgarian newspaper enterprises, 1846–1992. Organization Studies, 22(3), 419–444. Dollinger, M. J. (1990). The evolution of collective strategies in fragmented industries. Academy of Management Review, 15(2), 266–285. Douglas, M. (1986). How institutions think. Syracuse, NY: Syracuse University Press. Downs, A. (1957). An economic theory of democracy. New York: Harper & Row. Doz, Y. L., & Baburoglu, O. (2000). From competition to collaboration: The emergence and evolution of R&D. In: D. Faulkner, & M. d. Rond (Eds), Cooperative strategy: Economic, business, and organizational issues (pp. 173–192). New York: Oxford University Press. Dunifer, S. (2004). About free radio Berkeley. Retrieved June 19, 2004, from http://www.freeradio.org/index.php Dutton, J. E., & Dukerich, J. M. (1991). Keeping an eye on the mirror: Image and identity in organizational adaptation. Academy of Management Journal, 34(3), 38. Dutton, J. E., Dukerich, J. M., & Harquail, C. V. (1994). Organizational images and member identification. Administrative Science Quarterly, 39(2), 239–263. Dvorkin, J. A. (2000). What listeners say and where NPR’s money comes from. Retrieved June 8, 2004, from http://www.npr.org/yourturn/ombudsman/000620.html Edmonson, R. (2000). Rising up: Class warfare in America from the streets to the airwaves. San Francisco, CA: Librad Press. Erev, I., Bornstein, G., & Galili, R. (1993). Constructive intergroup competition as a solution to the free rider problem – a field experiment. Journal of Experimental Social Psychology, 29(6), 463–478.
476
JO-ELLEN POZNER AND HAYAGREEVA RAO
Federal Communications Commission. (1999). Notice of proposed rulemaking in the creation of a low power radio service. Federal Communications Commission. (2000). Report and order. Washington, DC: FCC. Federal Communications Commission: Experimental Radio, Auxiliary, Special Broadcast and Other Program Distributional Services. (2002). Title 47, Chapter 1, Part 74 1231. Fiol, C. M., & Huff, A. S. (1992). Maps for managers – where are we – where do we go from here. Journal of Management Studies, 29(3), 267–285. Fisher, M. (1998). Blackout on the dial. American Journalism Review. http://www.ajr.org/ Article.asp?id=1327 Flugennock, M. (1998). Against FCC/NAB suppression of community radio. Washington, DC (10.05.90). Retrieved June 19, 2004, from http://www.sinkers.org/microradio/ Fombrun, C. J. (1986). Structural dynamics within and between organizations. Administrative Science Quarterly, 31(3), 403–421. Fombrun, C. J., & Zajac, E. J. (1987). Structural and perceptual influences on intraindustry stratification. Academy of Management Journal, 30(1), 33–50. Galaskiewicz, J. (1985). Interorganizational relations. Annual Review of Sociology, 11, 281–304. Greve, H. R. Pozner, J. -E., & Rao, H. (2006). Vox populi: Resource partitioning, oganizational proliferation and the cultural impact of the insurgent micro-radio movement. American Journal of Sociology (Forthcoming). Gripsrud, G., & Gronhaug, K. (1985). Structure and strategy in grocery retailing – a sociometric approach. Journal of Industrial Economics, 33(3), 339–347. Haines, H. H. (1997). Black radicalization and the funding of civil rights: 1957–1970. In: D. McAdam, & D. A. Snow (Eds), Social movements: Readings on their emergence, mobilization and dynamics. Los Angeles: Roxbury. Hall, J. (2005). Telephone interview. J.-E. Pozner (Ed.). June 15, Chicago, IL. Hannan, M. T. (2005). Ecologies of organizations: Diversity and identity. Journal of Economic Perspectives, 19(1), 51–70. Hannan, M. T., Carroll, G. R., & Polos, L. (2003). The organizational niche. Sociological theory, 21(4), 309–340. Hardy, C., & Phillips, N. (1998). Strategies of engagement: Lessons from the critical examination of collaboration and conflict in an interorganizational domain. Organization Science, 9(2), 217–230. Hunt, C. S., & Aldrich, H. E. (1998). The second ecology: Creation and evolution of organizational communities. Research in Organizational Behavior, 20(20), 267–301. Ingram, P., & Inman, C. (1996). Institutions, intergroup competition, and the evolution of hotel populations around Niagara Falls. Administrative Science Quarterly, 41, 629–658. Ingram, P., & Rao, H. (2004). Store wars: The enactment and repeal of anti-chainstore legislation in America. American Journal of Sociology, 110, 446–481. Ingram, P., & Roberts, P. W. (2000). Friendships among competitors in the Sydney hotel industry. American Journal of Sociology, 106(2), 387–423. Ingram, P., & Simons, T. (2000). State formation, ideological competition, and the ecology of Israeli workers’ cooperatives, 1920–1992. Administrative Science Quarterly, 45(1), 25–53. Kennard, W. E. (2000a). Statement of FCC Chairman William Kennard on low-power FM. Kennard, W. E. (2000b). The voice of the people. Washington Post, p. A23. Kramer, R. M., & Brewer, M. B. (1984). Effects of group identity on resource use in a simulated commons dilemma. Journal of Personality and Social Psychology, 46(5), 1044–1057.
Fighting a Common Foe: Enmity, Identity and Collective Strategy
477
Leblebici, H., Salancik, G. R., Copay, A., & King, T. (1991). Institutional change and the transformation of interorganizational fields: An organizational history of the U.S. radio broadcasting industry. Administrative Science Quarterly, 36(3), 31. Leggett, N. E., Leggett, J. F., & Schellhardt, D. J. (1997). Petition for a microstation radio broadcasting service. Retrieved June 19, 2004. http://www.fcc.gov/Bureaus/Mass_Media/ Filings/rm9208.pdf Levinthal, D. A., & March, J. G. (1993). The myopia of learning. Strategic Management Journal, 14, 95–112. Lindahl, F. W. (1987). Accounting standards and Olson’s theory of collective action. Journal of Accounting and Public Policy, 6, 59–72. Maijoor, S., & VanWitteloostuijn, A. (1996). An empirical test of the resource-based theory: Strategic regulation in the Dutch audit industry. Strategic Management Journal, 17(7), 549–569. Mayer, P. (1998). Talk hard: Pirate radio in America. Radio Department, School of Journalism, Columbia University, New York. McAdam, D., Tarrow, S., & Tilly, C. (2001). Dynamics of contention. Cambridge: Cambridge University Press. McKendrick, D. G., Jaffee, J., Carroll, G. R., & Khessina, O. M. (2003). In the bud? Disk array producers as a (possibly) emergent organizational form. Administrative Science Quarterly, 48(1), 60–93. Mediageek. (2004). LPFM bill passes Senate Commerce Committee. From http://www.mediageek.org/archives/002317.html#002317 Messick, D. M., & Mackie, D. M. (1989). Intergroup relations. Annual Review of Psychology, 40, 45–81. Minkoff, D. C. (2002). The emergence of hybrid organizational forms: Combining identitybased service provision and political action. Nonprofit and Voluntary Sector Quarterly, 31(3), 377–401. Murray, E. A. (1978). Strategic choice as a negotiated outcome. Management Science, 24(9), 960–972. Notice of Proposed Rulemaking in the Creation of a Low Power Radio Service, January 28, 1999. Oliver, C. (1988). The collective strategy framework – an application to competing predictions of isomorphism. Administrative Science Quarterly, 33(4), 543–561. Oliver, C. (1990). Determinants of interorganizational relationships – integration and futuredirections. Academy of Management Review, 15(2), 241–265. Olson, M. (1965). The logic of collective action. Cambridge, MA: Harvard University Press. Opel, A. (2004). Micro radio and the struggle over broadcast policy. Westport, CT: Praeger. Pennings, J. M. (1981). Strategically interdependent organizations. In: P. C. Nystrom & W. H. Starbuck (Eds), Handbook of organizational design, (Vol. 1, pp. 433–435). New York: Oxford University Press. Peteraf, M., & Shanley, M. (1997). Getting to know you: A theory of strategic group identity. Strategic Management Journal, 18, 165–186. Pfeffer, J., & Nowak, P. (1976). Joint ventures and interorganizational interdependence. Administrative Science Quarterly, 21(3), 398–418. Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations: A resource dependence perspective. New York: Harper & Row.
478
JO-ELLEN POZNER AND HAYAGREEVA RAO
Phillips, N., Lawrence, T. B., & Hardy, C. (2000). Inter-organizational collaboration and the dynamics of institutional fields. Journal of Management Studies, 37(1), 23–43. Polos, L., Hannan, M. T., & Carroll, G. R. (2002). Foundations of a theory of social forms. Industrial and Corporate Change, 11, 85–115. Porac, J. F., & Thomas, H. (1990). Taxonomic mental models in competitor definition. Academy of Management Review, 15(2), 224–240. Porac, J. F., & Thomas, H. (1994). Cognitive categorization and subjective rivalry among retailers in a small city. Journal of Applied Psychology, 79(1), 54–66. Porac, J. F., Thomas, H., & Badenfuller, C. (1989). Competitive groups as cognitive communities – the case of Scottish Knitwear Manufacturers. Journal of Management Studies, 26(4), 397–416. Prometheus Project. (2000). Low power signals: Special interest noise. From http://www.prometheusradio.org/artnoise.shtml Radio Mutiny. (2004). Your dial was made for revolution: An introduction to radio mutiny. From http://www.prometheusradio.org/piratedial.html Reece, D. (1998). KREV fans rally for radio diversity: Buyout of progressive outlet sparks action, CD. Billboard, April 4. Reese, S. (2005). Telephone interview. J. -E. Pozner (Ed.). June 15, Cambridge Springs, PA. Reger, R. K., Duhaime, I. M., & Stimpert, J. L. (1992). Deregulation, strategic choice, risk and financial performance. Strategic Management Journal, 13(3), 189–204. Reger, R. K., & Huff, A. S. (1993). Strategic groups – a cognitive perspective. Strategic Management Journal, 14(2), 103–123. Rosenblum, A. (1997). Exclusive! Radio mutiny pirates speak out after one year of operation. From http://www.radiodiversity.com/radiomutiny.html Ruef, M. (2000). The emergence of organizational forms: A community ecology approach. American Journal of Sociology, 106(3), 658–714. Ruggiero G. v. FCC. (2002). Greg Ruggiero v. Federal Communicayion Commission. United States of America No. 00-1100, United States Court of Appeals for the District of Columbia Circuit. Sakolsky, R. (1992). Zoom black magic liberation radio: The birth of the micro-radio movement in the USA. In: B. Girard (Ed.), A passion for radio: Radio waves and community. Montreal: Black Rose Books. Schellhardt, D. (2005). Telephone interview. J.-E. Pozner (Ed.). July 5, Chicago, IL. Schneiberg, M., & Bartley, T. (2001). Regulating American industries: Markets, politics, and the institutional determinants of fire insurance regulation. American Journal of Sociology, 107(1), 101–146. Sharlet, J. (2003). Big world: How clear channel programs America. Harper’s Magazine, pp. 37–45. Sherif, M., Harvey, O. J., White, B. J., Hood, W. R., & Sherif, C. W. (1961). Intergroup conflict and cooperation. Norman, OK: Institute of Group Relations. Shipilov, A., Rowley, T. J., & Aharonson, B. (2006). When do networks matter? A study of tie formation and decay. In: J. A. C. Baum, S. D. Dobrev, A. v. Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 481–519). Oxford, UK: JAI. Simons, T., & Ingram, P. (2004). An ecology of ideology: Theory and evidence from four populations. Industrial and Corporate Change, 13(1), 33–59. Skinner, R. J. (1998). Proposal for creation of the low power FM (LPFM) broadcast service. Retrieved June 19, 2004. http://www.fcc.gov/Bureaus/Mass_Media/Filings/rm9242.pdf
Fighting a Common Foe: Enmity, Identity and Collective Strategy
479
Smith, J., & Rosenfeld, H. (1999). Radio for people, not profit. Dollars and Sense: The Magazine of Economic Justice, p. 223. Spangler, M. (1997). FCC tries to sink pirate operations. Radio & Records. http://www.radiodiversity.com/fcccrackdown.html Sterling, C. H., & Kittross, J. M. (1978). Stay tuned: A concise history of American Broadcasting. Belmont, CA: Wadsworth. Streeter, T. (1996). Selling the air: A critique of the policy of commercial broadcasting in the United States. Chicago: University of Chicago Press. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In: S. Worchell, & W. G. Austin (Eds), Psychology of intergroup relations. Monterey, CA: Brooks. Thompson, J. D. (1967). Organizations in action: Social science bases of administrative theory. New York: McGraw-Hill. triDish, P. (2005). Telephone interview. J.-E. Pozner (Ed.). June 22, Philadelphia, PA. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford, England: Basil Blackwell. Tymon, P. (1998). Those opposed. Retrieved June 19, 2004, from http://www.diymedia.net/ feature/lpfm/f071298.html Van de Ven, A. H. (1976). On the nature, formation, and maintenance of relations among organizations. Academy of Management Review, 1, 24–36. Van de Ven, A. H., & Walker, G. (1984). The dynamics of interorganizational coordination. Administrative Science Quarterly, 29(4), 598–621. Warner, M. (2004). A something among the sweet nothings. New York Times, December 22. Whetten, D. A., & Leung, T. K. (1979). Instrumental value of interorganizational relations – antecedents and consequences of linkage formation. Academy of Management Journal, 22(2), 325–344.
This page intentionally left blank
480
WHEN DO NETWORKS MATTER? A STUDY OF TIE FORMATION AND DECAY$ Andrew V. Shipilov, Tim J. Rowley and Barak S. Aharonson ABSTRACT Interorganizational partner selection decisions are plagued with uncertainty. When making partnering decisions, firms strive to answer two questions: does the prospective partner have resources which can be used to generate value in the relationship; and will the partner be willing to actively share these resources and cooperate in good faith? Answers to these questions help reduce three types of uncertainty – partner capability uncertainty, partner competitiveness uncertainty and partner reliability uncertainty. For a relationship to benefit both partners, they have to possess complimentary resources of comparable quality, avoid explicit competition as well as be willing to engage in the cooperative behaviors within the confines of their relationship. In this paper, we examine the importance of prospective partners’ characteristics (differences in size, status and specialization) as well as their network characteristics (existence of a common partner and membership in
$
This research was supported by a grant from the Social Sciences and Humanities Research Council of Canada.
Ecology and Strategy Advances in Strategic Management, Volume 23, 481–519 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23015-8
481
482
ANDREW V. SHIPILOV ET AL.
the same clique) to the formation and longevity of their social relationships, as these characteristics reduce firms’ value generation and partner reliability uncertainty. Research on interorganizational alliances and networks is dominated by investigations of the partnering process; and scholars’ attention has primarily focused on two features of this process – tie formations and partner uncertainty mitigation. Concentrating research on these aspects of the partnering process has generated two important contributions. One insight is a clear understanding of the types and sources of information firms use to reduce partnering uncertainty: Firms tend to examine partner attributes to reduce capability uncertainty (do they possess valuable and complementary skills?) and rely on network attributes to mitigate partner reliability uncertainty (will they engage in cooperate behavior leading to mutual gain?) (Gulati & Gargiulo, 1999; Gulati, 1995). Perhaps more importantly, this research provides an explanation of why network structures tend to be ‘‘cliquey’’ in shape and stable over time. Owing to their desire to reduce partnering uncertainty, firms are likely to form ties with their past partner and their partners’ partners, which creates dense local region within a stable network (Walker, Kogut, & Shan, 1997; Chung, Singh, & Lee, 2000; Li & Rowley, 2002). While the advantage of this focused approach is a deep understanding of key factors in the partner selection process, the trade-off is a lack of breadth. Partner selection research has led to a model of partnering behavior and network dynamics that emphasizes tie formation but not tie termination processes. This imbalance is criticized because network dynamics are as much about detaching from existing relationships as they are about forming new ties (Burt, 2000). Several scholars have lamented that tie formation, longevity and termination are confounded in the literature, even though they involve different types of information and uncertainty (Burt, 2000; Powell, White, Koput, & Owen-Smith, 2004). In addition, by stressing the importance of partner uncertainty reduction in the tie formation process, other factors and motives that may influence partner selection may be underplayed. For example, alliances provide a means of coordinating actions among rivals, thus giving firms strategic motives for coordinating efforts with particular partners beyond, or even in conflict with, partner-specific concerns about their reliability and capabilities. As an initial step toward increasing our breadth of knowledge concerning interorganizational partnering and its influence on network structures and dynamics, we separate the tie formation and renewal/termination processes. In this study, we suggest that the logics of attachment are different depending on
When do Networks Matter? A Study of Tie Formation and Decay
483
whether considering a new tie – partners that have had no prior ties with each other – or potentially renewing an existing/past tie – partners that have at least one ongoing or past tie with each other. Our argument is that firms will emphasize different types of uncertainty reduction and source different types of information when forming new ties compared to renewing existing/past ties. More specifically, we test the idea that network characteristics, which assist in reducing partner reliability uncertainty, are more important for new than for ongoing relationships. We argue that information about partner reliability, contained within the network of interfirm relationships, is more relevant for the formation of early stage ties between partners. Once the relationship has been formed, partners experience each other’s reliability through direct interactions, which reduce the value of network-based information. Further, while research shows that partner-specific concerns drive firms to find the best partner (Li & Rowley, 2002), we augment this work with the insight from ecology and game theory research that firms are motivated by the potential to moderate competitive intensity – i.e., form and maintain ties with rivals. Thus, they are not only managing partner uncertainty, but also competitive uncertainty through their alliance decisions. Our study contributes to the partner selection process as well as network dynamics. Current research has taken the field a great distance by focusing on the tie formation process. However, the result is an oversimplified view of how networks influence organizations. Research implies that the relationship between partner selection and network structure is recursive: partnering choices lead to the emergence of the network, but subsequently, the network affects partnering choices. Once the network structure emerges, it determines, in part, its own evolution (Galaskiewicz & Burt, 1991; Gulati & Gargiulo, 1999; Li & Rowley, 2002), because the logic of attachment guiding partnering choices is encoded in the network topology (Powell, Koput, Smith-Doerr, & Owen-Smith, 1999; Baum, Rowley, Shipilov, & Chuang, 2005). In this study, we question this argument. We suggest that network attributes are significantly discounted once first-hand experience is available for evaluating a partner’s reliability. Supporting evidence would suggest that network attributes affect new ties, and thus, are influential in the emergence of new networks as organizations form their initial ties. Once well established, however, the network plays a less significant role in its own evolution. By disaggregating partner selection into new and ongoing relationships, we strive to differentiate the factors influencing the initial emergence of a network from its subsequent evolution. Increasing our understanding of how networks are created and how they evolve is important because social structures create opportunities and constraints for organizations embedded within them.
484
ANDREW V. SHIPILOV ET AL.
Because these opportunities and constraints are unevenly distributed across networks, social structures are sources of competitive (dis)advantages depending on organizations’ positions in their networks. Thus, the creation and evolution of networks is paramount to understanding how organizations can derive network-based strategic advantages. And the differences in the process by which partnerships are formed, as opposed to renewed/terminated, are fundamental to how networks develop and change. In the sections that follow, we test our argument through two empirical comparisons. We compare the influence of network factors on the formation of new ties with their influence on the dissolution of existing relationships as a means of examining how network typologies influence partner selection and therefore network dynamics. In addition, to gain a better understanding of our findings we evaluate them against the results for non-network factors included in our analysis. Our hypotheses are tested on a population of Canadian investment banks forming syndicates to underwrite public offerings over a 38-year period (1952–1989).
THEORY AND HYPOTHESES The maxims ‘‘do not go it alone’’ and ‘‘cooperate with your competitors and win’’ have dominated the mindset of managers in many modern industries. Researchers studying this phenomenon have focused on how organizations manage the uncertainty inherent in partner selection decisions (e.g., Baum et al., 2005) and how the formation of relationships can help actors augment their performance (e.g., Shipilov, 2006; Pozner & Rao, 2006). Specifically, the majority of studies investigate the types of uncertainties related to partner selection and the criteria organizations use to mitigate the fear of partnering with unreliable and incapable partners. Two most commonly explored types of uncertainty are partner capability and partner reliability uncertainty. The third type – partner competitiveness uncertainty – has received more attention in the game theory literature, although it currently also makes a foray into the networks literature (Gimeno, 2004). Below, we examine each of these types of uncertainty in detail and then formulate hypotheses linking them to firms’ partnering decisions. Partner Capability Uncertainty A core rationale for forming a partnership is to access and combine capabilities (Barney & Hansen, 1994). By pooling resources and capabilities with those of
When do Networks Matter? A Study of Tie Formation and Decay
485
other companies, organizations can initiate projects that they could not have successfully done alone, ultimately creating value and enjoying performance improvements (Chung et al., 2000). Furthermore, the resource dependence perspective suggests that organizations attempt to form relationships with others that possess complementary resources, skills and capabilities (Pfeffer & Salancik, 1978). Mutual gain is then possible when partners can complement each other’s weaknesses (Hamel, Doz, & Prahalad, 1989). Prior research on the dynamics of interorganizational networks has found support for the assertion that firms strive to partner with firms that have complementary resources. Nohria and Garcia-Pont (1991) find that in the global automobile industry, firms in different strategic groups form alliances in order to gain complementary resources at each other’s disposal. In a three-industry study, Gulati and Gargiulo (1999) find that firms active in different industry segments, and serving customers in different geographical areas, were more likely to form alliances than firms serving similar geographical areas or catering to similar industry segments. Chung et al. (2000) also found that the likelihood of an alliance formation between two investment banks increased when these banks were catering to different investor types (e.g., insurance companies, mutual funds, pension funds) and serving different geographical areas. Status and functional dissimilarity. Assessing whether a potential partner possesses high-quality capabilities is an ambiguous process, because few tangible measures are available. Consequently, organizations turn to factors that are associated with the potential partner’s overall quality. Podolny (1993) illustrates that organizations consider their potential partners’ status as a proxy for quality when more direct measures are not available. An organization’s status as perceived by others is dependent on the status of other firms with whom the focal firm interacts. For example, organizations with ties to prestigious partners, which are perceived as being able to evaluate others’ abilities, are often viewed by third parties as possessing valuable skills, resources or capabilities (Stuart, 2000). Because organizations are status seeking, and their status rankings are a function of the status rankings of their partners, lower status organizations seek ties with higher status organizations, who attempt to avoid ties with lower status players. Consequently, organizations tend to collaborate with similar status partners. In addition to assessing the quality of a potential partner’s capabilities, organizations evaluate whether the combination of capabilities in the partnership will be complementary. Because organizations utilize partnerships to access capabilities they do not possess, they tend to search for partners that are functionally dissimilar (Li & Rowley, 2002). In particular, research
486
ANDREW V. SHIPILOV ET AL.
suggests that when organizations make these comparisons, they consider firm attributes such as size, scope, market focus and product mix (Greve, 1999). Size is a particularly salient factor because it is easily observed. The larger the firm, the more organizational capital it possesses (Levinthal, 1991) in terms of financial resource legitimacy and goodwill (Baum, 1996). Greater scale implies greater efficiency in the production of goods or in delivering services, stemming from, for example, the ability of large firms to purchase and use specialized equipment; savings from operational expansions and quicker returns on investments in capacity expansion; greater employee specialization based on the intricate division of labor; extraction of rents from experiential learning and the benefits of high frequency with which same tasks are carried out; and finally, reduction in per unit overhead costs (Dobrev & Carroll, 2003). However, this efficiency comes at a price: larger firms suffer from strong inertial pressures. Efficiencies that the firm has achieved as a result of its scale could limit this firm’s ability to adapt to new products or demands of their customers. In contrast, balancing on the verge of extinction, small firms are subject to lower inertial pressures as they take risks in developing new products or services (Levinthal, 1991). Given a choice of potential alliance partners, large firms should be attracted to smaller firms in order to access smaller firms’ entrepreneurial talent and capitalize on their flexibility. At the same time, smaller firms would also be more likely to benefit from relationships with larger firms, because such arrangements would allow smaller companies to access large firm’s resources and organizational capital. As Dobrev and Carroll (2003) point out, greater differences on the size gradient between a large and a small firm mean higher risks of failure for a small organization, as it becomes less and less able to extract scale benefits from its operations. Thus, small firms’ relationships with large firms could enable the former to draw upon at least some scale benefits available to the large partner, ultimately improving the small firm’s survival chances. Furthermore, small and large firms are less likely to compete within the confines of their relationships. Baum and Mezias (1992) formulate the model of size-localized competition, the main assumption of which is that greater similarity in size leads to greater potential for competition among firms. In contrast, greater difference in size between firms means that they draw upon different resources and exploit different customer segments (Dobrev, Kim, & Carroll, 2003). For example, while large firms might focus on mass production and engage in scale competition, small ones might cater to specialized market segments, produce goods and services whose appeal comes from perceived producer authenticity and custom-made products to individual clients
When do Networks Matter? A Study of Tie Formation and Decay
487
(Boone, Brocheler, & Carroll, 2000; Carroll & Swaminathan, 2000; Dobrev, 2000; Dobrev & Carroll, 2003; Dobrev et al., 2003). When a large and a small firm form a relationship, together they can access both mass markets and the customized market segments, which would result in the emergence of important synergies, such as using spare capacity of the large firm to lower the costs of manufacturing custom-made products for the small firm’s markets, or exploiting the insights from the small firm’s experience in catering to narrow market segments in order to provide some level of product differentiation within the large firm’s mass market offerings.
Partner Competitiveness Uncertainty While the partner selection literature identifies firms’ strategic desire to access and combine resources as a key rationale for forming alliances, a game theory perspective suggests that partnerships can reduce competitive uncertainty through mutual forbearance (Gimeno, 2004). The Don Michael Corleone’s maxim that one should ‘‘keep the friends close and enemies closer’’ may underlie partnering decisions. By forming ties with rivals, firms attempt to co-opt their partners thereby reducing the intensity and uncertainty of competitive interactions. This logic is drawn from the literature on multimarket contact, which similarly shows that firms competing in more than one market coordinate and signal their actions in ways that reduce competitive intensity (Jayachandran, Gimeno, & Varadarajan, 1999). Baum and Korn (1999) identify two logics in the multimarket contact literature explaining the occurrence of mutual forbearance – deterrence (Edwards, 1955) and tacit cooperation (Simmel, 1950). Deterrence strategies are more likely to emerge when firms face one another in a web of markets, because the prospect of an advantage in one market has to be weighted against the danger of retaliatory attacks by the same firm in other markets. Retaliators can counterattack in markets where their potential losses are small relative to those of the aggressors, forcing the aggressor to bear a higher cost for its initial actions. Moreover, Simmel (1950) argued that the potential for cooperation among rivals increases when they interact in multiple domains, since each will gain by allowing the other to be superordinate (dominant) in some domains in exchange for similar treatment in other domains. Multimarket contact enables firms to reduce rivalry via implicit means of signaling and learning (Korn & Rock, 2001; Stephan & Boeker, 2001), whereas partnering with potential (or actual) competitors reduces rivalry between them through more direct coordination. For example, Gimeno (2004) examined
488
ANDREW V. SHIPILOV ET AL.
relationship-building behaviors of airlines and found that ties among direct competitors (defined as the joint membership in the frequent travel programs) are more likely to be formed when their own competition is active in building such relationships. When two competitors form a relationship, they can coordinate their market entry and exit decisions as well as set prices. Within the confines of such a relationship, the partners can also explicitly agree on the distribution of markets between themselves by determining a priori the dominant and subordinate roles (Simmel, 1950) for themselves in different segments. If one partner is observed to be engaging in competitive behavior against another (e.g., entry into the other partner’s market), the management can resort to the alliance governance structures in order to solve the problem, as opposed to immediately starting the vicious cycle of retaliatory price cuts or market share wars. Given the emergence of a tie between direct competitors, they both have lower levels of competitive uncertainty as potential for the bitter competition between them has now been lowered. Firm specialization. According to the logic of competitive uncertainty reduction, the more similar the market niches that firms occupy, the more likely they are to form relationships with each other. Organizational ecologists show that niche overlap in terms of resource requirements and market specialization are important factors defining rivalry. According to Carroll (1985), organizations could be divided into two groups: specialists and generalists. Generalist firms compete in a variety of market segments and serve a wide variety of customers, whereas specialists compete within individual market segments and serve individual customer groups. For their survival, specialists and generalists require vastly different resource endowments. Generalists need access to heterogeneous resources on a large scale, whereas specialists can survive by exploiting homogeneous resources in niches that are too small for the entry of generalists. Thus, in the absence of co-optive mechanisms, the more similar firms are on the specialist–generalist dimension, the more intensively they are likely to compete. According to the logic of competitive uncertainty reduction, in order to lower the levels of competition for heterogeneous resources, generalists would be likely to form relationships with other generalists. At the same time, specialists exploiting resource niches will also be likely to form relationships with other specialists in order to lower the intensity of competition among them. In addition, relationships between firms of similar levels of specialization are likely to be stable because permanent elimination of ties between direct competitors could return them to the mode of engaging in rivalrous behaviors that, without relational governance mechanisms, could become detrimental to the performance of both firms.
When do Networks Matter? A Study of Tie Formation and Decay
489
Partner Reliability Uncertainty Partner-selection research provides evidence that firms attempt to mitigate partner reliability uncertainty when building relationships with their allies. In interfirm ties, success requires that partners are willing to cooperate – e.g., by negotiating the alliance arrangements in good faith, resolving conflicts, providing access to resources and avoiding such blatant instances of non-cooperation as a breach of the contract (Khanna, Gulati, & Nohria, 1998; Miles & Snow, 1978; Mohr & Spekman, 1994; Oxley, 1997). The partner also has to be prepared to reciprocate the contributions made by the other partner(s). Because organizations form partnerships to create competitive advantage, the cost of forming a tie with a non-cooperative partner, or worse, one engaging in learning races, can threaten the achievement of objectives or even survival (Khanna et al., 1998). Moreover, evaluating a potential partner’s reliability – willingness to engage in a cooperative, reciprocating exchange for mutual gain – requires tacit knowledge not easily ascertained prior to forming a tie (Larson, 1992). Thus, organizations seek information that can mitigate the risk of mistakenly forming a tie with a non-cooperative partner, i.e., reduce partner reliability uncertainty. Experience, third parties and networks. Interorganizational partnering patterns suggest that organizations use their past experience, their partners’ experience and network governance mechanisms to mitigate partner reliability uncertainty. When selecting partners, organizations show a propensity for their past partners and their partners’ partners (Gulati & Gargiulo, 1999). When parties to a prospective dyad share a common partner, they can utilize reliable information about each other from that partner (Baker, 1990). In this case, uncertainty about the reliability of a prospective partner is greatly reduced when a firm can turn to its own contacts and ask them about the experiences they had in relationships with the prospective ally. Similarly, organizations’ own partnering experiences provide first-hand knowledge which mitigates the risk of partnering again with their past partners. Li and Rowley (2002) find evidence that US investment banks discriminate among their past partners, preferring to renew relationships in which their partners engaged in a series of reciprocal exchanges. In addition, the surrounding network not only reduces partnering uncertainty because it is a repository of information, but also because it can serve as a governance mechanism fostering cooperation. As Gulati and Gargiulo (1999) point out, when two partners share common third-party ties, there is reputational lock-in whereby non-cooperative behavior by either partner will be efficiently revealed beyond the dyad and serve as an effective deterrent (Burt &
490
ANDREW V. SHIPILOV ET AL.
Knez, 1995). Similarly, within densely interconnected network regions – cliques – communications are efficient, and embedded organizations are able to collectively monitor and sanction non-cooperative behavior (Rowley, 1997). Prior work has shown that cliques help organizations improve performance and expose them to partnering options that exist within the clique (Baum, Shipilov, & Rowley, 2003; Provan & Sebastian, 1998). Partners are more likely to cooperate when the tie is embedded within a dense clique of common third parties, because shirking behavior will have negative reputation consequences leading to fewer partnering opportunities (Walker et al., 1997). Thus, when organizations may not be able to evaluate particular partners’ willingness to cooperate, they can, instead, rely on the network structure to govern cooperative behavior (Coleman, 1988; Li & Rowley, 2002; Rowley, Greve, Rao, Baum, & Shipilov, 2005).
Disentangling New and Renewed Ties Our review of interorganizational partnering research reveals a model emphasizing the logic of interorganizational attachment and decision criteria particular to reducing partner reliability and capability uncertainty. The model does not distinguish between tie formation and renewal/termination processes (Burt, 2000). Powell et al. (2004) analysis of the partnerships in the biotechnology industry suggests that organizations may apply different criteria depending on the stage of the tie in its life cycle, and supports our claim that the logic of initial tie formation should be disentangled from that of subsequent tie formations and terminations. The partner renewal/termination process – how organizations evaluate and choose their past or current partners for future collaborative opportunities – is distinguishable by the criteria used to evaluate partners. The initial tie formation process – how organizations evaluate and choose new partners – focuses on mitigating partner uncertainty despite the lack of first-hand information. At this stage of a relationship, organizations rely on second-hand proxies such as firm attributes, third-party endorsements, clique membership and status positions. Once a tie has been formed, however, the evaluations are made on the basis of criteria specific to the partnership. Decision criteria move from external proxies (initial tie formation decisions) to factors internal to the partners (subsequent tie renewal/termination decisions). Stinchcombe (1965) observed that collaborative agreements do not immediately lead to cooperation once a new relationship is established, but rather partners attempt to build trust and effective interactions over time. They do this by observing
When do Networks Matter? A Study of Tie Formation and Decay
491
the degree to which cooperative exchange is established in the relationship and form subsequent partnering decisions based on these experiences. Empirical research provides evidence of this shift in criteria. Larson’s (1992) study of entrepreneurial organizations shows that after forming an initial tie, partners engage in a series of incremental contributions to the relationship: each partner observes the other’s willingness to add value to the partnership, and usually responds in kind. Thus, the likelihood of tie renewal/termination is based on the tit-for-tat actions specific to the partnership (Keohane, 1986). Mohr and Spekman (1994) find evidence that organizations pay attention to the size of their partners’ investments in policies to oversee contingencies related to conflict resolution, the distribution of value and personnel turnover. While it may be intuitively obvious that firms use first-hand experience when it is available, research also suggests that it is preferred to other information sources such as network attribute information. Thus, network-based information is discounted in favor of first-hand experience, once partners have established an exchange history. Li and Rowley (2002) find that investment banks rely heavily on the information received from past ties to reconstitute their most cooperative and successful partnerships. Larson’s (1992) investigation of tie formations shows an incremental process of partners increasingly committing to the tie as a history of positive cooperation is established. Intuitively, firsthand knowledge has several advantages: the search costs are relatively low and the information gained, which is specific to the organization’s own circumstances, is not contaminated by factors unique to others’ experiences. Direct experience provides both the partners with better information about each other’s willingness to collaborate than any information transmitted by a single common partner in a simple triadic relationship or by a number of common partners within the clique. Thus, the criteria used to mitigate partner reliability uncertainty depend on whether the organization is evaluating a new partnership or an existing one. Second-hand information gleaned from partners or clique membership to deduce the partner’s reliability is replaced by first-hand accounts, once a relationship has been formed. A key implication is that once direct experience is available, the existence of a common partner or common membership in a clique is not likely to affect the longevity of the relationship. In contrast, reducing partner capability and competitiveness uncertainty does not change as dramatically from new to existing ties, because first-hand experience does not provide substantially better information than other measures. The reliability demonstrated by a partner is a characteristic specific to the relationship between the organization and that partner, and is
492
ANDREW V. SHIPILOV ET AL.
developed over time and across multiple ties (Larson, 1992; Ring & Van de Ven, 1992). However, it would not be correct to assume that factors affecting firms’ formation of relationships is completely different from those affecting firms’ decision to continue or dissolve the relationship. Although experience of working with a given partner does affect firms’ opinion of the partner’s reliability, partner capabilities cannot always be accurately judged on the basis of experience alone. The types of partner capabilities required and the intensiveness of their competition are determined by the objectives of the new collaborative opportunity and resource endowments required by partners at any given moment, regardless of past collaborative experiences. Whether two past partners have an appropriate mix of capabilities, given a new opportunity, is unclear because they might need to contribute different sets of resource endowments and capabilities or to combine them in different ways. In addition, partner capabilities and levels of specialization are likely to evolve over time as they obtain new skills, expand to different markets (thus becoming generalists), refocus their operations (thus becoming specialists) and pursue other different strategies. The quality of a partner’s products may change as well. Thus, whether a particular past partner is capable of contributing to a new collaborative venture is determined on a case-by-case (opportunity-by-opportunity) basis and second-hand information sources, such as functional characteristics (e.g., size, specialization, strategy, etc.), remain as vital inputs to the partner selection process. In summary, we expect factors mitigating partner capability uncertainty to be influential for both the formation of new ties as well as for the renewal of existing ties. The most common sources of such information are found in the functional characteristics of the partners, for example, size (Haveman, 1993) and the network – status (Podolny, 1993). Factors mitigating partner competitive uncertainty – their levels of specialization – will affect both the formation and dissolution of ties as well. In contrast, factors mitigating partner reliability uncertainty will influence tie formation to a greater extent than they would influence tie renewal/termination. These factors are usually found in the surrounding network – common third-party ties and clique membership (Gulati & Gargiulo, 1999). Keeping in mind that tie termination is the event representing the opposite of tie renewal, we propose the following hypotheses (see Table 1): Partner capability uncertainty H1. Functional dissimilarity positively influences tie formation and negatively influences the likelihood of tie termination.
493
When do Networks Matter? A Study of Tie Formation and Decay
Table 1. Hypotheses
Hypothesis Hypothesis Hypothesis Hypothesis Hypothesis
1 2 3 4 5
Summary of Hypothesized Relationships and Results. Variable
Tie Formation
Size difference Status difference Specialization difference Third-party tie Position in a clique
Tie Decay
Hypothesized Effect
Actual Effect
Hypothesized Effect
Actual Effect
+ + +
+ + +
+ + NS NS
+ + NS NS
H2. Status similarity positively influences tie formation and negatively influences the likelihood of tie termination. Partner competitiveness uncertainty H3. Similarity in specialization of firms positively influences tie formation and negatively influences the likelihood of tie termination. Partner reliability uncertainty H4. Common third-party ties positively influence tie formation and have a weaker positive or no effect on tie termination. H5. Common network clique membership positively influences tie formation and has a weaker positive or no effect on tie termination.
RESEARCH SETTING AND DATA Industry Background We gathered data on the population of Canadian investment banks forming syndicates to underwrite public offerings between 1952 and 1989. Numerous studies provide rich accounts on the functioning of IPO markets and investment banks (e.g., Eccles & Crane, 1988; Ellis, Michaely, & O’Hara, 2000; Podolny, 1993; Pollock, Porac, & Wade, 2004; Shipilov, 2005 2006), thus here we provide only a brief overview. Investment banks act as financial intermediaries linking issuers (organizations) wishing to raise funds on capital markets to investors. They add value to corporations raising capital in primary markets by effectively pricing and placing their issues. The industry is characterized as ‘‘relationship oriented’’ because banks commonly collaborate
494
ANDREW V. SHIPILOV ET AL.
in underwriting deals. Relationships are not only common practice but also a vital resource, as ties are conduits to underwriting opportunities, and they contribute to banks’ reputations and performance (Podolny, 1993). Banks form syndicates to facilitate the placement of new securities and to reduce financial risks to individual underwriters. Banks organizing syndicates are called lead managers and usually contribute up to 50% of the capital required to fund the issue (Ellis et al., 2000). The rest of the capital is contributed by other syndicate members (co-lead managers) whose participation is important for reducing risks to individual banks and for broadening the distribution of shares to different investor groups (Pollock et al., 2004). One needs to distinguish, however, between relationships and syndication ties in this particular setting. Syndication ties are short-term arrangements between investment banks when they pool their resources to underwrite a single public offering. By definition, such ties end when the syndication process is complete and the underwriting syndicate has disbanded. Relationships between investment banks, however, comprise a series of banks’ participation in multiple underwriting syndicates. A relationship between two banks can be said to exist when these banks have participated in one or more underwriting syndicates. Furthermore, an underwriting relationship between two banks will end if these banks no longer jointly participate in underwriting syndicates. Even though the Canadian investment banking industry is extensively scrutinized by both outsiders (media, government, investor groups) and by its participants through the Canadian Bankers Association, Investment Dealers Association of Canada or the Canadian Capital Markets Association, banks participating in the underwriting syndicates can still behave non-cooperatively toward one another as there are a lot of gray areas in the conduct of investment business. For example, lead managers invite other banks to join their syndicates, hoping that invitees would reciprocate invitations in the future and allow the former to participate in their own syndicates as co-lead managers. Some banks’ non-cooperation could be manifested in their failure to reciprocate past invitations or in the repeat instances of inviting their partners to participate in those syndicates that are a priori likely to generate lower investor interest and lower profits for all of the underwriters involved. Another instance of non-cooperation would be the syndicate members’ underutilization of their own contacts to place the newly issued securities. Every bank is linked to a number of different investor groups (individual investors or institutional investors such as pension funds) whose acquisition of new shares can contribute to the increase in the
When do Networks Matter? A Study of Tie Formation and Decay
495
trading activity, ultimately leading to an improvement in the issue’s performance. If a member of a syndicate is not actively pushing the issue to its various investor contacts, levels of issue trading may not be as high as expected and profits of all syndicate members may suffer from this noncooperative behavior. The final source of non-cooperation could be manifested in banks’ aggressive attempts to hire star investment bankers away from their current places of employment. When several banks work together on preparing new issues, the bankers from different firms get to know one another’s skills and capabilities. A non-cooperative bank could utilize this knowledge in an attempt to improve its own competitive position by offering significantly better compensation packages to the best professionals working in other banks, ultimately damaging competitive position of its syndication partners. Such instances of non-cooperative activities are perfectly legal, however, they detrimentally influence operations of affected parties and can be avoided by the network-based mechanisms reducing partner reliability uncertainty. Some of our theoretical arguments outlined above deal with the emergence of synergies when firms of extremely different sizes form relationships. In the IPO syndication side of the investment banking industry, such synergies may not be obvious as banks contribute money to finance public offerings, and money is one of the most homogeneous resources. However, as Pollock et al. (2004) point out, underwriters bring much more than cash to the issue, that is, they contribute investor contacts, distribution capabilities, industry-specific skills, syndication management skills and so on. Larger banks tend to have contacts with different investor groups, whereas small banks tend to have contacts with select groups of investors, whose participation in the issues nonetheless could be of great importance. Indeed, the higher the diversity of investors buying the newly issued securities, the better is the issue performance, thus, large banks would often need the assistance of smaller underwriters to sell the issue to the select groups of investors. In addition, smaller banks could also be better able to serve smaller clients whose small-cap issues may not be of interest to larger underwriters, although the expertise of small-cap investing of its partners could prove to be very valuable for a large underwriter. In return for their contacts and expertise, small banks may receive preferential treatment from the large lead managers and, despite their small asset base, be allowed into the syndication of potentially very lucrative issues that would ordinarily require commitments of large amounts of money from the underwriters. Finally, the propensity of Canadian investment banks to form relationships based on functional dissimilarity can be predicted from the organizational
496
ANDREW V. SHIPILOV ET AL.
learning perspective, originally applied by Rowley, Behrens, and Krackhardt (2000) to understanding the outcomes for firms’ partnering decisions. Industry volatility has a strong effect on the type of relationships from which firms will benefit. For example, in rapidly changing industries, firms benefit from exploratory learning which is manifested in building weak relationships to heterogeneous (dissimilar) partners. Such ties allow firms to maintain strategic flexibility and react to fleeting opportunities. In the more stable industries, firms benefit from exploitative learning by maintaining network ties to a homogeneous group of partners. When firms build relationships with functionally dissimilar allies, this implies the reliance on the exploratory networking behaviors as it leads to the construction of networks containing heterogeneous partners. As Shipilov (2005) demonstrated, between 1952 and 1989 Canadian investment banking suffered a series of shocks: the oil shock in the early 1970s (1970– 1975), the recession of 1982–1983 which was caused by fiscal restraint from the government and the 1987 global financial crisis. These disruptions increased uncertainty among investment banks and created an environment where one would expect exploratory logic with respect to functional specialization to dominate banks’ relationship-building decisions.
Network Operationalization We constructed interorganizational networks for each observation year based on banks’ underwriting syndicate memberships. These networks were constructed from adjacency matrices capturing the number of times each bank participated in a syndicate with each other bank for four-year moving periods (i.e., 1952–1955, 1953–1956, 1954–1957, etc.). For the purposes of our analysis, the matrices were symmetric, that is, if there was a tie from bank i to bank j, then there was a tie from bank j to bank i. We used four-year windows for four reasons. First, syndicate ties represent only the visible manifestation of relationships; banks participating in syndicates together in any given year are also likely to interact in other ways with one another in years proximate to the syndicate. Second, because syndicates can remain intact up to six months or more prior to the date of the offering, those that conclude in any given year may have been formed in prior years. Third, the four-year window permits us to gauge more accurately the strength of network ties by incorporating information on repeated ties over a number of years. Constructing the network based on four-year moving periods should thus permit us to represent the network more reliably and accurately. Finally, prior studies (Baum et al., 2005; Rowley et al., 2005) that looked at the dynamics within the Canadian
When do Networks Matter? A Study of Tie Formation and Decay
497
investment banking industry have demonstrated that analyses conducted using the four-year windows yielded similar results to those using alternative specifications (such as three- or five-year windows).
Modeling the Formation and Dissolution of Relationships Our sample covers the entire population of Canadian investment banks and all ties that were formed between 1952 and 1989. Using information on relationships between investment banks, we created a data panel representing a life history of individual dyads. In other words, we identified each and every relationship that existed between two banks and traced the history of this relationship within an observation window. Specifically, we recorded the formation of a relationship between two banks when they participated in the same underwriting syndicate for the first time. The relationship was considered ongoing if these banks together continued to underwrite public offerings in the future. One major complication that we encountered was identifying the timing of a relationship’s dissolution. Sometimes, a tie between two banks would disappear in one period and then reappear a few periods later. Making sure that we correctly coded the year when a relationship dissolved, never to reappear, was particularly crucial toward the end of our observation window. For example, if two banks ceased to cooperate in 1987, does this mean that their relationship deteriorated forever, or that this relationship would reappear in 1991? While our data did not allow us to observe partnering behavior after the end of the observation window (i.e., in 1990), we were still able to identify a time period beyond which ties were not likely to be restored. Specifically, based on the data between 1952 and 1989, we created a distribution of the number of restored ties as a function of the time elapsed from initial tie dissolution. As our calculations have shown, if a tie was not restored within 10 years after its initial dissolution, there was only a 5% chance that it would be restored later. Thus, we limited our analysis to dyads occurring between 1952 and 1980. If a tie decayed prior to 1980, and we did not observe this tie’s restoration until 1990, there was less than a 5% probability that it would reemerge beyond our observation window.
Theoretical Variables Our theoretical variables were computed on a dyadic level. Specifically, we computed variables for the differences in banks’ size (Hypothesis 1), and
498
ANDREW V. SHIPILOV ET AL.
status (Hypothesis 2) to test hypotheses related to value generation uncertainty. To test Hypothesis 3 about banks’ competitive uncertainty, we computed a variable that captured differences in banks’ levels of specialization. Finally, to test Hypotheses 4 and 5, partner reliability uncertainty, we computed variables capturing the existence of a common third-party partner for the two banks as well as a variable capturing both banks’ membership of the same clique. Functional dissimilarity. We used organizational size to capture functional dissimilarity among underwriting partners because the capabilities are quite different for large and small banks. Larger banks provide substantial capital and legitimacy, which attracts issuers and a broad range of investors. Smaller banks contribute niche specialization useful for accurately pricing IPO offerings, attracting investors favoring that sector and, overall, reducing underwriting risk. In addition, organizational size provides a gauge for the degree of competition between partners. Small and large banks compete in substantially different segments. Because small banks solicit IPO deals that large banks deem too small, they are unlikely to bid on the same offerings. Given this differentiation, smaller banks could refer parts of their smaller deals to the larger underwriters without fearing their larger partner attempting to secure future business from the issuer. We captured differences in banks’ size by computing the differences in their market shares. Operationally, we measured a bank’s market share for a given year based on the inflation-adjusted dollar value of the syndicates in which it participated in that year. We allocated 50% of the underwriting value to the lead bank, and split the remaining value among the co-lead banks equally.1 Then, to compute size difference variable, we calculated the absolute difference between logged market shares of banks in a dyad. We used this more complicated proxy for functional difference because traditional means of assessing banks’ size, such as their asset values or the number of people employed, were not available to us. Between 1952 and 1989, the vast majority of banks in our sample were privately held, thus the assets/employment information was not publicly available. Status similarity. Status homophily is particularly profound in the investment banking industry. Quality of investment banking services provided by an underwriter is difficult to determine a priori, thus banks send status signals to their partners as well as to outsiders. When quality cannot be directly observed, an increasing number of ties to higher status underwriters enhance the status of the focal bank, whereas ties to lower status underwriters distract from the bank’s own status.
When do Networks Matter? A Study of Tie Formation and Decay
499
We measured a bank’s status using Bonacich’s (1987) eigenvector centrality. While this approach, which relies on symmetric syndicate ties, is potentially less precise than one based on asymmetric tombstone rankings (see Podolny, 1993), it is consistent with past network-based measurement of status in investment and commercial banking (Jensen, 2003; Li & Rowley, 2002). Since this measure was computed based on symmetric ties, it was not affected by the differences in lead and co-lead management roles that the banks were taking throughout each individual time period. We computed each bank’s centrality for each four-year network, and assigned the value to the final year of each four-year period. Thus, the 1952–1955 network was used to measure banks’ centrality for 1955, the 1953–1956 network for positions in 1956 and so forth. We normalized the measure, dividing it by the maximum difference in each four-year network, to facilitate comparisons over time (Borgatti, Everett, & Freeman, 1999). Given this specification, an investment bank’s network status is a function of the number and the status of the banks with which it forms underwriting syndicates. In turn, the status of these partners is a function of the number and the status of their syndicate partners, and so on. Status difference has then computed in two steps. First, we calculated absolute differences between normalized eigenvector status scores of both parties to a dyad. Second, this difference was divided by the sum of the statuses for both parties. The second step was necessary as bank raw status scores had extremely high variance, with the lowest score of zero and the maximum score of 97. By dividing status differences of dyad members by the sum of status scores, we standardized this variable based on the statuses of parties to a dyad. Specialization difference. Differences in banks’ specialization could affect their propensity to form ties (Hypothesis 3). Following Shipilov (2006), we operationalized a bank’s specialization based on the range of industries in which it underwrote its public offerings. This range was identified based on a bank’s activities in seven economic sectors: manufacturing, non-manufacturing, natural resources, utilities, financial, technology and government. We then defined a bank’s specialization using a variant of the Herfindahl index: Sector specializationit ¼
j¼7 X Sjt 2 j¼1
kit
(1)
where kit is the total number of syndicates in which bank i participated in period t, Sjt the number of syndicates in which bank i participated in period t, that were in sector j. If the value of this variable is one (its maximum), this
500
ANDREW V. SHIPILOV ET AL.
indicates that a bank’s syndicates were all in the same single sector; lower values indicate that a bank participated in syndicates more broadly distributed across the various sectors. The specialization difference variable was then computed by calculating the absolute value of the difference between the specializations of two parties to a dyad. Third-party ties. To test Hypothesis 4, we created a dummy variable, third-party tie, which was set to 1 when banks in a dyad shared the same common partner(s) and set to 0 when banks in a dyad did not share the same partner. Clique membership. To test Hypothesis 5, we identified interbank cliques using the approach reported in Rowley et al. (Rowley, Baum, Shipilov, Rao, & Greve, 2004). While the concept of cliques in networks is widely accepted, a precise, formal definition of operationalizing it was not available to us a priori. Because cliques are relatively dense or cohesive substructures and cohesiveness demands that organizations are close to one another, many researchers use social distance, or reachability, to identify cliques (Wasserman & Faust, 1994). The social distance separating two organizations in a network is the count of the number of intermediaries (other organizations) along the shortest path connecting them – e.g., two organizations are two steps away from one another if they are not connected to each other but have a tie to the same third party. The most restrictive clique structures require that all members of the subgroup are tied to every other member such that the maximal social distance between all pairs is one. The core question for identifying cliques is how close organizations in a particular setting must be to one another to constitute a cohesive group. We identified cliques using the N-clan procedure implemented in UCINet 5.0 (Borgatti et al., 1999). An N-clan is a clique in which the geodesic distance between all nodes is no greater than N for paths within the subgraph. We set N ¼ 2 such that all members of a clique are within two steps of every other member. Identifying cliques in which the maximal social distance among clique members is two steps is appropriate in our empirical setting for two reasons. First, research has shown that the likelihood of an investment bank forming a tie with another bank is relatively high for its past partners and its partners’ partners, but declines significantly when moving beyond a social distance of two steps (Chung et al., 2000; Li & Rowley, 2002). These findings suggest that banks deem their partners’ partners to be close enough to overcome the uncertainty of potential allies’ reliability and capabilities. Thus, the partner selection patterns in the industry should tend to produce networks in which banks are self-organized in two-step cliques.
When do Networks Matter? A Study of Tie Formation and Decay
501
Using the cliques generated by the N-clan routine for each four-year network, we constructed a genealogy of the cliques, linking them over time. A clique was considered ongoing from one four-year period to the next if it retained more than 50% of its membership from the previous period. Thus, if a clique identified for the 1952–1955 network shared more than 50% of its members with a clique identified for the 1953–1956 network, we considered the 1953–1956 clique a continuation of the 1952–1955 clique. Using this procedure, we identified 25 distinct cliques, with an average duration of 5 years (SD ¼ 5.5; range 1–20). The cliques averaged 21.8 members (SD ¼ 12.5; range 4–51), of which, on average, over 70% remained the same from one year to the next. Two hundred ninety nine banks were assigned to cliques over the course of the observation period. The average density of syndicate ties within these groups was 0.27 (SD ¼ 0.10), while the average density of crossclique ties was 0.006 (SD ¼ 0.001), meaning the cliques are cohesive and close, which is consistent with the theoretical concept of cliques. Eighty-two percent of the sample banks were assigned to one clique in each year; however, 13% were assigned to two cliques and 4% to three or more (maximum five in less than 1% of cases). To capture two banks’ membership of the same clique(s), we computed a variable, clique overlap, which was equal to the number of cliques in which two banks belonged at the same time.
Control Variables Many other factors could influence stability of ties between the two banks. Thus, we included a range of control variables in our models. These controls included specifications for the characteristics of a dyad (strength of a relationship, duration of a relationship), industry network characteristics (number of nodes in a network, network clustering coefficient), industry characteristics (value of all public offerings, number of public offerings, number of syndicates formed among investment banks, value of TSE 300 index, a Canadian stock market indicator equivalent to S&P 500 in the US) and a variable capturing the passage of time. Among other dyad-level variables, we computed tie strength by counting the number of syndicates in which both banks jointly participated at any given period. Tie duration was set equal to the number of time periods during which there was a tie between bank i and bank j. If a tie between these two banks disappeared for some time, but was later restored, the count of tie duration was continued. For example, if a tie existed in 1952–1956, 1953–1957 then disappeared until 1956–1959, then variable tie duration in
502
ANDREW V. SHIPILOV ET AL.
1956–1959 would be set to three. To capture possible non-linearity for the relationship between tie duration and the likelihood of the relationship to decay, our models contained both linear and quadratic specification of tie duration. Changes in the total number of firms in a network could potentially affect partnering choices for individual firms. Thus, we calculated a control variable, number of nodes, which captured a number of banks in a given period of time that were active in underwriting public offerings in Canada. Density of ties in the industry could also affect a firm’s partnering decisions with greater average density contributing to industry governance mechanisms that increase firms’ propensity to form relationships and decrease the likelihood for these relationships to decay. To control this effect, following Baum et al. (2003), for each bank we first computed its clustering coefficient as a ratio of the number of actual ties between the bank’s partners over the number of possible ties between the bank’s partners. The denominator in that ratio was defined as N*(N1) where N was a number of bank’s partners. Once the clustering coefficient was computed for each bank, we then calculated the average clustering coefficient for the entire network by summing all banks’ clustering coefficients and dividing this sum by the total number of banks in a network. We controlled for several aspects of industry dynamics to hold constant the munificence of the investment environment. We computed a variable, value of all public offerings, which captured dollar value of all issues in Canada within the current observation period. A large number of syndicates formed between investment banks creates increased partnering opportunities, thus we created a variable, number of syndicates. Finally, the general health of the financial markets could affect the propensity of banks to engage in public offerings syndication deals, so we included a variable equal to the value of TSE 300 index. Finally, we controlled for the passage of time by using a counter, time, starting at one at the beginning of our observation period (1952–1956) and then increasing by a value of one for every subsequent period.
Estimation We tested our hypotheses using a two-stage regression approach. In the first stage, we predicted the likelihood for the formation of a relationship between two firms as a function of differences in their size, status, existence of a common third-party partner, both banks’ membership in the same cliques as well as a set of control variables. Based on the coefficients for theoretical and
When do Networks Matter? A Study of Tie Formation and Decay
503
control variables, we then constructed an instrument predicting the likelihood for the formation of relationships between the two firms. This instrument was then included as a type of Heckman-correction variable in the second stage of our analysis, where we modeled the impact of theoretical and control variables on the likelihood of tie dissolution between the two firms. In the first stage of our analysis, we obtained estimates for factors affecting the formation of relationships by comparing properties of realized dyads with a random sample of non-realized dyads (Jensen, 2003). Specifically, a random sample of 3,790 non-realized dyads was compared to 758 realized dyads that were actually formed (one realized dyad to five non-realized dyads), which resulted in the total of 4,548 dyads in our dataset. For 177 dyad years (3.8% of all observations), we were unable to calculate the specialization variable due to the missing data. Other than with the specialization variable, however, we did not face any further complications with missing information. Taking into account missing specialization observations, the complete dataset consisted of 4,371 dyad years. In the second stage, we compared all decayed dyads with all ongoing dyads between 1952 and 1980 with a combined sample consisting of 228 decayed relationships (that did not subsequently reappear after 1980) and 3,838 dyad years for the ongoing relationships, which resulted in the total sample of 4,066 dyad years. The first-stage regression analysis was conducted by fitting a logistic regression model (logit command in STATA), whereas the second-stage analysis was conducted by fitting a random effects time series logistic regression model (xtlogit command in STATA).
Results Table 2a reports means, standard deviations and correlations between variables in the dataset of realized and potential dyads, which is used to predict the probability of relationship formation between the firms. Table 2b reports descriptive statistics and correlations between variables in a dataset consisting of continued and terminated relationships, which is used to predict the probability of decay in the relationship between the firms. Correlations between theoretical variables are generally small in magnitude, which suggests that multicollinearity is not a cause for concern in our data. Table 3a contains hierarchically nested models estimating the impact of theoretical and control variables on the likelihood of tie formation between investment banks. Model 1 is a baseline containing control variables. In Model 2 we entered size difference, in Model 3 we entered status difference, Model 4 enters
504
Table 2a.
Descriptive Statistics and Correlations in the Analysis of Tie Formationa. Mean
a
Min
Max
1
2
Formed Tie ¼ 1 0.17 0.37 0.00 1.00 1.00 Size difference 0.02 0.04 0.00 0.34 0.22 1.00 Status difference 0.79 0.29 0.00 1.00 0.11 0.08 Specialization difference 0.33 0.26 0.00 0.80 0.03 0.18 Clique overlap 0.18 0.41 0.00 3.00 0.28 0.11 Third-party tie 0.26 0.44 0.00 1.00 0.33 0.16 IPO market size 1.02E+11 1.79E+11 3.44E+09 7.61E+11 0.00 0.11 Number of syndicates 114.52 113.34 20.00 444.00 0.00 0.09 Average clustering 0.20 0.08 0.03 0.37 0.01 0.03 Number of banks 96.98 24.24 34.00 140.00 0.00 0.15 TSE 300 1679.31 894.95 122.36 3348.85 0.00 0.13 Period ID 76.47 10.12 56.00 88.00 0.00 0.10
N ¼ 4371.
3
4
5
6
7
8
1.00 0.13 0.31 0.28 0.02 0.00 0.04 0.02 0.01 0.01
1.00 0.04 0.01 0.02 0.03 0.04 0.02 0.03 0.03
1.00 0.64 0.02 0.02 0.06 0.02 0.02 0.01
1.00 0.07 0.08 0.14 0.04 0.08 0.09
1.00 0.78 0.23 0.70 0.70 0.49
1.00 0.27 0.61 0.77 0.57
9
10
11
12
1.00 0.23 1.00 0.41 0.85 1.00 0.70 0.69 0.88 1.00
ANDREW V. SHIPILOV ET AL.
1 2 3 4 5 6 7 8 9 10 11 12
S.D.
Mean 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 b
Descriptive Statistics and Correlations in the Analysis of Tie Decayb. S.D.
Min
Max
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Decayed tie ¼ 1 0.05 0.22 0.00 1.00 1.00 Size difference 0.05 0.06 0.00 0.34 0.03 1.00 Status difference 0.68 0.30 0.00 1.00 0.04 0.18 1.00 Specialization difference 0.28 0.26 0.00 0.79 0.07 0.08 0.40 1.00 Clique overlap 0.85 0.97 0.00 5.00 0.02 0.18 0.45 0.27 1.00 IPO market size 1.19E+10 1.23E+10 3.44E+09 4.71E+10 0.06 0.02 0.01 0.09 0.07 1.00 Number of syndicates 53.91 23.40 20.00 113.00 0.04 0.07 0.03 0.09 0.09 0.50 1.00 Average clustering 0.17 0.10 0.03 0.37 0.08 0.07 0.07 0.07 0.08 0.46 0.42 1.00 Number of banks 81.37 17.00 34.00 109.00 0.04 0.15 0.02 0.11 0.03 0.57 0.59 0.01 1.00 Third-party tie 0.59 0.49 0.00 1.00 0.04 0.18 0.43 0.31 0.51 0.16 0.11 0.27 0.03 1.00 Duration of a tie 6.05 5.12 1.00 25.00 0.09 0.13 0.13 0.11 0.10 0.04 0.04 0.21 0.10 0.24 1.00 Duration of a tie2 62.81 104.04 1.00 625.00 0.10 0.12 0.11 0.10 0.07 0.10 0.00 0.25 0.08 0.22 0.95 1.00 Tie strength 1.68 2.95 0.00 38.00 0.02 0.14 0.34 0.15 0.27 0.09 0.07 0.05 0.05 0.15 0.17 0.25 1.00 TSE 300 950.32 385.50 122.36 2027.70 0.03 0.02 0.04 0.09 0.02 0.78 0.37 0.42 0.56 0.19 0.16 0.17 0.04 1.00 Period ID 69.30 7.20 56.00 80.00 0.04 0.05 0.08 0.08 0.01 0.63 0.40 0.85 0.22 0.29 0.29 0.30 0.04 0.73 1.00
N ¼ 4066.
When do Networks Matter? A Study of Tie Formation and Decay
Table 2b.
505
506
ANDREW V. SHIPILOV ET AL.
specialization difference, Model 5 enters a third-party tie variable and finally Model 6 enters a clique overlap variable. Each subsequent model represents a statistically significant improvement as compared to both the previous model and to the baseline. Results in full Model 6 indicate that greater difference in size between investment banks positively affects the likelihood of tie formation between them (po0.001), differences in status have a significant negative influence on the propensity of banks to form a tie (po0.1), similarity in banks’ levels of specialization has a positive effect on their propensity to form ties (po0.01), existence of third-party ties has a positive effect on the likelihood of tie formation (po0.001), and finally, both banks’ membership in the same clique increases the likelihood of these banks forming a relationship (po0.001). Table 3b contains hierarchically nested models estimating the impact of theoretical and control variables on the likelihood of tie decay between investment banks. Model 1 is a baseline containing control variables. In Model 2 we entered size difference, in Model 3 we entered status difference, Model 4 contains specialization difference, Model 5 contains third-party tie, and finally, Model 6 contains a clique overlap variable. In the full Model 6, third-party tie and clique overlap are not significant differences in size and in status point in the hypothesized direction. Specifically, the likelihood of decay of a relationship between two firms decreases with an increase in size difference between them (po0.01). Greater status difference (po0.05) increases the likelihood of decay of ties between firms. Finally, greater specialization difference has a positive impact on the likelihood of tie decay (po0.01). Coefficient estimates of some control variables in this table are also worth noting. A positive coefficient for a linear specification and a negative coefficient for a quadratic specification for duration of a tie (po0.001) suggest the existence of a U-shaped relationship between tie duration and tie decay. Very young and very old ties are less likely to decay as compared to ties of an average duration. Once the tie is being formed, the goodwill that exists between the parties enables this tie to survive, but only up to a certain point. If firms do not realize how to extract long-term benefits from their collaboration, their tie is likely to decay when it has reached a level of ‘‘adolescence,’’ but once the tie has survived the ‘‘adolescent’’ threshold, it is likely to be more stable in the future. Perhaps surprisingly, we found that stronger ties (i.e., relationships between investment banks that involve multiple deals in a single year) have a high probability of decay. Moreover, an increase in the size of the market for public offerings in Canada leads to an increase in the probability of relationships to decay. This effect could be attributed to
Variable
Model 1
Estimates for Tie Formation. Model 2
Model 3
Model 4
—
Status difference
—
11.96 (0.967) —
Specialization difference
—
—
12.90 (1.02) 1.16 (0.132) —
Third-party tie
—
—
—
Clique overlap
—
—
—
—
PO market size
8.71 1026 (4.35 1013) 5.84 1016 (0.0007) 1.07 1012 (1.00) 1.79 1015 (0.0035) 3.37 1016 (0.0002) 2.72 1014 (0.018) 1.61+ (0.95)
4.13 1016 (4.35 1013) 0.00015 (0.0007) 0.312 (1.03) 0.0032 (0.004) 0.00008 (0.0002) 0.008 (0.018) 1.74 (0.98)
7.59 1014 (4.38 1013) 0.00002 (0.0008) 0.321 (1.04) 0.004 (0.004) 0.000011 (0.0002) 0.0002 (0.019) 1.33 (1.00)
8.30 1014 (4.39 1013) 0.00005 (0.0008) 0.743 (1.05) 0.005 (0.0039) 0.00003 (0.0002) 0.005 (0.018) 1.45 (1.01)
1.21 1013 (4.55 1013) 0.0002 (0.0008) 3.35 (1.13) 0.004 (0.0038) 0.0002 (0.0002) 0.03 (0.019) 3.50 (1.07)
10.16 (1.000) 0.245+ (0.151) 0.555 (0.173) 1.367 (0.114) 0.451 (0.108) 1.68 1013 (4.56 1013) 0.0002 (0.0008) 3.26 (1.13) 0.003 (0.0038) 0.0002 (0.0002) 0.028 (0.019) 3.57 (1.07)
2049 NA
1961 88
1897 64
1848 49
1693 155
1684.7 8.7
Average clustering Number of banks TSE 300 Period ID Constant Chi square D Chi square with previous model
507
po0.01.
10.27 (1.000) 0.372 (0.151) 0.585 (0.173) 1.631 (0.094) —
Model 6
Size difference
Number of syndicates
13.29 (1.04) 1.078 (0.134) 0.538 (0.16) —
Model 5
When do Networks Matter? A Study of Tie Formation and Decay
Table 3a.
508
Table 3b. Variable Size difference Status difference Specialization difference
Clique overlap Duration of a tie Duration of a tie2 Tie strength PO market size Number of syndicates
Model 2
Model 3
1.70 (1.29)
3.09 (1.40) 0.527+ (0.309) 1.04 (0.28) —
Model 5 3.45 (1.45) 0.656 (0.343) 1.09 (0.292) 0.184 (0.180) —
Model 6
3.45 (1.45) 0.656 — — (0.343) — — 1.09 (0.292) — — — 0.215 (0.195) — — — — 0.041 (0.101) 0.671 0.658 0.676 0.674 0.672 0.656 (0.108) (0.109) (0.11) (0.11) (0.11) (0.11) 0.064 0.064 0.063 0.065 0.064 0.064 (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) 0.053 0.083 0.071 0.073 0.074 0.049 (0.021) (0.021) (0.020) (0.023) (0.023) (0.023) 5.04 1011 5.18 1011 4.81 1011 5.09 1011 5.03 1011 5.03 1011 (9.47 1012) (9.51 1012) (9.96 1012) (9.96 1012) (1.00 1011) (1.00 1011) 0.022 0.021 0.022 0.023 0.023 0.023 (0.006) (0.006) (0.006) (0.006) (0.006) (0.006) —
2.74 (1.37) 0.876 (0.295) —
Model 4
ANDREW V. SHIPILOV ET AL.
Third-party tie
Model 1
Estimates for Tie Decay.
Number of banks TSE 300 Period ID Probability of tie to form Constant Chi square D Chi square with previous model po0.05. po0.01.
9.19 (2.18) 0.011 (0.008) 0.0006 (0.0005) 0.068+ (0.034) 0.104 (0.322) 7.14 (1.84) 789.52 NA
9.21 (2.19) 0.0102 (0.008) 0.0006 (0.0005) 0.067+ (0.034) 0.181 (0.326) 7.001 (1.83) 788.61 0.91 NS
9.05 (2.21) 0.009 (0.008) 0.0006 (0.0005) 0.074 (0.035) 0.34 (0.344) 8.03 (1.88) 758.4 30.21
8.74 (2.22) 0.009 (0.008) 0.0005 (0.0005) 0.070 (0.035) 0.391 (0.35) 7.93 (1.89) 743.8 15
8.83 (2.22) 0.010 (0.008) 0.0005 (0.0005) 0.071 (0.035) 0.342 (0.35) 8.06 (1.89) 743.2 0.6 NS
8.95 (2.22) 0.010 (0.008) 0.0005 (0.0005) 0.071 (0.035) 0.342 (0.35) 8.07 (1.89) 743.2 0 NS
When do Networks Matter? A Study of Tie Formation and Decay
Average clustering
509
510
ANDREW V. SHIPILOV ET AL.
the possibility that munificent environments create multiple underwriting opportunities for investment banks. In order to take advantage of these new opportunities, banks need to form ties with new partners. Since a bank can participate in a limited number of syndicates in any given year, in order to form relationships with new partners, it has to abrogate relationships with the old ones. Conversely, when markets for public offerings shrink, investment banks stick to collaborating with their past partners, as they become uncertain about their ability to secure deals involving underwriters they have never collaborated with in the past. Finally, there is a negative association between the ties’ propensity to decay, average clustering and number of syndicates at the industry level. As the density of relationships between firms increases, the propensity of ties to decay decreases, suggesting the existence of an industry network-level governance mechanisms that solidifies relationships between investment banks. In Tables 3a and 3b, coefficients of theoretical variables provide evidence in support of Hypotheses 1–5. Specifically, we find that differences in size and similarities in status increase the likelihood of the formation of relationships between the firms and decrease the likelihood of decay in these relationships. Similarity in firms’ competitive positions, as reflected in the levels of their specialization, increases the likelihood of tie formation and reduces the likelihood of tie decay. In contrast, the existence of a common third-party tie and the firms’ membership of cliques have a positive effect on the formation of ties between them, but have no effect on the longevity of these relationships. A closer look at the descriptive statistics in Tables 2a and 2b prompted us to conduct an additional sensitivity analysis. For example, the mean number of banks in the industry equals about 97 (SD ¼ 24) in the analysis of tie formation, but in the analysis of tie decay, this number equals about 81 (SD ¼ 17). Such differences, attributable to different time windows used in the analyses (1952–1989 for tie formation and 1952–1980 for tie decay), suggest that the number of firms active in the industry in any given period varies. As Shipilov (2005) points out, between 1952 and 1990, the Canadian investment banking industry witnessed a large number of entries and exits, which could be attributable, in part, to the fluctuations in the demand for public offerings. To see whether industry-level entry–exit dynamics materially affected our results, we conducted a supplementary analysis in which we entered the counts of new entrants and firms that exited the industry in every period of time into our analysis of tie formation (Table 3a) and tie decay (Table 3b). However, the addition of these two variables did not materially alter the results of our analyses.
When do Networks Matter? A Study of Tie Formation and Decay
511
In a final supplementary analysis, we checked for the impact of new firms’ entrance into the market on the propensity of tie formation between these firms and the established market participants. A priori, we expected to find that a dyad including a new entrant should have a lower probability of forming than a dyad that includes firms with some experience of competing in the market. This finding would be consistent with our uncertainty arguments, suggesting that every new entrant will suffer from reliability, competitiveness and capability uncertainties, simultaneously. After all, when a firm has just entered the market, the incumbents will need time to assess whether there is a fit between the capabilities of the new entrant and those of their own; whether the entrant is likely to engage in competitive behavior in their own market and whether the entrant is a reliable ally to work with. To test for this assertion, we included a dummy variable new entrant to denote a dyad member that has entered into the market for the first time in a given period. In the analysis of the determinants for tie formation, this dummy was negative, suggesting that dyads were less likely to form when at least one prospective partner was a new entrant.
DISCUSSION AND CONCLUSIONS This study is motivated by a contradiction separating interorganizational partner selection theory and empirical observation. The theory suggests that organizations show a propensity for forming ties with their past partners (Gulati, 1995), their partners’ partners (Uzzi, 1996) and high status others (Podolny, 1993), all of which tend to solidify the existing network structure (Li & Rowley, 2002; Rowley et al., 2005; Walker et al., 1997). Empirical analysis, however, shows that interorganizational networks tend to change over time (Burt, 2002; Madhavan, Koka, & Prescott, 1998; Rowley & Baum, 2004). To reconcile these differences, we differentiated ties that marked the beginning of a new relationship – no previous ties between the partners – from ties representing the continuation (rejuvenation) of an existing (previous) relationship. Our argument is that network factors mitigating partner reliability uncertainty and leading to network stability carry less weight in the decision process once the relationship has been established and partners have access to first-hand knowledge. These arguments are applicable to the evolution of horizontal networks, for example, those formed as a result of investment banks’ participation in IPO syndication teams, firms’ joint efforts for the establishment of R&D consortia or joint ventures aimed at entering foreign markets.
512
ANDREW V. SHIPILOV ET AL.
Specific to the investment industry, banks expose themselves to risks when they invite unfamiliar partners to join their syndicates. For example, it is common practice for co-lead managers to invite their past partners (lead managers) to take part in their own syndicates, thus providing the latter with additional underwriting fees. A lead manager partnering with unfamiliar colead will thus face uncertainty about the willingness of the co-lead to reciprocate its offer. When both banks belong to the same clique, or at least share a single common partner, information about each other’s past reciprocal behaviors is distributed across clique members, which enables both parties to lower the uncertainty about each other. While such acts of non-cooperation are not very frequent, our interviews with bankers have shown that they do take place and can affect the bottom line of the underwriters. The results, which compare the influence of partner and network attributes across tie formation and termination events, support our general claim. Our hypothesis that size differences between partners increases the likelihood of two banks forming a tie and decreases the likelihood that they will terminate (Hypothesis 1) is supported. This finding suggests that partner attributes indicating whether a potential or existing partner possesses complementary skills/capabilities are important for initial and subsequent tie formation decisions. We find that status difference (Hypothesis 2), another factor used to mitigate partner capability uncertainty, is significant in both the tie formation and decay models. We also discover that firms form stable relationships with partners that possess similar levels of specialization, which helps firms to reduce the levels of competitive uncertainty (Hypothesis 3). Interestingly, both the network attribute factors in our model – third-party ties and position in same clique (Hypotheses 4 and 5) – are significant in the tie formation model but not in the tie decay model. These findings suggest that these network factors are used to mitigate uncertainty associated with forming initial ties, but subsequent to establishing the relationship network factors, they are less relevant. Our explanation for these findings is based on two insights from existing research. First, banks have a strong preference for first-hand knowledge over secondary proxies (Podolny, 1993; Chung et al., 2000; Li & Rowley, 2002), and so once they form a tie, they use their direct experience to evaluate their partners’ reliability. Second, the process of establishing a cooperative partnership is incremental and organic. Whether cooperation is established in the relationship depends on whether partners are successful in the fragile process of reciprocal and escalating contributions that build trust and long-term perspectives (Larson, 1992; Ring & Van de Ven, 1992). Our findings are consistent with the claim that partner-specific experience rather than network constraints and embeddedness determine whether
When do Networks Matter? A Study of Tie Formation and Decay
513
allies engage in cooperative exchanges. Thus, network factors provide less value, once partners have formed their initial tie. One important limitation that we have to address in this study is our inability to directly measure performance of individual relationships. For example, firms may be willing to continue collaborating with partners when such relationships provide them with substantial profits or potential for future profits. A relationship might be plagued with high levels of partner reliability and capability uncertainty; however, both firms would be willing to partner as long as they make money now from this relationship. An example of such a relationship could be the one between a small niche bank working in the rapidly developing emerging market or industry and a large established investment powerhouse. A small bank, despite all the functional and status differences, as well as the lack of indirect connections to the large bank, would be willing to establish and continue relationship with a large bank because the latter would buy from it high-risk issues from this emerging market. Similarly, despite all the differences, a large bank may be willing to continue cooperating with the small partner because doing so provides it with access to an emerging market or industry as well as generating profits (e.g., by buying risky issues from the small bank at a hefty discount and reselling them to the large bank’s own investor groups). Unfortunately, the data that we used did not contain accounting information (e.g., exact amount of underwriting fees or profits) on each IPO transaction, thus we were unable to assess performance of each individual relationship. Future research in the settings where such data is available could shed more light on the influence that performance of ties could have on their longevity, given different types of uncertainty affecting the relationships. Another limitation of this study rests in our usage of dyadic-level data. While we control for various dyad-level characteristics (such as age of a dyad or differences in size of dyad members), we cannot include firm-level characteristics (ages and sizes of individual banks) in the models, since the dyad-level data are derived from firm-level variables. Since we are interested in factors affecting the formation and decay of dyads, the lack of firm-level characteristics does not present a problem on an empirical level, but conceptually it would be interesting to disentangle the effects of differences in firms’ characteristics from the effects of each individual firm’s age and size on the longevity of relationships. This study fits within the stream of research that examines cooperative strategies of individual firms. A highly complementary approach to understanding the processes underlying cooperativeness would involve examination of collective strategies, defined as group-level actions targeted at the achievement of common goals. Examination of collective strategies shifts
514
ANDREW V. SHIPILOV ET AL.
the focus of analysis from individual dyads to the processes taking place in collectivities as a whole. For example, Rao (1998) examines the emergence of consumer watchdog organizations, whereas Pozner and Rao (2006) study drivers for success of collective action of low-power FM radio activists targeted at large commercial radio stations. A powerful insight from this stream of research is that individual actors can engage in collective action without the formation of direct dyadic relationships among them. If within dyads actors rely on relationship-specific mechanisms of coordination (e.g., negotiations and exchange of resources), group coordination within collective action movements is primarily achieved through mutual ideological identification and/or recognition of a common threat. For the emergence of consumer watchdog organizations or a successful low FM radio activism, social movement participants did not have to maintain direct social ties to one another, but it was their common ideological identity and the existence of common threats that prompted them to act in unison. A promising line of enquiry, which can integrate the two streams of cooperative and collective action research, can focus on the evolution of networks in environments requiring collective action. Interesting questions that can be addressed, for example, would include the influence of social ties on the diffusion of collective identities and, conversely, the impact of mutual ideological identification on the formation and stability of social relationships. Furthermore, the integration of the two approaches can help clarify the role of organizational characteristics in the nature of relationships between network actors. Both streams of research highlight the importance of organizational attributes (size, status) as signaling mechanisms, although the two perspectives emphasize different uses of these attributes. The present paper, in line with the tradition of studying cooperative relationships, demonstrates that firms evaluate peers based on their size or status, and seek to form relationships on the basis of these criteria. Studies of collective action, Pozner and Rao (2006) included, emphasize actors’ search for common enemies along these same dimensions. An integrated approach could show, however, that the two uses of organizational attributes take place simultaneously in that firms seek to identify partners on the basis of size and status similarity, and once the relationship is formed, they can solidify it on the basis of encounters with a common enemy whose differences vis-a`-vis the focal group can be vividly exposed along the same organizational attributes. One core insight from partner selection research is that the partner selection process is encoded in the network typology: an organization’s position in its network, its direct ties and density of connection around it determine with whom it will form ties. Our study, however, suggests that network factors are
When do Networks Matter? A Study of Tie Formation and Decay
515
more influential when the network is initially emerging than when it is evolving, and when a bank is searching for new partners rather than reestablishing existing ties. Another core insight from this paper stems from different facets of partnering uncertainty, which we have examined. While most of the prior studies have looked at the ability of relationships to provide firms with access to partners’ resources and ensure their cooperativeness (Baum et al., 2003; Gulati & Gargiulo, 1999; Gulati, 1995; Shipilov, 2006), we also examined the logic of the competition-reducing role of relationships between firms of similar specialization levels. We find that each type of uncertainty: partner capability, partner reliability and competitiveness uncertainty play important, albeit distinct, roles in the decisions of firms to form and to dissolve relationships. Failure to separately examine each of these types of uncertainty has prevented researchers from developing a more comprehensive understanding of factors that influence network evolution. The implication of these finding contradicts the common claim that the influence of network factors on partner selection leads to stability in tie formation decisions and network structures. In contrast, our study suggests that network structure has little influence on whether the network is dynamic or stable. Instead, banks rely on their direct experience with partners to determine whether to reaffirm the relationship or search for a new partner. Stability is a function of building cooperative ties based on reciprocal exchanges of mutual gain (Larson, 1992; Li & Rowley, 2002) rather than network information and governance. Moreover, our study suggests that the networkbased advantages banks gain from their network positions are less deterministic than implied in previous work. It is not the case that networks constrain partner selection and diminish organizations’ freedom to manipulate their networks and positions to their advantage. Banks use their experience to ascertain the value of current relationships and use these evaluations to decide whether to form ties that affirm or evolve the existing structure. In sum, our goal in this study has been to better understand the contingencies influencing how and to what degree the various factors influence partner selection decisions. We place this study within the context of network emergence and dynamics, and hope to motivate future studies linking partnership decisions to the evolution of network structures and network-based advantages.
NOTES 1. We computed three alternative market share specifications, one equally splitting the value of deals among all syndicate members, the others assigning either 25%
516
ANDREW V. SHIPILOV ET AL.
or 75% of the syndicate’s value to the lead bank. Since the average correlation among these specifications was 0.98, we used the 50–50 specification.
ACKNOWLEDGMENT We are grateful to Joel Baum, Stanislav Dobrev and Arjen van Witteloostuijn for their suggestions and encouragement, and to Stan Li for his help with data collection and coding.
REFERENCES Baker, W. E. (1990). Market networks and corporate behavior. American Journal of Sociology, 96, 589–625. Barney, J. B., & Hansen, M. H. (1994). Trustworthiness as a source of competitive advantage. Strategic Management Journal, 15, 175–190. Baum, J. A. C. (1996). Organizational ecology. In: S. Clegg, C. Hardy, & W. Nord (Eds), Handbook of organization studies, London: Sage. Baum, J. A. C., & Korn, H. (1999). Dynamics of dyadic competition. Strategic Management Journal, 20, 251–278. Baum, J. A. C., & Mezias, S. (1992). Localized competition and organizational failure in the Manhattan hotel industry. Administrative Science Quarterly, 37, 580–604. Baum, J. A. C., Rowley, T., Shipilov, A. V., & Chuang, Y. (2005). Dancing with strangers: Aspiration performance and the search for underwriting syndicate partners. Administrative Science Quarterly, 50(4), 673–676. Baum, J. A. C., Shipilov, A. V., & Rowley, T. (2003). Where do small worlds come from? Industrial and Corporate Change, 12(3). Bonacich, P. (1987). Power and centrality: The family of measures. American Journal of Sociology, 92, 1170–1183. Boone, C., Brocheler, V., & Carroll, G. (2000). Custom service: Application and tests of resource-partitioning theory among Dutch auditing firms for 1896 to 1992. Organization Studies, 21(2), 355–381. Borgatti, S., Everett, M., & Freeman, L. C. (1999). UCINet 5 for Windows: Software for social network analysis. Natick: Analytic Technologies. Burt, R. (2000). Decay functions. Social Networks, 22, 1–28. Burt, R. (2002). Bridge decay. Social Networks, 24(4), 333–363. Burt, R., & Knez, M. (1995). Kinds of third-party effects on trust. Rationality and society, 7, 225–292. Carroll, G. (1985). Concentration and specialization: Dynamics of niche-width in populations of organizations. American Journal of Sociology, 94, 723–753. Carroll, G., & Swaminathan, A. (2000). Why the microbrewery movement? Organizational dynamics of resource partitioning in the U.S. brewing industry. American Journal of Sociology, 106(3), 715–762.
When do Networks Matter? A Study of Tie Formation and Decay
517
Chung, S. A., Singh, H., & Lee, K. (2000). Complimentarity, status similarity and social capital as drivers of alliance formation. Strategic Management Journal, 21(1), 1–22. Coleman, J. (1988). Social capital in the creation of human capital. American Journal of Sociology Supplement, 94, S95–S120. Dobrev, S. (2000). Decreasing concentration and reversibility of the resource partitioning process: Supply shortages and deregulation in the Bulgarian newspaper industry, 1987– 1992. Organization Studies, 21(2), 383–404. Dobrev, S., & Carroll, G. (2003). Size (and competition) among organizations: Modelling scalebased selection among automobile producers in four major countries, 1885–1981. Strategic Management Journal, 24(6), 541–558. Dobrev, S., Kim, T., & Carroll, G. (2003). Shifting gears, shifting niches: Organizational inertia and change in the evolution of the U.S. automobile industry, 1885–1981. Organization Science, 14(3), 264–282. Eccles, R., & Crane, D. (1988). Doing deals: Investment banks at work. Boston, MA: Harvard Business School Press. Edwards, C. D. (1955). Conglomerate bigness as a source of power. Princeton, NJ: Princeton University Press. Ellis, K., Michaely, R., & O’Hara, M. (2000). When the underwriter is the market maker: An examination of trading in the IPO aftermarket. Journal of Finance, 55, 1039–1074. Galaskiewicz, J., & Burt, R. S. (1991). Interorganization contagion in corporate philanthropy. Administrative Science Quarterly, 36, 88–105. Gimeno, J. (2004). Competition within and between networks: The contingent effect of competitive embeddedness on alliance formation. Academy of Management Journal, 47(6), 820–842. Greve, H. (1999). The effect of core change on performance: Inertia and regression toward the mean. Administrative Science Quarterly, 44(3), 590–614. Gulati, R. (1995). Does familiarity breed trust? The implications of repeated ties for contractual choice in alliances. Academy of Management Journal, 38(1), 85–112. Gulati, R., & Gargiulo, M. (1999). Where do inter-organizational networks come from? American Journal of Sociology, 104(5), 1439–1493. Hamel, G., Doz, Y., & Prahalad, C. K. (1989). Collaborate with your competitors and win. Harvard Business Review, 67(1), 137–139. Haveman, H. (1993). Organizational size and change: Diversification in the savings and loan industry after deregulation. Administrative Science Quarterly, 38, 20–50. Jayachandran, S., Gimeno, J., & Varadarajan, P. R. (1999). Theory of multimarket competition: A synthesis and implications of marketing strategy. Journal of Marketing, 63(3), 49–66. Jensen, M. (2003). The role of network resources in market entry: Commercial banks’ entry into investment banking 1991–1997. Administrative Science Quarterly, 48, 466–497. Keohane, R. (1986). Reciprocity in international relations. International Organization, 40, 1–27. Khanna, T., Gulati, R., & Nohria, N. (1998). The dynamics of learning alliances: Competition, cooperation and relative scope. Strategic Management Journal, 19, 193–210. Korn, H., & Rock, T. (2001). Beyond multimarket contact to mutual forbearance: Pursuit of multimarket strategy. In: J. A. C. Baum & H. R. Greve (Eds), Advances in Strategic Management. Oxford, UK: JAI. Larson, A. (1992). Network dyads in entrepreneurial settings: A study of the governance of exchange processes. Administrative Science Quarterly, 37, 76–104.
518
ANDREW V. SHIPILOV ET AL.
Levinthal, D. (1991). Random walks and organizational mortality. Administrative Science Quarterly, 36, 397–420. Li, S. X., & Rowley, T. J. (2002). Inertia and evaluation mechanisms in interorganizational partner selection: Syndicate formation among U.S. investment banks.. Academy of Management Journal, 45(6), 1104–1120. Madhavan, R., Koka, B., & Prescott, J. (1998). Networks in transition: How industry events (re)shape interfirm relationships. Strategic Management Journal, 19, 439–459. Miles, R. E., & Snow, C. C. (1978). Organizational strategy, structure, and process. New York: McGraw-Hill. Mohr, J., & Spekman, R. (1994). Characteristics of partnership success: Partnership attributes, communication behavior, and conflict resolution techniques. Strategic Management Journal, 15(2), 135–152. Nohria, N., & Garcia-Pont, C. (1991). Global strategic linkages and industry structure. Strategic Management Journal, 12(Summer Special Issue), 105–124. Oxley, J. (1997). Appropriability hazards and governance in strategic alliances: A transaction cost approach. Journal of Law Economics and Organization, 13(2), 387–409. Pfeffer, J., & Salancik, G. (1978). The external control of organizations: A resource dependence perspective. New York: Harper and Row. Podolny, J. M. (1993). A status-based model of market competition. American Journal of Sociology, 98(4), 829–872. Pollock, T., Porac, J., & Wade, J. (2004). Constructing deal networks: Brokers as network ‘‘architects’’ in the U.S. IPO market and other examples. Academy of Management Review, 29, 50–72. Powell, W., White, D., Koput, K., & Owen-Smith, J. (2004). Network dynamics and field evolution: The growth of inter-organizational collaboration in the life sciences.. American Journal of Sociology, 110, 1132–1205. Powell, W. W., Koput, K. W., Smith-Doerr, L., & Owen-Smith, J. (1999). Network position and firm performance: Organizational returns to collaboration in the biotechnology industry. Research in the Sociology of Organizations, 16, 129–159. Pozner, J-E., & Rao, H. (2006). Fighting a common foe: Enmity, identity and collective strategy. In: J.A.C. Baum, S. Dobrev & A. van Witteloostuijn (Eds), Ecology and strategy: Advances in strategic management (Vol. 23, pp. 445–479). Oxford, UK: JAI. Provan, K. G., & Sebastian, J. G. (1998). Networks within networks: Service link overlap, organizational cliques, and network effectiveness. Academy of Management Journal, 41, 453–463. Rao, H. (1998). Caveat emptor: The construction of nonprofit consumer watchdog organizations. American Journal of Sociology, 103(4), 912–961. Ring, P. S., & Van de Ven, A. (1992). Structuring cooperative relationships between organizations. Strategic Management Journal, 13, 483–498. Rowley, T., Baum, J. A. C., Shipilov, A. V., Rao, H., & Greve, H. R. (2004). Competing in groups. Managerial and Decision Economics (25), 453–471. Rowley, T., & Baum, J. A. C. (2004). Sophistication of interfirm network strategies in the Canadian investment banking industry. Scandinavian Journal of Management, 25(6–7), 103–124. Rowley, T., Behrens, D., & Krackhardt, D. (2000). Redundant governance structures: An analysis of structural and relational embeddedness in the steel and semiconductor industry. Strategic Management Journal, 21, 369–386.
When do Networks Matter? A Study of Tie Formation and Decay
519
Rowley, T., Greve, H., Rao, H., Baum, J., & Shipilov, A. V. (2005). Time to break up: Social and instrumental antecedents of firm exits from cliques. Academy of Management Journal, 48(3), 499–521. Rowley, T. J. (1997). Moving beyond dyadic ties: A network theory of stakeholder influences. Academy of Management Review, 22, 887–910. Shipilov, A. V. (2005). Should you bank on your network? Relational and positional embeddedness in the making of financial capital. Strategic Organization, 3(3), 279–309. Shipilov, A. V. (2006). Network strategies and performance of Canadian investment banks. Academy of Management Journal, forthcoming. Simmel, G. (1950). The sociology of Georg Simmel. New York: Free Press. Stephan, J., & Boeker, W. (2001). How multimarket contact affects firms’ market entry decisions. In: H. R. Greve & J. A. C. Baum (Eds), Multiunit organization and multimarket strategy. Oxford, UK: JAI. Stinchcombe, A. (1965). Organizations and social structure. In: J. March (Ed.), Handbook of organizations. Chicago, IL: Rand McNally. Stuart, T. E. (2000). Interorganizational alliances and the performance of firms: A study of growth and innovation rates in a high-technology industry. Strategic Management Journal, 21(8), 791–811. Uzzi, B. (1996). The sources and consequences of embeddedness for the economic performance of organizations: The network effect. American Sociological Review, 61, 674–698. Walker, G., Kogut, B., & Shan, W. (1997). Social capital, structural holes and the formation of an industry network. Organization Science, 8(2), 109–125. Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and applications. Cambridge, UK: Cambridge University Press.
This page intentionally left blank
520
PART VIII: SCALE AND SCOPE
521
522
SCALE AND SCOPE ECONOMIES IN THE BRITISH MOTORCYCLE INDUSTRY, 1899–1993 Filippo C. Wezel and Arjen van Witteloostuijn ABSTRACT This paper extends organizational ecology by making an attempt to disentangle the consequences of scale and scope economies for organizational survival under different product market configurations. We test our hypotheses by analyzing the mortality rates of 643 UK motorcycle producers during the 1899–1993 period. The findings obtained offer two specific contributions. First, by separating the performance impact of scale from scope economies we clarify the complex mechanisms behind the survival consequences of different organizational strategies. Second, we show how the intensity of both scale and scope forces is relative to the aggregate market-level product configuration. The implications of these findings for organizational ecology and strategic management, and their cross-fertilization, are further discussed.
INTRODUCTION Organizational ecology has a long tradition of studying the macro- and micro-level determinants of organizational survival (Carroll & Hannan, Ecology and Strategy Advances in Strategic Management, Volume 23, 523–548 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23016-X
523
524
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
2000). Several theoretical fragments within organizational ecology have argued that long-run survival heavily depends on cost conditions. Take, as a case in point, Carroll’s (1985) resource-partitioning theory. In this theory, scale economies are a key driver of the survival of generalist organizations. However, as some authors have argued (e.g., Boone & van Witteloostuijn, 2004; van Witteloostuijn & Boone, 2006), organizational ecology’s theoretical apparatus can be further improved by building on more fine-grained cost-based insights. Indeed, the above authors refer to industrial organization and strategic management as sources of inspiration. Strategic management scholars have underscored, and still do so, scale and scope economies as important determinants of competitive advantage (Venkatraman & Subramaniam, 2002). Largely influenced by the industrial organization tradition, early strategy research focused on the competitive advantage obtained by leveraging (dynamic) economies of scale via learning and/or experience curves (e.g., Henderson, 1979; Lieberman, 1987). Later in the field’s history, a growing literature on organizational capabilities argued how cost and knowledge sharing across multiple products (i.e., scope economies) can be the main source of competitive advantage (e.g., Amit & Schoemaker, 1993; Holcomb, Holmes, & Hitt, 2006). As already mentioned, organizational ecology studies have been concerned with scale and scope arguments as well. In particular, they have demonstrated that, under specific conditions, large scale or multi-product organizations exhibit survival advantages because they benefit from a lowcost position or broad-scope risk spreading, respectively. In the first place, resource-partitioning studies have provided evidence for the survival benefits of large-scale generalist profiles – at least before the emergence of strong competitive pressures, after which they are forced to engage in tough competition (Carroll & Swaminathan, 2000; Carroll, Dobrev, & Swaminathan, 2002). In the second place, empirical studies have revealed that organizational niche width (reflected in, e.g., the number of products) decreases organizational failure in industries characterized by coarsegrained environmental changes (Dobrev, Kim, & Hannan, 2001; Wezel & van Witteloostuijn, 2004). Cross-fertilization between strategic management and organization ecology provides two distinct advantages (see also Boone, Carroll, & van Witteloostuijn, 2004; Boone & van Witteloostuijn, 2004). First, oftentimes organizations seek both potential sources of competitive advantage: scale and scope economies. The question then is which of both forces is more important than the other in shaping long-run performance – i.e., what is the relative importance of scale and scope economies? Second, the application
Scale and Scope Economies in the British Motorcycle Industry
525
of ecological reasoning may help to overcome the firm-centric view that dominates many strategic management theories. In spelling out the logic of our argument, we will emphasize how the importance of scale and/or scope economies depends critically upon the population-level distribution of competitors across the industry’s product niches. Our key argument is that the pattern of product distribution at the industry-level matters, too, in determining the survival of an individual organization. Industries marked by high homogeneity of products are populated with similar organizations, and thus are more competitive than those characterized by more heterogeneity (Boone, Wezel, & van Witteloostuijn, 2005). Moreover, highly homogeneous industries (in terms of their product market distribution) are expected to exhibit tough competition in the abundant resource areas, leaving untapped market pockets that are ideal for leveraging scope economies. This may suggest that the extent to which scale and scope economies will increase the likelihood of organizational survival, depends on the relative position in the industry’s product niche ‘space’ of the focal organization vis-a`-vis its competitors. To test this, we will add interaction terms to main-effect benchmark model specifications. The paper is organized as follows. In the next section, we will first define our key concepts. Subsequently, we will provide our study’s theoretical background, formulating a set of hypotheses about mortality rates, which is the mirror image of the survival, of a focal organization’s scale and scope economies (or lack thereof), as well as their interaction with a measure of the shape of the population-level product distribution. Subsequently, we will introduce this chapter’s empirical setting: the British motorcycle industry in the 1899–1993 period. As we were not able to obtain firm-level cost structure data for such a long span of time, we will explain why we believe that our measures offer reasonable proxies. After that, we are ready to present our empirical evidence. Finally, we conclude with an appraisal.
THE KEY CONCEPTS DEFINED Scale and Scope Economies Scale and scope economies are central to many industrial organization, organizational ecology and strategic management arguments. Consequently, definitional issues must be settled up-front. Scale and scope economies can be the result of lower costs or positional benefits. In industrial organization,
526
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
scale cost economies are determined by cost advantages that come with the large-scale development (R&D scale), production (manufacturing scale) and selling (marketing scale) of a single product. An example of a scale benefit is market power: a firm that size-dominates the market, can sustain high pricecost margins. Scope economies pertain to the benefit and/or cost advantages that are associated with the combined production and selling of a portfolio of different products. A scope cost advantage is, for instance, any cost reduction that results from the multi-product sharing of input resources (such as shared raw material and production machinery). A benefit advantage from scope may, for example, be derived from multi-market branding by transferring a strong brand name from one market or product to the other, or joint R&D that generates new process technologies or product features that spill over from one market or product to the other. In our context, the cases of the resource-partitioning and niche-width theories are particularly relevant. In resource-partitioning theory (Carroll, 1985), on the one hand, ‘‘the generalist organization that secures a toehold in the dense market center possesses a potentially sustainable competitive advantage over all competitors’’ (Carroll & Hannan, 2000, p. 263). Notice that scope economies play a (limited) role in the theory, too, as the fight for the market center gives the surviving generalists the opportunity to expand into the neighboring areas left uncovered after the failure of competitors. In niche-width theory (Freeman & Hannan, 1983), on the other hand, scope economies are more relevant. There, the argument is that firms operating in multiple niches can benefit from a buffer against (coarse-grained) environmental shifts. In Table 1, both theories’ different sets of definitions are reproduced, in a summary fashion, in the second and third columns. In our setting, we will mix the scale and scope arguments, by building on the above theories, to estimate their absolute and relative impact upon an organization’s likelihood of survival.
Population-Level Product Diversity Ecology and strategy tend to differ in the way in which they define and treat the environment. In the present chapter, we analyze the impact of population-level product (or, conversely, homogeneity) or heterogeneity as reflected in the industry’s aggregate product distribution. We argue that any scale or scope economies advantage stemming from specific positions in resource space has to be evaluated vis-a`-vis the distribution of all competitors in the industry – i.e., the aggregate product diversity.
Scale and Scope Economies in the British Motorcycle Industry
Table 1. Theories
527
Resource-Partitioning and Niche-Width Theory’s Key Definitions. Economies
Forms
Niche width
Scope economies
A generalist reaps scope economies by being located in multiple niches, which provides a buffer against course-grained environmental changes
Resource partitioning
Scale and (limited) scope economies
A generalist reaps scale economies by being located in the large resource-abundant center, and scope economies by expanding into neighboring areas left uncovered by failed competitors
We define the degree of homogeneity in the industry with reference to the distribution of organizations in product space. This space has two dimensions: the number of different products or niches, and the distribution of offerings across these niches. Take the extreme cases. On the one hand, if all organizations offer the very same product (hence, the market is associated with a single niche only), then the industry is associated with a maximum degree of product homogeneity. On the other, if organizations are equally distributed across a series of niches, then the market is characterized by a maximum degree of product heterogeneity. Of course, the population-level product homogeneity – heterogeneity dimension is continuous, with many different values between these extreme cases.
THEORETICAL BACKGROUND The Survival Consequences of Scale and Scope Economies Industrial organization and strategic management scholars agree that multiproduct firms tend to exhibit greater profitability. A mix of scale and scope advantages can be put forward to justify this claim. Offering multiple products reduces substitutability with competing ones, diminishes demand elasticity and hence increases firm profits. In particular, industrial organization suggests that high product differentiation leads to high profitability as a consequence of increased barriers to entry (Bain, 1956). The economics literature suggests that multi-product firms increase their profits by pooling resources, staff and expertise and, thus, by benefiting from economies of
528
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
scale and/or scope (Panzar & Willig, 1981). Since the seminal work of Baumol, Panzar, and Willig (1982), economies of scale and economies of scope have been studied in a wide range of industries and fields (e.g., Murray & White, 1983; Deller, Chicoine, & Walzer, 1988; Cohn, Rhine, & Santos, 1989; Fournier & Mitchell, 1992; Wholey, Feldman, Christianson, & Engberg, 1996). Research in strategic management has built on these findings, stressing the benefits associated with scale and scope economies (Venkatraman & Subramaniam, 2002). A parallel literature in organizational ecology has pointed in a similar direction. According to niche-width theory, generalist organizations hedge against environmental changes by offering products in multiple niches (Hannan & Freeman, 1977). By and large, the empirical literature about nichewidth evolution and organizational form performance deals with the question as to how the vital rates of generalists and specialists are affected by such features as niche width, niche position and niche crowding. The backbone of this research tradition lies in niche-overlap theory (McPherson, 1983; Baum & Singh, 1994a, 1994b), which argues that competitive intensity is proportional to the organizations’ degree of overlap along critical dimensions. McPherson (1983) pointed out that competition between two voluntary organizations for new recruits is proportional to their socio-demographic similarities. Baum and Singh (1994b) showed that positional overlap across segments of the child-care population produced higher failure levels. Similar findings have been found regarding technological (Podolny, Stuart, & Hannan, 1996) and geographical (Baum & Mezias, 1992) overlap. In this stream of research, generalism and specialism have been defined according to a gradient of size: large organizations have been labeled as generalists, and small ones as specialists (Boone, Bro¨cheler, & Carroll, 2000). Other studies have used the number of products (Dowell & Swaminathan, 2000; Sorenson, 2000), whereas more recent research has opted for a continuous measure of the organizational niche as related to the market range covered along a technological dimension, such as engine size in the automobile industry (Dobrev et al., 2001; Dobrev, Kim, & Carroll, 2002). This body of research agrees that, holding market concentration constant (in terms of the sales distribution), generalist organizations exhibit survival advantages vis-a`-vis their specialist counterparts. From a resource-partitioning perspective, a generalist organization is more likely to exhibit a potentially sustainable advantage by positioning its offer in the most resource-abundant niche in the product space – i.e., in the center of market. From the angle of niche-width theory, under specific conditions, scope economies may be obtained by leveraging a competitive
Scale and Scope Economies in the British Motorcycle Industry
529
position in multiple niches. In so doing, we envision different types of generalism, depending upon whether the focal organization’s product portfolio or niche width does or does not include an offer in the most abundant portion – or center – of the industry’s product space. In this way, we elaborate on concepts, definitions and insights from resource partitioning and niche-width theories – i.e., scale economies in the market center and a multiple-niche position with scope economies across space, respectively. Our argument is that being located in the market center offers an extra scale advantage to organizations located there (resource-partitioning theory), which is different from the scope advantages derived from spreading across multiple niches (niche-width theory). In resource-partitioning theory, an organization may exploit the advantages of a large-scale production technology especially when it produces standardized mass products for the center of the market. The underlying logic is that cost-reducing large-volume production and low-price marketing strategies are a key to obtain the central organizational objective of growing large by developing market share in order to be able to reap further scale economies. This claim is in line with the findings of Dobrev and his colleagues (2002, 2004), who found that locating far from the market center (in either way – in their automobile industry setting: either above or below the centrist segment of cars) positively impacted upon the disbanding rate of US automobile producers during the 1885–1981 period. Conversely, under the assumption of an environment characterized by coarse-grained changes and following niche-width theory’s argument, the advantage obtained by leveraging scope economies is, in principle, independent from the relative abundance of each niche, but rather relates to the sharing of resources, skills and knowledge for the development, production and marketing of different products for different niches (see Dobrev et al., 2001; Wezel & van Witteloostuijn, 2006). Hence, we propose our first two main-effect hypotheses. Hypothesis 1. (scope economies): Positioning in multiple product niches (other than the center) decreases organizational mortality rates. Hypothesis 2. (scale economies): Locating in the market center decreases organizational mortality rates. The Survival Consequences of the Market-Level Product Distribution As we explained above, the shape of the industry-level product distribution matters for organizational performance, too. In effect, the distribution of products at the population level implies a trade-off at the organizational
530
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
level (for a thorough discussion, we refer to Boone et al., 2005). Increasing product homogeneity at the signals the potential for different types of spillover effects – e.g., infrastructure, knowledge and information externalities. A high level of industry-level homogeneity indicates crowding in the product space – i.e., competition – because similar organizations are competing for similar scarce resources (Hawley, 1950; Hannan & Freeman, 1977). This trade-off line of reasoning suggests that at high levels of populationlevel product homogeneity further homogeneity-enhancing increments generate competitive pressures that increase mortality rates. In other words, there is some kind of ‘optimal’ distribution of organizations in product space – in which the trade-off forces of clustering together or differentiating away are balanced. From this logic, a curvilinear effect of population-level product or organizational homogeneity on mortality can be derived (Boone et al., 2005). Hence, we expect a U-shaped main effect of industry-level product homogeneity on organizational mortality. Hypothesis 3. (population-level product homogeneity): The relationship between population-level product homogeneity and organizational mortality rates is U-shaped. Scale Economies, Scope Economies and Product Homogeneity in Interaction So far, we discussed main-effect hypotheses only, either relating to scale and scope economies (Hypotheses 1 and 2) or population-level product homogeneity (Hypothesis 3). However, implicit in our reasoning is that the market’s product distribution – i.e., population-level product homogeneity or heterogeneity – moderates the positive survival effects of scale and scope economies. That is, we argue that the impact of scale and/or scope economies, being associated to specific positions in product space, depends upon the relative position of the focal organization vis-a`-vis all of its competitors. Below, we explore this interaction–effect logic. From the perspective of strategic management, the case related to competition at high industry-level homogeneity is particularly interesting. Specifically, we expect that the survival consequences of being located in the center of the market will erode at increasing population-level product homogeneity. Under these conditions, in effect, a location in the center of the market (i.e., betting on scale economies) amplifies the negative implications of similarity for the likelihood of organizational survival. On the contrary, being positioned in other niches than the center of the market (i.e., seeking
Scale and Scope Economies in the British Motorcycle Industry
531
scope economies) then brings the benefits of differentiating away from competitors, and thus reduces the negative consequences of competition for organizational survival. So, the impact of scale and scope economies on the likelihood of organizational survival depends critically upon the shape of the population-level product distribution – i.e., the degree of industry-level product homogeneity or heterogeneity. Indeed, resource-partitioning studies have demonstrated that at high levels of (market share) concentration, specialist non-center organizations perform better than their generalist center counterparts (e.g., Boone, Carroll, & van Witteloostuijn, 2002, 2004). It is interesting to note that our logic may provide an interpretation of two findings in the extant literature. On the one hand, niche width has been shown to decrease organizational survival chances at increasing market share concentration (Dobrev et al., 2001). On the other, however, Dobrev and his colleagues (2002, 2004) have shown how in the US automobile industry the survival consequences of being located far away from the market center increase with market share concentration. We interpret these two seemingly conflicting findings as proof for the need to unravel the two different underlying forces related to market center positioning (associated with a positive effect on mortality at high concentration) and to niche width (with a negative impact on mortality at high concentration). The above studies focus on size-based measures of concentration, though, as normally applied in industrial organization and strategic management research as well. In this chapter, we adopt a line of reasoning similar to Boone, Wezel, & van Witteloostuijn (2005), which suggests that the benefits of differentiation not only unfold from increasing market share concentration, but also from concentration in product space – i.e., increasing organizational similarity. The magnitude of the effect of a focal organization’s product portfolio on the likelihood of survival thus depends on the shape of the aggregate market-level product distribution: the impact of scale versus scope economies is moderated by the relative position of the focal organization vis-a`-vis its competitors. The above logic leads to the following pair of interaction-effect hypotheses. Hypothesis 4. (scale economies* population-level product homogeneity): As population-level homogeneity increases, the negative impact of scale economies on organizational mortality is reduced. Hypothesis 5. (scope economies* population-level product homogeneity): As population-level product homogeneity increases, the negative impact of scope economies on organizational mortality is reinforced.
532
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
DATA, MEASUREMENT AND METHOD Data To test our theoretical argument, we opted for the United Kingdom (UK) motorcycle industry for several reasons. First, the different characteristics of motorcycles permit a straightforward partitioning of the market in meaningful niches, which we need to measure scale and scope economies. Second, the availability of accurate records of the overall histories and vital events in this industry offers opportunities to study the effects of both types of economies on organizational survival performance over the complete history from 1895 to 1993. Third, detailed information on products is available that we can exploit to calculate niche-specific product densities, from which we can subsequently compute annual population-level product homogeneity – heterogeneity variables. As a side effect, the significant body of research on the change dynamics of automobile manufacturers (e.g., Dobrev et al., 2001, 2004), an industry that in many ways resembles the motorcycle producer population, facilitates the accumulation and comparison of empirical results. Figures 1(a) and 1(b) depict the history of this paper’s industry in terms of density, and the number of entries and exits of UK motorcycle producers. The data used to generate these figures includes 643 motorcycle producers during the period from 1895 to 1993. The main source of information is British Motorcycles since 1900 (Collins, 1998), which includes the date of birth and disbanding of each firm in the UK. Data at the product level were obtained from The Register of Machines of the VMCC (Hume, 1991). Other sources used are The Complete Illustrated Encyclopedia of the World’s Motorcycles (Tragatsch, 1977, 2000), Historic Motorcycles (Burgess Wise, 1973), The Ultimate Motorcycle Book (Wilson, 1993) and Encyclopedia of Motorcycling (Bishop & Barrington, 1995). The year in which the first model of producer x was mentioned in the sources, was coded as the year of birth of this firm x, and the year in which firm x’s last model disappears from the sources’ registers, was coded as its year of death.
Measurement Dependent Variable In line with the organizational ecology tradition, Organizational Mortality is the dependent variable. The main exit events usually associated with the
533
Scale and Scope Economies in the British Motorcycle Industry 200 180 160
Density
140 120 100 80 60 40 20 1895 1898 1901 1904 1907 1910 1913 1916 1919 1922 1925 1928 1931 1934 1937 1940 1943 1946 1949 1952 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991
0
Time
(a) 60
50 entries exits 40
30
20
10
(b)
Fig. 1.
year 1896 1899 1902 1905 1908 1911 1914 1917 1920 1923 1926 1929 1932 1935 1938 1941 1944 1947 1950 1953 1956 1959 1962 1965 1968 1971 1974 1977 1980 1983 1986 1989 1992
0
(a) Density of UK Motorcycle Producers from 1895 to 1993; (b) Entries and Exits of UK Motorcycle Producers from 1895 to 1993.
ending of an organization’s history are disbanding, exit to another industry, and a merger or acquisition. While our data set is clear about disbanding, this is not always so regarding the mode of exit. Those firms known to have ended through an acquisition or a merger were treated as right-censored
534
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
observations, a standard practice in organizational ecology studies. Following Dobrev, Kim, and Hannan (2001), we assumed ‘unknown events’ – about eight percent of our population – to be governed by processes very similar to those related to pure exits. Therefore, the outcome event of interest in our analyses is ‘‘disbanding/exit to another industry, defined to include events of unknown type’’ (Dobrev, 2001, p. 1311). Independent Variables Each organization in a population occupies an organizational niche characterized by a set of organizational capabilities and a location in the population’s resource space (Baum & Singh, 1994a, 1994b). In line with previous studies (Podolny et al., 1996; Dobrev et al., 2001), we define the organization’s niche with reference to the technological space, which in the motorcycle industry is very closely related to the issue of product range. In particular, like Dobrev et al. (2001) did in their study of the European automobile industry, we measure the range of motorcycles offered by each producer by the spread of the engine capacity of the models they have on offer. In doing so, according to a standard practice in this industry, we segment the market into seven main niches: up to 50cc, 51–125cc, 126–250cc, 251–350cc, 351–500cc, 501–600cc, and more than 600cc. Not all seven niches existed over the entire time period of our study. For example, the ‘more than 600cc’ type appeared only in 1900. The ‘up to 50cc’ category, for instance, emerged only in 1923 and disappeared between 1926 and 1936. We opt for superimposing the above seven categories, characterized by their technological potential, over the entire observation window: then, as soon as the first motorcycle was invented, these categories became technologically feasible. Following this rationale, we end up with a complete description of all the market niches covered by each of the motorcycle producers in the UK for each and every year of the history of this industry from 1895 to 1993. Following previous studies, we define the organization’s product portfolio (or niche) width with reference to technological space. Owing to lack of longitudinal firm-level cost data, we decided to construct proxies using this technological space information. For one, those firms offering a product in the most densely populated niche are defined as leveraging Scale Economies (Hypothesis 2). Our justification for such a positional measure is the following. High product density in the center niche may signal that being located there is attractive. Of course, this is an imperfect measure of the resource peak. We would very much have liked to use a measure of demand per cc-category. Regrettably, this type of information is unavailable. Therefore, we build a measure on the assumption that
Scale and Scope Economies in the British Motorcycle Industry
535
segment-level demand is positively correlated with the number of producers offering one or more models in the associated cc-segment of the market. The assumption is that niche-level demand is positively correlated with the number of organizations offering one or more products in the associated segment of the market. This assumption relies on straightforward entry and exit economics, arguing that – ceteris paribus – large or growing markets attract many entrants, while small or declining markets force many incumbents to exit (Baumol et al., 1982). Notice also that such a measure allows the peak to freely change location over time – e.g., from the high to the low cc-segments, or from the tails to the center. Additionally, we measure the range of motorcycles offered by each producer – Scope Economies – by the spread of the engine capacity of their motorcycles (Hypothesis 1). To account for the advantage of scope economies relative to the time-varying potential of the market, we divided this measure by the width of the market in that specific year. As all the firms located in the center are positioned in at least one other niche as well, by adding the Scale Economies dummy we are able to show how competitive advantages (i.e., superior survival chances) stem from two different forces – one related to production in the resource-abundant market center (i.e., Scale Economies), and the other obtained via producing multiple products in niches different from the market center (i.e., Scope Economies). The correlation between these two variables is moderate – i.e., 0.40 – suggesting that we are indeed measuring two different aspects of an organization’s strategy. Of course, our measures of scale and scope economies are imperfect proxies. However, we believe that our measures are reasonable proxies. As firm-level cost data are generally very difficult to obtain, particularly in the context of longitudinal studies, such proxies are needed to carry out any ecological research that seeks to explicitly take on board the cost-related arguments that are so essential for the theories involved. Not only do longitudinal firm-level cost data tend to be unavailable: the same holds true for niche-level demand information. Needless to say, lacking a niche-specific measure of market demand, we were concerned about the reliability of our peak measure. However, additional information does suggest that our measure makes sense. First, according to our measure, 1947 marks the year of a radical relocation of the market center from the 351–500cc segment to the 51–125cc niche. Indeed, the history of this industry suggests that right around this time one of most significant changes in demand took place. Because of the need for cheap utility transport, after WWII sales of small-capacity motorcycles boomed and ‘‘basic, cheap and mostly two strokes [motorcycles] buzzed about the streets’’
536
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
(Tragatsch, 2000, p. 46). Second, by the same token, during the 1920s our peak’s location remains stable at the 251–350cc category, which is nicely in line with information reported on the ‘leading models of the 1930s’ (see the appendix in Tragatsch, 2000): out of the 167 UK models listed, 38 percent fell within the 251–350cc category, by far the most represented. Third, we expect center organizations to be, on average, larger than their noncenter counterparts if the latter focus on scale economies, as argued above. To double-check the robustness of our center measure, we therefore computed the correlation between the number of plants of each producer (our proxy for size: see below for more details) and a center location dummy. We interpret the positive and statistically significant coefficient (b ¼ 0.0975 with t ¼ 6.7 and po0.001) as providing support for our measure. Similarly, organizations located in the market center exhibit a higher average number of plants (1.9), compared to their counterparts without a center position (1.4). Building on our cc-classification of niches, we measure the overall degree of population-level product homogeneity or heterogeneity through a Hirschmann–Herfindahl Index (HHI) of product similarity. This HHI indicates how concentrated the supply of products is in a given year by (a) counting the percentages of producers located in each cc-segment and (b) subsequently summing the squared percentages. This is our measure of Population-Level Product Homogeneity. In biology, the same measure is widely used. The Simpson-index maps the diversity of species in a habitat, putting together two dimensions: richness (number of species in the habitat) and evenness (distribution of species across the habitat). Similarly to the HHI, the Simpson-index combines both dimensions to map the distribution of species controlling for their number. Indeed, the Simpson-index and the HHI produce identical results. In our empirical setting of the British motorcycle industry in the 1895–1993 period, though, the ‘vertical’ dimension is driving the variance of the HHI over time, given that we keep the horizontal dimension (i.e., cc-classes) constant over the complete time window. As the number of cc-categories is fixed to seven in this case, the minimum value equals 0.14 (i.e., 7n(1/7)2, implying maximum heterogeneity or diversity). Fig. 2 (see above) provides the evolution of the HHI measure. The linear and squared terms of the HHI are used to test for the U-shaped effect of population-level product homogeneity of Hypothesis 3 on exit rates. An interaction product term of each focal firm’s competitive position (as described above) with the HHI squared, Scale Economies* PopulationLevel Product Homogeneity and Scope Economies* Population-Level Product Homogeneity, is added to test Hypotheses 4 and 5.
Scale and Scope Economies in the British Motorcycle Industry
537
0.59 0.54 0.49 0.44 0.39 0.34 0.29 0.24 0.19
1898 1901 1904 1907 1910 1913 1916 1919 1922 1925 1928 1931 1934 1937 1940 1943 1946 1949 1952 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991
0.14
Fig. 2.
Hirschmann-Herfindahl Measure of Population-Level Product Homogeneity in the UK Motorcycle Industry from 1899 to 1993.
Control Variables A few control variables are introduced. Organizations can enter an industry following different paths. For instance, previous research on the American automobile industry (Carroll, Bigelow, Seidel, & Tsai, 1996) found that the life chances of lateral entrants were higher than of those entering de novo. Thus, inspired also by recent findings of Dobrev, Kim, and Solari (2004), we include three dummy variables to code different entry modes: DeNovo inexperienced organizations, DeAlio lateral entrants, and DeIpso firms that entered through a merger or resurgence of motorcycle manufacturers. The age of an organization is measured as the tenure of the firm in the industry. Unfortunately, the archival sources we used contain only the year of the event. Thus, following Petersen’s (1991) advice, we benchmark organizational Age from the midpoint of the year of entry. Following this logic, those firms entering and exiting in the same year receive a tenure value of 0.5 in their year of founding. Similar to Barnett and McKendrick (2001), we define the intensity of competition among firms in proportion to their degree of overlapping product range. An organization that, for instance, produces 125, 250 and 350cc motorcycles receives a value of 1/3 of competition from those firms that it meets in only a single niche – e.g., 125cc – and a value of 1
538
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
from those rivals that offer the same range of products. These Niche Overlap measures are recorded on a yearly basis. To account for the confounding effects of mutual forbearance, following Baum and Korn (1996), we divided our measure by the annual count of the number of organizations that do overlap with the focal one in at least two niches. We controlled for the remaining effects of diffused legitimation by counting Density as the number of incumbent UK motorcycle producers in each year t. We squared this value to capture the effects of competition. We further control for the negative impact of firm size on organizational mortality by creating a time-varying variable, counting the number of plants owned by the focal firm – Size. Owing to our dataset’s century-long time window, we were unable to collect conventional size measures. The number of plants is a sensible surrogate measure, though. For instance, two studies of Mata and Portugal (1994, 2000) reveal a strong positive and statistically significant correlation between number of plants and standard size measures (e.g., number of employees). We log this measure to reduce its variance. As the reliability of our HHI measure depends on the minimum efficient scale of the industry, we compute a variable named Minimum Efficient Scale (MES), defined as the yearly number of motorcycle sales divided by the total number of plants in the industry. We further controlled for the number of Total Motorcycle Sales (log) as well as the market breadth by adding a timevarying variable indicating the Market Width. The British motorcycle industry had a dominant production area in the geographic triangle between the cities of Birmingham, Coventry and Wolverhampton. Almost half (278 out of 639) of all manufacturers, including Ariel, BSA, (Royal) Enfield, Norton and Triumph, were located there (see Wezel, 2005, for the dynamics of foundings within this region). To control for this, we added a dummy variable District to identify those organizations having their headquarters in that area. To control for the general economic climate and for latent demand for transportation, we introduce a variable measuring the gross domestic product per person – GDP per Person – using data obtained from Mitchell (1998). Finally, three dummies control for the influence of WWI, WWII and the Japanese era – Post-1958.
Method The HHI is artificially high during the early years of the industry because of the small number of initial incumbent organizations. To avoid overestimating the effect of the HHI during the early years, we decided not to use the
Scale and Scope Economies in the British Motorcycle Industry
539
first four years, but rather to start our observation window in 1899 [cf. Boone et al. (2000) for a similar procedure in their resource-partitioning study of audit firm concentration]. Because of that, the final data set includes 598 exits, implying 4,678 firm-year observations. We divide the life of each producer in organization-years through the spell-splitting technique (Allison, 1984; Tuma & Hannan, 1984). To allow the baseline hazard rates of the different categories of producers to vary in an unrestricted fashion, we model this rate according to a piecewise exponential formulation. More precisely, the age of an organization is divided into intervals, and the hazard is constant within each interval but can vary across intervals. We define a set of J intervals, dividing the age variable at precise points (a1, a2, a3, a4, y, aj), where a0 ¼ 0 and aj ¼ N. The interval J is given by [aj1, aj), and the hazard of the firm i is defined by rðtÞ ¼ mj exp½b0 x; per aj
1
aoaj ,
where aj ¼ log mj. This formulation allows the intercept of the log-hazard function to vary at different cut-off points. After examining life tables, we divided the age of the firm in the following six segments: Age 0–1 (0.5–1 years), Age 1–2 (1–2 years), Age 2–3 (2–3 years), Age 3–6 [3–6 years), Age 6– 10 [6–10 years) and Age 10–N (10-onward years). We then estimated the hazard rate (r) for organization i, as a function of a vector of firm’s characteristics and environmental variables (x). All the estimates were obtained using the software package STATA 8.
EMPIRICAL RESULTS Table 2 presents the maximum likelihood estimates of the piecewise exponential model of mortality rates for 643 producers during the period from 1899 to 1993. Benchmark Model 1 contains control and niche-width variables only, hence revealing the effect of Scope Economies (Hypothesis 1). Model 2 shows, in addition to this benchmark, the estimates of the location in the market center – i.e., the impact of Scale Economies (Hypothesis 2). Model 3 introduces the main effect of Population-Level Product Homogeneity through its linear and squared term (Hypothesis 3). Subsequently, Models 4 and 5 add the interactions of the Scale Economies and Scope Economies variables with the measure of Population-Level Product Homogeneity, respectively (Hypotheses 4 and 5). Finally, Model 6 presents the full specification. Below, we will discuss the results of Models 1–5. Note that the
Variables
Maximum Likelihood Estimates of Piecewise Exponential Model for Exit Rates of UK Motorcycle Producers in 1899–1993. Model 1 Std. Err.
Std. Err.
0.15 0.29 0.33 0.86 0.86 0.88 0.88 0.88 0.88 0.16 0.09 0.08 0.13 0.00006 0.12
0.21 0.67 0.37 6.03 6.25 6.25 6.53 6.35 6.75 0.63 0.61 0.26 0.60 0.000088 0.46
0.15 0.30 0.33 0.88 0.87 0.89 0.89 0.89 0.89 0.16 0.09 0.08 0.14 0.000066 0.12
0.11 0.0033 0.0007 0.006 0.00006 0.26
0.88 0.0149 0.0018 0.011 0.00015 0.44 0.30
0.11 0.0039 0.0007 0.006 0.00007 0.27 0.09
Model 3 Std. Err. 0.34 0.64 0.03 2.14 2.33 2.32 2.64 2.49 2.94 0.69 0.62 0.30 0.29 0.000057 0.36 0.92 0.0143 0.0014 0.018 0.00024 0.52 0.27 21.57 40.24
903.3
Note: Two-tailed t-tests. Standard errors are in parentheses. po0.10. po0.05.
891.1
0.15 0.31 0.29 1.09 1.09 1.10 1.10 1.09 1.09 0.15 0.10 0.08 0.14 0.000065 0.13 0.11 0.0042 0.0007 0.007 0.00008 0.27 0.09 4.98 12.36
Model 4 Std. Err. 0.33 0.62 0.06 2.61 2.80 2.79 3.10 2.94 3.44 0.69 0.62 0.30 0.27 0.000058 0.37 0.92 0.0144 0.0015 0.019 0.00025 3.21 0.25 20.51
0.15 0.31 0.29 1.15 1.15 1.16 1.15 1.15 1.15 0.15 0.10 0.08 0.15 0.000064 0.13 0.11 0.0042 0.0006 0.007 0.00008 2.45 0.09 5.34
42.69
13.33
18.45
11.08
889.8
Model 5 Std. Err. 0.33 0.64 0.04 2.61 2.81 2.79 3.11 2.96 3.43 0.69 0.62 0.30 0.29 0.000057 0.36 0.92 0.0144 0.0014 0.018 0.00024 0.53 0.48 19.81
0.15 0.30 0.29 1.19 1.19 1.20 1.19 1.19 1.19 0.15 0.10 0.08 0.15 0.000064 0.13 0.11 0.0042 0.0007 0.007 0.00008 0.27 0.62 5.30
Model 6 Std. Err. 0.33 0.62 0.06 2.71 2.91 2.89 3.20 3.05 3.54 0.70 0.62 0.30 0.28 0.000058 0.37 0.92 0.0144 0.0015 0.019 0.00025 2.77 0.02 20.01
0.15 0.31 0.29 1.20 1.20 1.21 1.20 1.20 1.20 0.15 0.10 0.08 0.15 0.000064 0.13 0.11 0.0042 0.0006 0.007 0.00008 2.93 0.75 5.52
42.35
12.75
42.86
13.40
3.26 890.5
2.94
17.26 1.13 889.5
11.40 3.55
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
WWI 0.17 WWII 0.70 Post-1958 0.43 Age 0–1 6.06 Age 1–2 6.32 Age 2–3 6.32 Age 3–6 6.61 Age 6–10 6.43 Age 10–8 6.82 DeIpso 0.62 DeAlio 0.60 District 0.26 GDP per Person 0.59 MES 0.00008 Total Motorcycle Sales 0.47 (log) Size (log) 0.89 Niche Overlap 0.0124 Market Width 0.0017 Density 0.009 Density2 0.00015 Scope Economies 0.75 Scale Economies Population-Level Product Homogeneity (PH) Population-Level Product Homogeneity2 Scope EconomiesPH Scale EconomiesPH Log-likelihood 906.9
Model 2
540
Table 2.
Scale and Scope Economies in the British Motorcycle Industry
541
overall pattern of findings is further confirmed by the full model specification’s estimates – i.e., Model 6. The estimates for the control variables in Model 1 are in line with our expectations. Among these, it is worth mentioning that the force of competition due to organizational similarity (even after discounting the effect of multi-market contacts) is mirrored in the significant and positive coefficient of the Niche Overlap variable. The estimates obtained for the main effects of Scale Economies and Scope Economies offer support for our Hypotheses 1 and 2 (Models 2 and 3). Adding these variables to our benchmark Model 1 significantly improves the fit of our model. For one, broad product portfolio producers enjoy a significantly lower hazard of exit (Hypothesis 1). For example, at the mean value of Scope Economies observed in our sample, the hazard of failure decreases by approximately 16 percent [exp( 0.806n21.42)]. Moreover, Scale Economies have an equally significant effect of lowering the risk of mortality (Hypothesis 2). On average, an organization offering products in the market center exhibits a decrease in the likelihood of failure by about 41 percent less, vis-a`-vis those firms not offering products in that segment of the market [exp(-0.36)]. Hence, both Hypothesis 1 and Hypothesis 2 are supported. By controlling for Scale Economies whilst measuring the impact of Scope Economies, we are able to disentangle the impact of both potential sources of survival advantages. Given lack of a fully blown theory, we did not formulate a hypothesis about the relative effect of scale versus scope economies. However, Model 3 reveals that, on average, the survival advantage largely comes from market center location (Scale Economies) rather than multiple-niche position (i.e., Scope Economies). In effect, after adding our proxy for Scale Economies, the estimate of the Scope Economies coefficient fails to reach statistical significance. Even when adopting different operationalizations of the latter variable (e.g., scaling the number of firm niches by the average number of niches of competitors or using a non-standardized measure), we found the same pattern of results.1 Needless to say, this implies explorative and preliminary evidence only that might well be specific for our British motorcycle industry setting. Future research must examine the robustness of this result, revealing the contingencies that determine the dominance of either scale or scope economies. The addition of the Population-Level Product Homogeneity variable in Model 4 significantly improves the fit of our model to the data (w2 [L3/ L2] ¼ 24.36 with 2 df and po0.001). The parameter estimates indeed reveal the predicted U-shaped effect of Population-Level Product Homogeneity on
542
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
mortality (similar to that of density), which reaches its minimum well within the range observed. As a consequence, Hypothesis 3 receives support. As far as the interaction Hypotheses 4 and 5 are concerned, the estimates obtained in our Models 4 and 5 for the interaction terms Scope Economies* Population-Level Product Homogeneity and Scale Economies* PopulationLevel Product Homogeneity only partly support our argument. On the one hand, as expected, the negative effect of Scope Economies on mortality is reinforced as the level of aggregate (industry-level) Population-Level Product Homogeneity increases. On the other hand, the estimate for the interaction with Scale Economies, although in the right direction, is far from reaching statistical significance. In line with Hypothesis 5 but in contrast with Hypothesis 4, therefore, Scope Economies rather than Scale Economies are especially rewarding under conditions of crowding into the market center. Fig. 3 graphically illustrates this conclusion by comparing the multiplier of the failure rate of scope economies at different levels of population-level homogeneity. We interpret the lack of support for Hypothesis 4 with reference to the complex relationship between our measures of the market center and scale economies. Our proxy of scale economies could capture other than traditional
1.2
Multiplier of the mortality rate
1
mean hhi below mean hhi above mean hhi
0.8
0.6
0.4
0.2
1
0.96
0.92
0.88
0.8
0.84
0.76
0.72
0.68
0.6
0.64
0.56
0.52
0.48
0.4
0.44
0.36
0.32
0.28
0.2
0.24
0.16
0.12
0.08
0
0.04
0
Relative_scope_economies
Fig. 3.
Interaction Between Scope Economies and Population-Level Product Homogeneity.
Scale and Scope Economies in the British Motorcycle Industry
543
scale-based benefits associated with a market center location as well – e.g., spillover effects of legitimacy and reputation. Although we trust our measure to match the theoretical construct discussed (see the methods section on this), it is likely that the above non-scale spillover benefits of clustering may partly obscure those related to crowding in product space. In future research, we hope to be able to include separate measures for the different pros and cons of market center location.
CONCLUSIONS Scale and scope economies are widely recognized as potential sources of competitive advantage in the strategic management literature (see Holcomb and his colleagues 2006). We add to this body of research by developing a theory and performing an initial empirical test in the context of an organizational ecology framework. In so doing, a threefold contribution is provided. First, theorizing on organizational ecology requires cost-based arguments. Scale and (limited) scope economies are central to understanding generalist consolidation in resource-partitioning theory (Carroll, 1985), and scope economies matter for the niche-width theory’s argument as to the survival advantage of the generalist form under coarse-grained environmental changes (Freeman & Hannan, 1983). By now, organizational ecology should start benefiting from developing a more precise cost-based argument, integrating insights from strategic management into its selection arguments (see also Boone & van Witteloostuijn, 2004; Wezel & van Witteloostuijn, 2004; van Witteloostuijn & Boone, 2006). The current paper makes an attempt to deepen our understanding of the survival consequences of scale and scope economies. Second, our findings underscore the impact of an under-explored environmental contingency: the aggregate distribution of organizations in the industry. Conversely to a static perspective on strategic differentiation (see, e.g., Deephouse, 1999), we develop the key argument that a focal organization’s (survival) performance, and the role of scale and scope economies, depend dynamically upon the position of this organization vis-a`-vis all of its competitors in the industry’s product space. Introducing a logic that builds on arguments from niche-width and resource-partitioning theories, we hypothesize how the pros and cons of center and non-center locations – i.e., scale and scope economies – work out in an environment characterized by population-level product homogeneity or heterogeneity. In this respect, our findings are consistent to those obtained by Boone et al. (2005), who found
544
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
that the survival benefits of organizational similarity decrease at high levels of population-level homogeneity. Indeed, we report evidence that the beneficial impact of scope economies (i.e., here defined as a multi-product offering outside the market center) is reinforced in circumstances of high population-level homogeneity. In so doing, we believe to add theoretical precision to the traditional Porterian logic as to the superiority of specific strategies under different rules of the competitive game (Boone & van Witteloostuijn, 2004; Boone et al., 2005). Third, our conception of the organizational search for profits through scale and scope economies is complementary to that of strategic management scholars. By building on similar assumptions (i.e., market imperfections and incomplete information) both approaches stress the importance of organizational positioning in product space for long-term survival. For instance, Holcomb and his colleagues (2006) argue that firms purposively change organizational scale and scope in search for competitive advantage, and adapt to the dynamics in the environment while doing so. This search process is driven by the accumulation of knowledge and capabilities (through scale and scope economies) that are then integrated and recombined to move to more favorable market positions. Our analysis differs, though, because we remain silent on whether or not firms are able to proactively move to favorable positions. In effect, in line with the ecological research, a few extra analyses confirmed that market position changes are conducive to organizational mortality. Needless to say, this finding does not exclude that managers purposively scan the environment in search for market imperfections they think they can benefit from, but rather suggests that the process leading to such positional changes is hazardous. Given these results, the question then becomes whether the accumulated resources and competencies by the focal firm are sufficient to overcome the risks of repositioning. Recent findings in ecological research indeed suggest that this might well be the case. Kim, Dobrev, and Solari (2003) and Wezel and van Witteloostuijn (2006) show how the sharp increase in mortality due to product portfolio expansions (in the first case measured with a dummy, and in the second study by adopting a continuous variable) is significantly reduced by the organization’s portfolio width, which may be seen as proxy of firm’s capabilities. Such findings will hopefully stimulate more cross-fertilization between strategic management and organization ecology. Of course, our study is far from perfect. Empirically, a key limitation is that, due to lack of information, we were unable to collect data on nichespecific demand. Similarly, standard size-based measures at the individual
Scale and Scope Economies in the British Motorcycle Industry
545
and at the aggregate level could not be included. As a result, our proxy for scale economies is probably rather noisy. In future research, we would like to improve upon such issues of measurement. Only then, we can systematically estimate the interaction and importance of product-based (our product space concentration logic) and size-based (e.g., resource-partitioning theory’s concentration argument) mechanisms. Theoretically, we believe that the general approach taken here – introducing industrial organization and strategic management logic into an otherwise organizational ecology framework – is very promising indeed. An early example of this approach was Boeker’s (1989) study of strategic groups, and Amburgey and Rao (1996) already convincingly launched a plea for such an approach. Our study is yet another example of the strength of such a tradition.2
NOTES 1. In a robustness check, we tested for the existence of an interaction between scale and scope with industry age. None of the two interactions, however, turned out to be statistically significant. 2. From the side of strategic management scholars, the argument oftentimes was and still is that organizational ecology’s inertia assumption is a fundamental hurdle for meaningful integration. This view does injustice to organizational ecology’s much more sophisticated and subtle argument, which is clearly witnessed by organizational ecology’s progress over the past three decades or so (see, e.g., the many organizational change effect studies).
ACKNOWLEDGMENT We would like to thank Joel Baum and Stanislav Dobrev for their insightful comments. Of course, all remaining errors are ours.
REFERENCES Allison, P. D. (1984). Event history analysis: Regression for longitudinal event data. Beverly Hills, CA: Sage. Amburgey, T. L., & Rao, H. (1996). Organizational ecology: Past, present, and future directions. Academy of Management Journal, 39, 1265–1286. Amit, R., & Schoemaker, P. (1993). Strategic assets and organizational rent. Strategic Management Journal, 14, 333–346.
546
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
Barnett, W. P., & McKendrick, D. (2001). The organizational evolution of global technological competition. Research Paper # 1682. Stanford University: Graduate School of Business. Bain, J. S. (1956). Barriers to new competition. Cambridge: Harvard University Press. Baum, J. A. C., & Korn, H. J. (1996). Competitive dynamics of interfirm rivalry. Academy of Management Journal, 39, 255–291. Baum, J. A. C., & Mezias, S. (1992). Localized competition and organizational failure in the Manhattan hotel industry, 1898–1990. Administrative Science Quarterly, 37, 580–604. Baum, J. A. C., & Singh, J. V. (1994a). Organizational niche overlap and the dynamics of organizational founding. Organization Science, 5, 483–502. Baum, J. A. C., & Singh, J. V. (1994b). Organizational niche overlap and the dynamics of organizational mortality. American Journal of Sociology, 100, 346–380. Baumol, J. W., Panzar, J. C., & Willig, R. D. (1982). Contestable markets and the theory of industry structure. New York: Harcourt Brace Jovanovich. Bishop, G., & Barrington, S. (1995). Encyclopedia of motorcycling. London: Bison Group. Boeker, W. (1989). Strategic change: The effects of founding and history. Academy of Management Journal, 32, 489–515. Boone, C., Bro¨cheler, V., & Carroll, G. R. (2000). Custom service: Application and tests of resource partitioning among Dutch auditing firms from 1880 to 1982. Organization Studies, 21, 355–382. Boone, C., Carroll, G. R., & van Witteloostuijn, A. (2002). Resource distributions and market partitioning: Dutch daily newspapers, 1968 to 1994. American Sociological Review, 67, 408–431. Boone, C., Carroll, G. R., & van Witteloostuijn, A. (2004). Size, differentiation and the performance of Dutch daily newspapers. Industrial and Corporate Change, 13, 117–148. Boone, C., Wezel, F. C., & van Witteloostuijn, A. (2005). An ecological theory of populationlevel diversity dependence. Working Paper. Tilburg University, Tilburg. Boone, C., & van Witteloostuijn, A. (2004). Towards a unified theory of market partitioning: Integrating resource partitioning and sunk cost theories of dual market structures. Industrial and Corporate Change, 13, 701–726. Burgess Wise, D. (1973). Historic motorcycles. New York: Hamlyn Publishing. Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in populations of organizations. American Journal of Sociology, 90, 1262–1283. Carroll, G. R., Bigelow, L. S., Seidel, M. D., & Tsai, L. B. (1996). The fates of de novo and de alio producers in the American automobile industry, 1885–1982. Strategic Management Journal, 17, 117–138. Carroll, G. R., Dobrev, S., & Swaminathan, A. (2002). Organizational processes of resource psartitioning. Research in Organizational Behavior, 24, 1–40. Carroll, G. R., & Hannan, M. T. (2000). The demography of corporations and industries. Princeton, NJ: Princeton University Press. Carroll, G. R., & Swaminathan, A. (2000). Why the microbrewery movement? Organizational dynamics of resource partitioning in the US brewing industry. American Journal of Sociology, 106, 715–762. Cohn, E., Rhine, S. L., & Santos, M. C. (1989). Institutions of higher education as multiproduct firms: Economies of scale and scope. Review of Economics and Statistics, 71, 284–290. Collins, P. (1998). British motorcycle since 1900. Surrey: Ian Allen Transport Library.
Scale and Scope Economies in the British Motorcycle Industry
547
Deephouse, D. L. (1999). To be different, or to be the same? It’s a question (and theory) of strategic balance. Strategic Management Journal, 20, 147–166. Deller, S. C., Chicoine, D. L., & Walzer, N. (1988). Economies of size and scope in rural lowvolume roads. Review of Economics and Statistics, 70, 459–465. Dobrev, S. D., Kim, T., & Solari, L. (2004). The two sides of the coin: Core competence as capability and obsolescence. In: J. A. C. Baum & A. M. McGahan (Eds), Advances in strategic management (Vol. 21, pp. 255–285). Greenwich CT: JAI Press. Dobrev, S. D., Kim, T.-Y., & Carroll, G. R. (2002). The evolution of organizational niches: U.S. automobile manufacturers 1885–1981. Administrative Science Quarterly, 47, 233–264. Dobrev, S. D., Kim, T.-Y., & Hannan, M. T. (2001). Dynamics of niche width and resource partitioning. American Journal of Sociology, 106, 1299–1337. Dowell, G., & Swaminathan, A. (2000). Racing and back-pedalling into the future: New product introduction and organizational mortality in the US bicycle industry, 1880–1918. Organization Studies, 21, 405–431. Fournier, G. M., & Mitchell, J. M. (1992). Hospital costs and competition for services: A multiproduct analysis. Review of Economics and Statistics, 74, 627–634. Freeman, J., & Hannan, M. T. (1983). Niche width and the dynamics of organizational populations. American Journal of Sociology, 88, 1116–1145. Hannan, M. T., & Freeman, J. (1977). The population ecology of organizations. American Journal of Sociology, 83, 929–984. Hawley, A. (1950). Human ecology. New York: Ronald Press. Henderson, B. D. (1979). Henderson on corporate strategy. Cambridge, MA: Abt Books. Holcomb, T. R., Holmes, R. M., & Hitt, M. A., Jr. (2006). Diversification to achieve scale and scope: The strategic implications of resource management for value creation. In: J. A. C. Baum, S. D. Dobrev & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 549–587). Oxford, UK: Elsevier. Hume, W. (1991). The register of machines of the VMCC (3rd ed.). London: The Vintage Motor Cycle Club. Kim, T.-Y., Dobrev, S. D., & Solari, L. (2003). Festina lente: Learning and inertia among Italian automobile producers, 1896–1981. Industrial and Corporate Change, 12, 1279– 1301. Lieberman, M. B. (1987). The learning curve, diffusion and competitive strategy. Strategic Management Journal, 8, 441–452. Mata, J., & Portugal, P. (1994). Life duration of new firms. Journal of Industrial Economics, 42, 227–246. Mata, J., & Portugal, P. (2000). The survival of new domestic and foreign-owned firms. Strategic Management Journal, 23, 323–343. McPherson, J. M. (1983). An ecology of affiliation. American Sociological Review, 48, 519–535. Mitchell, B. R. (1998). Abstract of British historical statistics. Cambridge, UK: Cambridge University Press. Murray, J. D., & White, R. W. (1983). Economies of scale and economies of scope in multiproduct financial institutions: A study of British Columbia credit unions. Journal of Finance, 38, 887–902. Panzar, J. C., & Willig, R. D. (1981). Economies of scope. American Economic Review, 71, 268–272.
548
FILIPPO C. WEZEL AND ARJEN VAN WITTELOOSTUIJN
Petersen, T. (1991). Time aggregation bias in continuous-time hazard-rate models. In: P. Marsden (Ed.), Sociological methodology (pp. 290–293). Oxford, UK: Blackwell. Podolny, J. M., Stuart, T., & Hannan, M. T. (1996). Networks, knowledge, and niches: Competition in the worldwide semiconductor industry, 1984–1991. American Journal of Sociology, 102, 659–689. Sorenson, O. (2000). Letting the market work for you: An evolutionary perspective on product strategy. Strategic Management Journal, 21, 577–592. Tragatsch, E. (1977). The complete illustrated encyclopedia of world’s motorcycles. London: Quantum Publishing. Tragatsch, E. (2000). The new illustrated encyclopedia of motorcycles. Edison, NJ: Quantum Publishing. Tuma, N. B., & Hannan, M. T. (1984). Social dynamics: Models and methods. New York: Academic Press. Venkatraman, N., & Subramaniam, M. (2002). Theorizing the future of strategy: Questions for shaping strategy research in the knowledge economy. In: The handbook of strategy and management (pp. 461–474). London: Sage Publications. Wezel, F. C. (2005). Location-dependence and industry evolution: Founding rates in the United Kingdom motorcycle industry, 1895–1993. Organization Studies, 26, 729–754. Wezel, F. C., & van Witteloostuijn, A. (2004). The ecology of niche-width strategies: Generalists, polymorphists and specialists in the UK motorcycle industry from 1895 to 1993. Working paper. Tilburg University, Tilburg. Wezel, F. C., & van Witteloostuijn, A. (2006). Portfolio width and expansion, and organizational failure: Evidence from the British motorcycle industry in 1895 to 1993. Long Range Planning, 39, 11–28. Wholey, D., Feldman, R., Christianson, J. B., & Engberg, J. (1996). Scale and scope economies among health maintenance organizations. Journal of Health Economics, 15, 657–684. Wilson, H. (1993). The ultimate motorcycle book. New York: Darlin Kindersley. van Witteloostuijn, A., & Boone, C. (2006). A resource-based theory of market structure and organizational form. Academy of Management Review, 31, 409–426.
DIVERSIFICATION TO ACHIEVE SCALE AND SCOPE: THE STRATEGIC IMPLICATIONS OF RESOURCE MANAGEMENT FOR VALUE CREATION Tim R. Holcomb, R. Michael Holmes Jr. and Michael A. Hitt ABSTRACT Research on diversification has produced insights into possible linkages between organizational scale and scope and firm performance. However, the paucity of research on strategy implementation has hindered our understanding of the broader performance implications of diversification. We extend the resource-based view and diversification research by examining how firms can exploit diversifying investments designed to achieve scale and scope economies. Successful firms more effectively structure their resource portfolio, bundle resources into capabilities, and leverage these capabilities when implementing a diversification strategy. We develop a model linking strategies by which firms expand product and geographic market scope to the actions they take to manage resources. We examine three actions – internal development, acquisitions, and strategic Ecology and Strategy Advances in Strategic Management, Volume 23, 549–587 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23017-1
549
550
TIM R. HOLCOMB ET AL.
alliances – and discuss the implications of these actions using the resource management framework.
INTRODUCTION Research in strategic management has long focused on the origins of performance differences among firms (Powell, 2001). Early research on the strategy/structure model addressed the relationship between corporate strategy, organizational structure, and performance (e.g., Chandler, 1962; Lawrence & Lorsch, 1967). This body of research suggests that although strategy alone or structure alone may explain some of the variation in firm performance, differences in firms’ performances are determined more by the nature of the strategy/structure fit (Yin & Zajac, 2004). Scholars studying the relationship between organizational scale and scope and firm performance have learned that a firm’s structure is both influenced by and influences diversifying actions to achieve scale and scope. However, despite decades of research, a complete understanding of the performance implications of changes to scale and scope through diversification remains elusive. One important shortcoming in the literature is that little research has addressed how diversifying firms implement the fit between strategy and structure to realize superior performance. Stated differently, research on diversification has not adequately addressed the means through which firms exploit the potential scale and scope advantages generated by diversification. This is unfortunate because scholars have long argued that the success of diversification depends on how it is implemented (Chandler, 1962; Ramanujam & Varadarajan, 1989; Hoskisson & Hitt, 1990). This relative neglect of implementation reflects the broader trend for many strategic management scholars to study strategy formulation rather than implementation (Hrebiniak & Joyce, 2001). As such, the lack of consensus on the implementation of diversification strategies may be more contentious than disagreements about the strategies alone (cf. Bourgeois, 1980). In this chapter, we consider the implementation challenges of diversification and seek a balance in focus between content and process. We refer to diversification content as the actual changes derived from firm strategies to expand into new product and geographic markets. Diversification process refers to the means by which firms change and exploit the potential scale and scope advantages achieved through such strategies. We draw on recent research from the resource-based view of the firm (RBV) to examine the implementation of diversification strategies. The RBV
Diversification to Achieve Scale and Scope
551
suggests that possessing valuable and rare resources provides the basis for value creation. However, merely owning resources does not guarantee superior performance (Priem & Butler, 2001). To realize superior performance, resources must be managed effectively (Sirmon, Hitt, & Ireland, 2007). Following Priem and Butler’s (2001) critique suggesting that the RBV was static and paid inadequate attention to the process and context in which resources are managed, recent work on RBV has sought to examine the ways in which resources can be successfully managed to create value (e.g., Sirmon et al., 2007). Resource management involves routines and processes associated with structuring the firm’s resource portfolio, bundling resources to form capabilities, and leveraging those capabilities to create and maintain value for customers and owners (Sirmon et al., 2007). Given the applicability of the RBV to diversification research (Porter, 1991; Peteraf, 1993; Bergh, 2001) and the process-oriented nature of the resource management literature, we believe this research can contribute to our understanding of diversification strategy implementation. Building on the RBV, we argue that a firm’s ability to exploit the content of diversification to realize superior performance depends on implementation processes, especially those associated with resource management. This work makes at least two contributions to the literature. First, we integrate resource management and diversification research to examine how successful firms implement their diversification strategies. In doing so, we extend not only the theory of resource management, but also the theoretical base of diversification strategies. Second, we examine resource management in the context of different strategic actions designed to achieve scale and scope. Strategic actions – such as internal development, acquisitions, and alliances – provide a context for examining the performance implications of diversification under different organizational and environmental conditions. We propose that superior resource management is necessary to exploit the potential scale and scope advantages these actions provide. The proposed theoretical model is depicted in Fig. 1. In the following section, we review prior work that examines linkages between product and international diversification and performance outcomes. Next, we discuss resource management as a means for firms to achieve superior performance through diversification. We then examine diversifying actions that create scale and scope and utilize the resource management framework to explain how these actions can be successfully implemented. We integrate insights from the organizational ecology literature in the arguments presented. Finally, we conclude by discussing implications of this work and future research extensions.
552
Environmental and Competitive Conditions (External Contingencies)
Fig. 1.
Product and International Diversification Strategies to Create Scale and Scope (Strategy Formulation)
Resource Management to Exploit Scale and Scope (Strategy Implementation)
Sustained Competitive Advantage and Economic Return (Value Creation)
Linking Diversification Strategies with Resource Management to Create Value. TIM R. HOLCOMB ET AL.
Diversification to Achieve Scale and Scope
553
DIVERSIFICATION IN PURSUIT OF SCALE AND SCOPE Strategic management researchers commonly examine performance variations as a function of firm behavior. Recent advances in strategic management research have recognized that environmental characteristics, such as uncertainty, munificence, and the competitive actions of rivals are both pervasive and dynamic (e.g., Gimeno, 1999; Wan & Hoskisson, 2003; Knott, 2003), which makes sustaining the benefits of organizational scale and scope over time especially difficult. Instead, many firms attempt to remain responsive to the changing competitive landscape, seeking to maintain a series of temporary competitive advantages (Eisenhardt, 1999; Knott, 2003). Thus, strategic management scholars frequently emphasize the need for firm strategies to change in response in shifting environmental contexts. Insights from the organizational ecology literature contribute to our discussion of scale and scope in at least two related ways. First, ecology research suggests the influence of environmental factors over which firms have limited control (Baum & Powell, 1995). These factors affect selection processes that enforce changes on organizations at the population level (Hannan & Freeman, 1977; Singh & Lumsden, 1990). Second, although some scholars now dispute the notion of structural inertia, work in this tradition continues to highlight the challenges of managing change (Haveman, 1993; Baum, 1996). For example, there is evidence that organizational momentum influences the strategic actions that firms pursue (Amburgey & Miner, 1992; Baum, Li, & Usher, 2000). Many internal and external barriers to change exist, resulting in long periods of incremental change punctuated by infrequent periods of more substantial change (Romanelli & Tushman, 1994; Henderson & Mitchell, 1997; Miller, 1993). Often, the disruptive effects of change can actually reduce rather than promote organizational viability (Singh, House, & Tucker, 1986; Miner, Amburgey, & Stearns, 1990; Amburgey, Kelley, & Barnett, 1993). These two insights suggest the possibility that organizational adaptation and environmental selection may be interdependent processes (Baum, 1996). In this chapter, we build on these concepts to examine both the implementation of diversifying strategies and the environmental context in which they occur to extend theoretical explanations of the performance implications of diversification. Both product and international diversification play key roles in the strategic behavior of firms (Hitt, Hoskisson, & Ireland, 1994). Product diversification refers to expansion into product markets that are new to the firm (Hitt, Hoskisson, & Kim, 1997). Changes to organizational scale and scope
554
TIM R. HOLCOMB ET AL.
through product diversification are reflected by expansion into multiple lines of business that fulfill different customer needs (Pitts & Hopkins, 1982). International diversification refers to expansion across national borders into different geographic locations to obtain new resources, exploit existing resources and capabilities, or both (Bartlett & Ghoshal, 1989; Zahra, Ireland, & Hitt, 2000). A firm’s level of international diversification is represented by the number of different markets in which it operates and their importance to the firm. Patterns of product and international diversification result from decisionmaking processes used to alter organizational scale and scope. Organizational scale refers to the size of a firm’s operations, resource endowments, production volume, and market power within a single line of business or corporate support activity. Scale economies are realized, for example, when an increase in production volume for a product reduces its average unit cost. Organizational scope generally refers to the range of products and business segments in which firms operate. Scope economies are present when resources shared across product and geographic markets result in a reduction of average costs. Scale and scope contribute to each firm’s unique resources, capabilities, and competitive positions and can allow firms to realize superior economic performance. However, merely possessing organizational scale and scope alone does not ensure superior performance. The literature suggests that economies are greater and performance is higher when firm activities are integrated in such a way that resources and value-adding routines are shared across multiple product and geographic markets (Panzar & Willig, 1981; Porter, 1985; Robins & Wiersema, 1995). Resource management facilitates such integration. We propose that diversification strategies that optimally structure the firm’s resource portfolio, bundle such resources into valuable and difficult-to-imitate combinations, and leverage these combinations across product-geographic markets can be a source of competitive advantage and value creation. In the following sections, we examine the linkage between diversification strategy and performance.
Product Diversification and Performance Product diversification has created conditions in which firms frequently confront a given rival in two or more segments. Multimarket competition is important because it influences both the motives and the outcomes of product diversification. When firms compete in multiple markets, competitive
Diversification to Achieve Scale and Scope
555
attacks in one market can often be countered with attacks in other markets, potentially reducing the motivation for firms to initiate attacks and increasing the options for retaliation (Edwards, 1955, Chen, 1996; Gimeno, 1999). Thus, firms that utilize product diversification to establish a presence in multiple markets may have an advantage over firms competing in a single market. Research on the performance implications of product diversification often invokes the concept of relatedness. Relatedness refers to the degree of commonality in the value creation functions of a firm’s different businesses (Markides & Williamson, 1996). The economic benefits of related diversification arise primarily from the ability of firms to exploit economies of scope (Teece, 1982; Jones & Hill, 1988). Indeed, firms that pursue related diversification often seek to extend resource advantages to new product segments and/or to exploit synergies across related segments to achieve superior performance (Hill, Hitt, & Hoskisson, 1992; Mahoney & Pandian, 1992; Robins & Wiersema, 1995). Examples of related segments include those that employ common distribution channels, engage in common advertising when products are compatible, or utilize common administrative processes. The sources of economic benefits from unrelated diversification lie in efficient and effective governance (Jones & Hill, 1988; Hill et al., 1992). Arguments for the economies of unrelated diversification are based on the proposition that labor, technology, and capital markets frequently suffer from opportunism and information asymmetries that create difficulties in the efficient allocation of resources among firms (Hill et al., 1992; Khanna & Palepu, 1997). It is argued that firms pursuing unrelated diversification can more optimally allocate such resources because common governance smoothes transactions between the corporate office and its business segments (Williamson, 1975, 1985). Nonetheless, despite the potential advantages of product diversification, its benefits are difficult to realize (Rumelt, 1974; Hoskisson & Hitt, 1990; Hitt et al., 1994). In a meta-analysis examining the effects of product diversification on firm performance, Palich, Cardinal, and Miller (2000) found consistent support for a curvilinear (inverse-U) relationship. Palich and his colleagues speculated that diversification efforts by firms during early stages of development are often made into related business segments, allowing firms to exploit synergies among similar products. However, as time passes, many firms begin to diversify into unrelated lines of business in which synergies are more elusive. When coordinating and monitoring multiple business units, firms confront transaction difficulties such as bounded rationality, uncertainty,
556
TIM R. HOLCOMB ET AL.
and complexity (Jones & Hill, 1988), which can neutralize the benefits of product diversification. As diversification activity continues, informationprocessing demands increase with growth in the number and diversity of product segments served, making multiproduct firms more difficult and costly to manage (Hitt et al., 1994; Sorenson, 2000; Bergh, 2001). In addition, increased product diversification may reduce innovation (Hoskisson & Hitt, 1988; Baysinger & Hoskisson, 1989). As the number of the firm’s product lines increases, information-processing demands and transactional difficulties are heightened. In response, managers often shift attention from strategic controls that incorporate long-term and strategically relevant criteria to financial controls that rely primarily on measures such as return on sales and return on investment (Hoskisson & Hitt, 1988). This creates a short-term and risk-averse orientation inside the firm that reduces its incentive to innovate (Hitt, Hoskisson, Johnson, & Moesel, 1996). Of course, this reduction is problematic because the long-term success of the firm often depends on its ability to innovate (Franko, 1989). With few exceptions, however, the authors of many prior studies have assumed away the environmental conditions within which diversification is pursued. Neglecting the environmental conditions that surround a firm’s actions is unfortunate because such conditions influence benefits available from scale and scope (Khanna & Palepu, 1997; Peng, 2001). Wan and Hoskisson’s (2003) examination of corporate diversification and home country diversification is a notable exception. Drawing on arguments from institutional economics, these authors examined the effects of munificence in home country environments on corporate diversification and found that product diversification actions had a negative effect on firm performance in relatively munificent market environments. Conditions prevalent where markets are more munificent discourage actions that deplete managers’ capacities to understand the firm’s different business units. When the environment in less munificent, there is a greater need for the internal markets diversification provides, thereby making product diversification a more attractive option (Khanna & Palepu, 1997; Wan & Hoskisson, 2003). Organizational ecology research has emphasized some of the benefits of diversification. For example, in this volume, Wezel and Witteloostuijn (2006) shed light on the degree to which industry concentration affects scale and scope economies. Using the resource partitioning and niche width theories, these authors demonstrate the survival consequences of scale and scope economies for firms based on the population-level distribution of rivals across an industry. In particular, survival chances in concentrated markets erode at increasing levels of concentration, allowing firms to
Diversification to Achieve Scale and Scope
557
achieve economies by diversifying into other markets niches. Thus, diversification allows a firm to generate scope economies and also differentiates the firm’s competitive position relative to rivals, thereby avoiding some of the negative effects of especially high levels of competition.
International Diversification and Performance Similar to product diversification, international diversification offers several advantages to firms that may affect their performance. For example, international diversification allows firms to take advantage of foreign market opportunities and imperfections, both in the exploitation of current resources and capabilities and in the search for new resources (Rugman, 1981). Thus, international diversification may enable firms to realize cost benefits achieved through economies of scale and scope (Caves, 1996). There is also recent evidence that international diversification may be driven by the global competitive moves of multimarket rivals (Gimeno, Hoskisson, Beal, & Wan, 2005). By capitalizing on market imperfections across geographic boundaries, firms with a presence in multiple markets may have advantages over firms operating exclusively in a single (i.e., domestic) market. Indeed, the scale and scope advantages that diversification provides may reduce the motivation for rival firms to compete aggressively (Chen, 1996; Gimeno, 1999; Gimeno & Woo, 1999) because the prospect of an advantage in one market may prompt retaliatory actions by rivals in other markets (Bulow, Geanakoplos, & Klemperer, 1985). In examining the international diversification-performance link, some studies have found a linear positive relationship (Errunza & Senbet, 1984; Delios & Beamish, 1999). Other research has found no linear effect of international diversification on firm performance (Morck & Yeung, 1991). Still others have found a negative relationship (Denis, Denis, & Yost, 2002). These results suggest that, although international diversification may offer several benefits to firms, the scale and scope advantages it creates may be especially difficult to manage. For example, escalating internationalization greatly increases coordination, distribution, and communication costs, and may also increase information-processing demands on the firm’s managers (Hitt et al., 1997). Internationalization exposes the firm to a diverse set of government policies, cultures, market requirements, and economic conditions. Internationally diversified firms therefore operate in several environmental contexts simultaneously, some of which may make conflicting demands. Thus, at high
558
TIM R. HOLCOMB ET AL.
levels, the added complexity of managing international diversification may nullify many of its benefits (Hitt, Bierman, Uhlenbruck & Shimizu, 2006). Recently, scholars have started to more explicitly consider the costs of internationalization and have examined a curvilinear relationship between international diversification and performance (Hitt et al., 1997; Lu & Beamish, 2001). In one of the earlier studies to explore this possibility, Hitt et al. (1997) found an inverted-U relationship between international diversification and firm performance. These scholars argued that early efforts to diversify internationally often produce positive economies of scope, scale, and experience. However, as diversification activity continues, performance benefits decrease with increased integration and coordination requirements and can turn negative at higher levels of internationalization. In contrast, Lu and Beamish (2001), studying a sample of small- and medium-sized enterprises (SMEs), argued and found a U-shaped relationship between international diversification and performance. These results suggest that international expansion initially hurts performance. Beyond a threshold, however, performance improves as experience provides opportunities for learning by doing, potentially allowing firms to develop the knowledge and capabilities required to manage international operations more successfully. Attempting to reconcile these earlier studies (e.g., Hitt et al., 1997; Lu & Beamish, 2001), Lu and Beamish (2004) theorized and found that the relationship between international diversification and firm performance was S-shaped. When firms expand into international markets, they confront a liability of foreignness (Zaheer, 1995). The organization ecology literature sheds some light on the shape of this relationship. It suggests that SMEs confront a liability of newness, with many also facing a liability of smallness (Stinchcombe, 1965; Hannan & Freeman, 1977; Baum, 1996). The joint effects of these liabilities can create significant costs that reduce firm performance at early stages of internationalization. As time passes, however, organizational structures develop, which allows some firms to overcome such liabilities. With the effects of these liabilities reduced, growing international diversification can lead to improvements in firm performance. At especially high levels of international diversification, however, the information processing and transaction difficulties associated with managing the international diversified firm reduce firm performance by creating diseconomies, coordination challenges, and lost flexibility. Additionally, interorganizational relationships may play an important role in helping firms achieve the benefits of international diversification. Ecologists and institutional theorists suggest that interorganizational relationships help legitimate organizational structures (Baum & Oliver, 1991).
Diversification to Achieve Scale and Scope
559
Combining perspectives from organizational ecology and institutional theory, Baum and Oliver (1996) found that interorganizational linkages confer both legitimacy and resources thereby enhancing the probability of survival by small and young organizations. Similarly, strategic management and international business research suggests that alliances with local firms in a foreign country help firms to overcome the liability of foreignness (Eden & Miller, 2004). Supporting this view, Lu and Beamish (2001) found that SMEs entering international markets through partnerships with local firms performed better than SMEs entering such markets through partnerships with firms from the home country. Other strategic management scholars have suggested that the benefits of such interorganizational relationships are contingent on internal and external conditions such as the firm’s product– market strategy and environmental demands for exploration and exploitation (Rowley, Behrens, & Krackhardt, 2000; Echols & Tsai, 2005). In addition, there is evidence that a firm’s current stock of resources can influence the success of international diversification. Intangible resources, such as technological know-how, management skills, brands, and goodwill may be especially influential (Lu & Beamish, 2004). For example, firms with an advanced stock of knowledge-intensive resources may have the absorptive capacity to capitalize on exploration opportunities in international markets (Hitt et al., 2006). Similarly, internationalization provides opportunities for firms to exploit their resources (Caves, 1996). For example, global market imperfections in the trade of intangible assets often enhance the value of such assets in foreign markets (Morck & Yeung, 1991; Kotabe, Srinivasan, & Aulakh, 2002). Given the costs of developing intangible resources and their potential as public goods (Caves, 1971), optimal returns to the exploitation of these resources is most likely when their appropriability regime is greatest (Teece, 1986). Thus, maximizing the returns to intangible assets may require deploying them into multiple geographic markets (Lu & Beamish, 2004). As this discussion suggests, building a multinational firm is a highly complex task (Tallman & Yip, 2001). However, an additional layer of complexity is introduced when firms pursue both product and international diversification concurrently.
Concurrent Product and International Diversification and Performance Many internationally diversified firms also operate in multiple product markets in an attempt to enhance the benefits of both forms of diversification
560
TIM R. HOLCOMB ET AL.
(Hitt et al., 1994). However, pursuing both strategies concurrently amplifies the complexity of managing the diversified firm. Indeed, there is evidence that product diversification negatively moderates the relationship between international diversification and performance (Vermeulen & Barkema, 2002; Wan & Hoskisson, 2003). Combining high levels of both product and international diversification also exposes the firm to diverse rivals and potentially intense competition across product and geographic markets, creating demands that make it especially challenging to coordinate the firm’s activities. Managers in firms with high levels of both product and international diversification may therefore have difficulty developing and maintaining the absorptive capacity to understand and adapt to the requirements of diverse competitive environments (Vermeulen & Barkema, 2002). When the simultaneous pursuit of diversification involves unrelated product diversification and/or institutionally and culturally distant international diversification, firms encounter additional complexities and costs and may compensate by underinvesting in innovation (Hitt et al., 1997). Indeed, few firms have the resources and capabilities to manage high levels of either product or international diversification simultaneously, much less the simultaneous pursuit of both types of diversification (Barkema & Vermeulen, 1998). However, other theory and evidence suggest that the simultaneous pursuit of product and international diversification can enhance firm performance. For example, Hitt and his colleagues (1997) found that product diversification positively moderated the relationship between international diversification and firm performance. Evidence presented by Kim, Hwang, and Bergers (1989) suggests that this effect is not contingent on whether firms pursue primarily related or unrelated diversification. One possibility is that international diversification allows the product-diversified firm to develop the scale necessary to drive down costs (Hitt et al., 1994). Scope economies may also emerge through the sharing of resources across international markets (Hitt et al., 1997). Additionally, international diversification is sometimes managed through multiple business units similar to those utilized by product-diversified firms. Thus, product diversification may provide valuable learning experiences that a firm can apply to its international diversification efforts (Tallman & Yip, 2001). Indeed, product-diversified firms may have built structures, routines, and processes necessary to share and combine knowledge and other resources across business segments, thereby enabling these firms to perform such operations more effectively across international markets.
Diversification to Achieve Scale and Scope
561
MANAGING SCALE AND SCOPE TO CREATE VALUE As the above review implies, organizational scale and scope resulting from diversification strategies does not ensure performance gains. Nevertheless, there are firms for which developing scale and scope through diversification does improve performance. Resource-based theory argues that interrelationships between resources held across business segments provide a foundation for the diversified firm (Robins & Wiersema, 1995). Diversification strategies are partly determined by the incentive to find new resources or additional uses for existing resources not fully utilized in existing product and geographic markets (Penrose, 1959; Gimeno & Woo, 1999; Zahra et al., 2000). However, superior performance may not materialize if rivals are motivated to compete aggressively and are able to implement strategies that create similar resources combinations (Bulow et al., 1985; Markides & Williamson, 1996; Gimeno, 1999). Although establishing multimarket contact may act as a deterrent to rivals (Gimeno & Woo, 1999), creating persistent heterogeneity relative to rivals may require utilizing either new or old resources in novel ways (Black & Boal, 1994; Morrow, Sirmon, Hitt, & Holcomb, 2005). Each business in a firm’s portfolio inherently influences the profitability of other business units within that portfolio (Bowman & Helfat, 2001). Through integration, firms are able to take advantage of such interdependencies by sharing and combining resources across segments. When these resource combinations are leveraged across multiple markets, this heterogeneity may be a source of superior returns (Peteraf, 1993; Miller, 2003). Also, the routines on which such processes are based may be highly complex, dependent on the firm’s unique history, and therefore difficult for rivals to imitate (Barney, 1991; Porter, 1996). Thus, we contend that focusing on the formulation of diversification strategies alone fails to consider the means by which firms derive value from diversifying investments in scale and scope. In this section, we attempt to address how firms exploit the scale and scope achieved through diversification by utilizing the resource management framework to examine the implementation of diversification strategies. At its core, the RBV is largely based on the argument that each firm’s resource and capability endowment is unique (Penrose, 1959; Peteraf, 1993). In this chapter, we define resources as either tangible or intangible assets owned or controlled by firms that allow them to implement their strategies (Barney, 1991; Grant, 1991). Capabilities refer to bundled combinations of resources that enable a firm to perform its activities (Grant, 1991; Sirmon
562
TIM R. HOLCOMB ET AL.
et al., 2007). Although a dominant perspective in management research, the RBV has been criticized. Three of the more common criticisms of the RBV are as follows: (1) much of the RBV research largely ignores how resources are managed to create competitive advantage, (2) the RBV presents a static view of an evolving and dynamic process, and (3) the RBV fails to consider external contingencies such as the competitive context and the maturity of market institutions (Barney & Arikan, 2001; Priem & Butler, 2001). Therefore, the RBV requires further development to explain the continuous management of the firm’s resources and the effects of external contingencies on this process. We address these issues by examining resource management within the context of diversification. As Fig. 2 illustrates, we represent resource management as a ‘‘closed-loop’’ process. Structuring the resource portfolio provides resources used by the firm to create capabilities through the bundling process. Bundling refers to the processes that firms use to integrate resources to create new capabilities and leveraging refers to the deployment of such capabilities to take advantage of market opportunities. Because leveraging converts the firm’s bundled resource combinations into value (Sirmon et al., 2007), leveraging is often instrumental in subsequent structuring activities (cf. Miller, Eisenstat, & Foote, 2002). Indeed, firm-specific advantages that result from capabilities
Strategic Actions to Create Scale and Scope Internal Development
Strategic Alliances
Acquisitions
External Contingencies: • Environmental Uncertainty • Environmental Munificence • Competitive Rivalry
Structuring the Resource Portfolio
“Closed-loop” Resource Management Process Leveraging Capabilities to Create Value
Fig. 2.
Bundling Resources to Form New Capabilities
Managing Resources to Exploit Scale and Scope Economies.
Diversification to Achieve Scale and Scope
563
that are effectively leveraged by the firm may influence its subsequent diversification strategies (cf. Gimeno et al., 2005). We explore resource management and the effects of different strategic actions on scale and scope economies more fully in the following sections.
Structuring the Resource Portfolio Diversification strategies allow firms to structure their resources to achieve differentiation. Structuring the resource portfolio is the process by which firms accumulate (Dierickx & Cool, 1989; Thomke & Kuemmerle, 2002), acquire (Barney, 1986; Denrell, Fang, & Winter, 2003; Makadok, 2001), and access (Das, Sen, & Sengupta, 1998; Hitt, Dacin, Levitas, Arregle, & Borza, 2000) resources. Accumulating resources refers to the continuing development of firms’ existing resources (Dierickx & Cool, 1989). Because firms have superior information about resources they control, they are more likely to recognize the value of their own resources (Alchian & Demsetz, 1972; Grant, 1991), which may be especially valuable when uncertainty makes it difficult to assess the value of external resources. Accumulating resources may also be crucial in low munificent environments when resources cannot easily be obtained from external factor markets (Sirmon et al., 2007). Diversification strategies that utilize accumulation processes involve recombining existing resources into new capabilities, allowing firms to capitalize on resources they currently own to exploit opportunities in new product and geographic markets. Also, when diversification involves accumulation, natural isolating mechanisms emerge because the firm has less need to interact intensely with other firms, further increasing the sustainability of achieved performance advantages (Thomke & Kuemmerle, 2002). Although accumulating resources and recombining them in valuable and difficult-to-imitate ways can facilitate entry into profitable new markets, especially in the short term, a firm’s reliance on accumulated resources may limit the amount of change that is possible. Significantly increasing the scope of a firm’s capabilities often requires new resources from outside the firm (Morrow et al., 2005). Acquiring and accessing resources are two means to utilize external resources. Acquiring resources refers to purchasing them from strategic factor markets (Barney, 1986). A common means of acquiring new resources is through acquisitions (Hitt et al., 1996). Diversification strategies that allow firms to acquire additional resources may improve a firm’s ability to compete by creating new capabilities or altering existing capabilities substantially. For
564
TIM R. HOLCOMB ET AL.
example, acquisitions infuse new resources into a firm, potentially revitalizing the firm and helping it overcome the harmful effects of rigidity (Vermeulen & Barkema, 2001). Moreover, because information is asymmetrically distributed in the environment, strategic factor markets often have incomplete information about the possible uses of resources and therefore do not always accurately price resources that can be used in new ways (Denrell et al., 2003). These conditions create different expectations about the future value of resources. Consequently, when firms possess superior information, there may be opportunities to acquire resources below their true market value (Makadok, 2001). By diversifying to obtain resources to integrate with those resources a firm already controls, it may be able to create value over and above the costs of both old and new resources. In addition, when firms diversify with the intent of creating new resource combinations, differences between the firm and its rivals are likely to increase. Such heterogeneity may be highly inimitable and thus can provide a basis for achieving competitive advantage. However, acquiring resources often represents a commitment to hold and manage them for an extended period, especially when the liquidity of resources is low. Thus, in highly dynamic environments, acquiring resources is often risky because the value of the given resources is generally contingent on the degree to which the resources match the requirements of the external environment (Black & Boal, 1994). Rapid technological changes, globalization, and increased competition have created conditions in which few firms possess all of the resources needed to remain competitive (Ireland, Hitt, & Vaidyanath, 2002). When firms lack certain resources, they may attempt to access resources without taking ownership and control over them. For example, the need to gain access to resources is a common motivation for strategic alliances (Das et al., 1998; Hitt et al., 2000). Resources accessed through alliances often demand lower financial commitments and may require less managerial attention than resources owned by firms, allowing firms to focus their efforts on extracting additional value from those resources they own. Accessing resources can be especially beneficial in dynamic environments by providing firms the flexibility to utilize resources on a short-term basis. Thus, accessing may be preferred when firms diversify into environments in which they confront uncertainty. Bundling Resources to Create Capabilities Bundling is the process by which unique capabilities are formed, with each capability representing a unique combination of resources that enable firms
Diversification to Achieve Scale and Scope
565
to pursue specific actions (e.g., marketing, R&D, etc.) that create value (Sirmon et al., 2007). Competition between firms is often driven by similarities or differences in the means by which firms use resources, not necessarily by similarities or differences in the resources they control (Peteraf & Bergen, 2003). When successful firms create new capabilities through diversification, they may achieve superior returns by performing value-creating tasks more effectively than rivals can, allowing them to effectively exploit those capabilities in the marketplace (Grant, 1991; Makadok, 2001; Blyler & Coff, 2003). Because capabilities are idiosyncratic to a firm, they cannot be easily bought and sold in strategic factor markets (Teece, Pisano, & Shuen, 1997). Many successful firms are able to explicitly integrate or configure their capabilities in ways that are difficult for competitors to imitate (Galunic & Rodan, 1998). Creating and exploiting unique capabilities require that diversifying firms embed them into an organizational structure that sustains, enriches, and substantially enhances their value (Miller, 2003; Sirmon et al., 2007). Proper structural alignment provides the opportunity to create synergies that make capabilities more valuable when leveraged across product and geographic markets than when managed separately. Indeed, better economies have been cited as one of the most salient potential benefits of sharing resources across different market segments (Gimeno & Woo, 1999). Creating unique resource combinations may require a delicate balance between centralization and decentralization in order to realize scale and scope economies through diversification. Firms pursuing product and international diversification often adopt decentralized structures (Chandler, 1962; Lu & Beamish, 2001). Granting autonomy to the firm’s division or subsidiary managers allows them the freedom to identify potentially valuable combinations of resources and configurations of the resulting capabilities. In addition, given their proximity and knowledge of the firm’s internal routines and processes, nonexecutive managers may be able to better facilitate the bundling of the firm’s resources and configuration of their capabilities. Nonetheless, coordination is needed throughout the entire organization, especially when diversification substantially alters the boundaries of the firm. Thus, a certain level of centralized decision-making may be required to allocate resources to the bundling process in an optimal manner. Different types of bundling processes often produce different capabilities. Bundling processes required when firms are attempting to produce incremental change differ from those required when the goal is to achieve a more substantial change in capabilities (Hamel & Prahalad, 1994). Henderson and Mitchell (1990) make a similar point in addressing the difficulty firms
566
TIM R. HOLCOMB ET AL.
have responding to different types of innovation. The identification of potentially valuable resource combinations is enhanced to the extent that firms have the ability to develop a comprehensive understanding of their resources, current capabilities, and processes (Winter, 2000; Zollo & Winter, 2002). With experience, firms assimilate knowledge from the diversification strategies they undertake. Although product and international diversification can enhance learning (Bartlett & Ghoshal, 1989; Bergh, 2001), both forms of diversification can also create conditions inside the firm that discourage learning. For example, diversifying actions often disrupt the firm’s operations. Such disruptions create uncertainty that stifles learning by making cause and effect associations difficult to discern (Fiol & Lyles, 1985; Newman, 2000). In addition, the process of diversification can affect a firm’s capacity to learn because expanding too rapidly can subject the firm to time compression diseconomies that limit the amount of knowledge that can be absorbed and applied in a given time period (Dierickx & Cool, 1989). However, expanding too slowly may cause the firm to lose some of the knowledge previously learned (Argote, Beckman, & Epple, 1990; Darr, Argote, & Epple, 1995). As such, some researchers have suggested that expanding at a regular pace without allowing too much time to elapse between specific actions is an optimal way for firms to learn through diversification (Hayward, 2002; Vermeulen & Barkema, 2002). Because successfully combining resources owned or controlled by the firm is difficult and may require considerable time and effort, the ability to understand and use feedback to improve integration becomes critical (Lei, Hitt, & Bettis, 1996). The efficient and effective utilization of feedback is enhanced when knowledge is shared openly among managers and employees. Knowledge sharing enables learning and often increases the absorptive capacity of firm employees, enhancing their ability to interpret and utilize feedback. Learning can also lead to the creation of new knowledge (Kogut & Zander, 1992) that helps the firm identify new resources and resource combinations. Thus, firms must be able to design structures and processes that encourage knowledge sharing (Teece et al., 1997). For example, the use of liaison personnel, employee transfers, and teams often improves knowledge sharing among the firm’s employees (Galbraith, 1973; Gupta & Govindarajan, 2000; Jin & Holmes, 2006). Incentive plans, internally consistent promotion schemes, and strategic controls can also enhance intrafirm communication (Hitt et al., 1996; Murtha, Lenway, & Bagozzi, 1998), which further increases the likelihood of configuring resources into unique capabilities.
Diversification to Achieve Scale and Scope
567
Leveraging Capabilities in the Market Leveraging involves processes used to deploy the firm’s capabilities in the market to take advantage of specific market opportunities (Sirmon et al., 2007). The complexities of markets and the presence of underserved niches (Carroll, 1985) create opportunities for firms to leverage their capabilities to serve the needs of customers and generate profits (Miller, 2003). Leveraging involves determining both where and how to deploy the firm’s capabilities in the market (Brush, Green, & Hart, 2001). Ideally, capabilities are fungible across different market settings, which allow firms to achieve scale and scope economies by deploying their capabilities to reach customers across multiple markets (Prahalad & Hamel, 1990; Miller et al., 2002). Value creation involves matching the firm’s capabilities to customers’ needs in ways that achieve a unique competitive position in the market relative to rivals (Porter, 1980, 1996). However, markets are dynamic, customer demand often shifts quickly and unexpectedly, and competitors may engage in aggressive actions to attack the firm’s market position (Smith, Grimm, & Gannon, 1992; Chen, 1996). Leveraging processes are improved by extensive scanning and learning to identify areas of the market where the firm’s capabilities can create the most value (Sirmon et al., 2007). This allows firms to target their capabilities away from undesirable markets and toward opportunities that are more promising. Diversified firms must also be proactive in responding to conditions that impede the leveraging process. For example, product-diversified firms attempting to aggressively leverage their capabilities in new markets that are important to rivals are likely to face retaliatory actions by rivals in those or other markets (Gimeno, 1999; Gimeno & Woo, 1999). In addition, diversification, especially internationalization, exposes the firm to a diverse set of institutional environments and cultural and language differences. Thus, routines used to leverage capabilities in some markets may not be applicable to conditions prevalent in other markets (Nelson & Winter, 1982; Barkema & Vermeulen, 1998; Greve & Baum, 2001). Moreover, responding to environmental dynamism and heterogeneity requires continuous monitoring of the firm’s products and their performance in different markets (Sorenson, 2000). Firms must be able to learn from both favorable and unfavorable experiences (Miller, 2003). Learning from favorable experiences allows firms to direct capabilities at those segments where the capabilities are most valuable. Learning from unfavorable experiences allows firms to adjust capabilities to better satisfy new or existing segments.
568
TIM R. HOLCOMB ET AL.
Diversifying into market segments that are similar to those in which the firm already has a presence can make leveraging more effective. Firms diversifying into similar segments can utilize valuable information that they already possess about customer needs in those segments. Firms can also serve such segments with existing value-added activities and learning processes developed previously (Kogut, 1984). In contrast, when firms expand into unfamiliar markets, accessing knowledge and other resources from external sources helps exploit opportunities. For example, firms may be able to supplement their knowledge of unfamiliar markets by hiring new employees with knowledge of such markets (Jin & Holmes, 2006). Additionally, some firms diversifying into unfamiliar markets may be able to alter that market’s competitive landscape to fit the firm’s product offerings. Indeed, some diversifying actions provide the firm with technology that, when combined with the firm’s current technology, help the firm develop innovative products that create either new market segments or new niches in a given market (Cantwell, 2001). Although leveraging is difficult, it is critical to realizing the benefits of investments in scale and scope. Indeed, a uniquely structured portfolio bundled into firm-specific capabilities may be of little use if the capabilities cannot be leveraged into desirable returns for the firm and its shareholders. Moreover, whereas few competitive advantages are sustainable in the long term (Eisenhardt & Martin, 2000), successful leveraging generates profits and learning opportunities that provide the financial resources and knowledge necessary to facilitate subsequent structuring and capability development actions, which can lead to new profit opportunities (Miller, 2003).
STRATEGIC ACTIONS TO CREATE SCALE AND SCOPE Firms enact competitive strategy through purposive actions that give rise to changes in scale and scope (Porter, 1980). Strategy orients a firm’s behavior and guides its actions. In this work, diversifying actions represent the organizing modes that firms use to pursue entry into new product and geographic markets. Different organizing modes provide access to different resources and affect the process by which firms manage diversification. It is important to study strategic actions for several reasons. Different strategic actions often require organizational structures, organizing methods,
Diversification to Achieve Scale and Scope
569
and routines that best fit each action. These factors, alone or in combination, affect knowledge sharing and internal cooperation (Conner & Prahalad, 1996; Grant, 1996). Knowledge sharing and internal cooperation influence resource allocation decisions and the coordination of the firm’s activities, both of which are embedded within structuring, bundling, and leveraging routines. As a result, the value derived by firms from different strategic actions may vary not only because of the actions, but also because of the idiosyncratic structures, methods, and routines required to execute such actions. Variations in organizational structures, organizing methods, and routines may explain organizational founding and mortality in an ecological sense (e.g., Bryce & Singh, 2001). Population dynamics depend on the actions a given firm pursues and also on the competitive actions taken by rivals. In this section, we examine three common and heavily studied strategic actions firms employ when diversifying their operations to create scale and scope: internal development, acquisitions, and strategic alliances.
Internal Development Internal development actions occur when existing resources are recombined and leveraged in order to diversify into new product or geographic business units by making a greenfield investment or creating a new start-up operation (i.e., plant, subsidiary, division, etc.). Internal development is often preferred when firms already possess valuable resources (Hennart & Park, 1993), when they seek to protect resources from expropriation by other firms (Teece, 1986), or when strategic factor markets are unlikely to provide all of the resources firms need (Sirmon et al., 2007). There are important reasons that firms pursue internal development. First, firms are sufficiently motivated to seek optimal uses of existing resources and related firm-specific advantages. Because managers possess superior knowledge about existing resources, valuable and difficult-to-imitate new combinations are possible (Denrell et al., 2003). Further, firms often pursue start-ups when diversifying into foreign markets in order to exploit firm-specific resource advantages embedded in the organization (Hennart & Park, 1993). Also, when diversifying into environments where the availability of resources is low, the use of existing resources may be preferred. At the extreme, internal development may be the only viable option for firms operating in markets characterized by low munificence. Recombining existing resources into new capabilities can be especially helpful in the internal development of new products (Morrow et al., 2005).
570
TIM R. HOLCOMB ET AL.
Existing managers and employees understand the firm’s current resources and capabilities and may therefore be better able to effectively bundle and leverage these resources. Exploiting new product and geographic markets with existing resources should enhance value creation because utilizing existing resources in new ways is likely to be unexpected by investors, and thus helps firms exceed investors’ expectations (Morrow et al., 2005). Transaction costs are also much lower in these cases. Further, avoiding collaboration with other firms protects the firm’s most valuable resources from current or potential competitors by buffering its proprietary technology (Hennart & Park, 1993). Thus, utilizing existing resources and resource combinations to diversify into new markets may be less costly than obtaining new resources externally (Barkema & Vermeulen, 1998). Finally, firms that diversify into related segments may experience greater success. Internal development actions often build upon and use many of the firm’s current routines and processes (Barkema & Vermeulen, 1998). Entry into a related market environment facilitates sharing these routines and processes (cf. Nelson & Winter, 1982), but diversifying into unrelated segments may make such sharing especially difficult or even irrelevant (Cohen & Bacdayan, 1994). Also, relatedness improves the likelihood that a firm’s resources will be applicable across segments (Robins & Wiersema, 1995). Thus, diversifying into related segments enables a new unit to share resources with the parent and with other units in the firm, which facilitates the likelihood of recombination to build new or improved capabilities (Vermeulen & Barkema, 2001). Relatedness also improves leveraging by allowing firms both to utilize their current knowledge of customer demand and to take advantage of existing value-added activities to service that demand. Internal development also has several disadvantages. Because this approach to diversification relies on existing resource endowments as the basis of competitive advantage, internal development often requires that firms already possess a valuable stock of resources (Hitt et al., 1994; Barkema & Vermeulen, 1998). If achieving, maintaining, or extending competitive advantage requires resources not currently possessed by the firm, internal development may not be the ideal form of expansion. Indeed, a firm’s reliance on existing resources often limits the amount of change possible and favors increased simplicity in firm operations (Miller, 1993). Thus, pursuing internal development alone can stifle attempts to identify and create novel resource combinations over time (March, 1991; Baum et al., 2000). Simplicity may result, as path-dependence channels strategic behavior and choice along a consistent and defined pathway. This simplicity may reflect organizational
Diversification to Achieve Scale and Scope
571
rigidity that limits a firm’s ability to adapt to changing environments (Miller, 1993; Vermeulen & Barkema, 2001). Combating the progressive simplicity associated with internal development may require creating and nurturing an organizational culture that promotes flexibility, experimentation, and cooperation among firm employees.
Acquisitions Although recombining existing resources in valuable and difficult-to-imitate ways can facilitate the development of profitable new market offerings, especially in the short term, firms frequently require new capabilities to remain competitive. Developing new capabilities commonly requires an infusion of new resources. A common means of obtaining new resources is through acquisitions (Hitt et al., 1996; Karim & Mitchell, 2000). Acquisitions may involve either one firm purchasing a majority interest in another firm or two firms surrendering their independence, combining their operations, and creating a new legal entity (Hitt, Harrison, & Ireland, 2001). Acquisitions not only add new resources, but also give acquiring firms control over the acquired firm’s resources. Thus, expansion through acquisition can be valuable when the competitive landscape requires new and different types of resources. Moreover, acquisitions can help to rejuvenate the firm and break unproductive internal rigidities (Vermeulen & Barkema, 2001). By combining the resources of the target with those of the acquiring firm, the newly combined firm often has the potential to generate capabilities that can be leveraged more effectively in the marketplace (Harrison, Hitt, Hoskisson, & Ireland, 1991). Further, acquisitions can also insulate firms from competitive attacks because such actions provide new, distinct, and valuable market positions that enhance firms’ ability to respond to rivals. Acquisitions also combine the market shares of the acquiring and target firms in markets that overlap. Increased market share can generate economies of scale and scope that increase competitiveness by driving down costs. Indeed, a larger market share may translate into enhanced market power. Market power provides the combined firm with greater bargaining power over suppliers and buyers, resulting in lower factor costs and price premiums for the firm’s offerings (Porter, 1980). Finally, acquisitions can help to quickly redeploy or replicate the firm’s existing capabilities into new markets (Helfat & Peteraf, 2003), thereby providing the newly combined firm with an advantage over slower moving rivals.
572
TIM R. HOLCOMB ET AL.
However, acquisitions are inherently risky (Pablo, Sitkin, & Jemison, 1996). Integration commonly requires significant time and effort, and acquisitions often fail because projected synergies do not materialize (Hitt et al., 2001). Further, acquisitions can be expensive, which makes it difficult for firms to generate value above the cost of the acquisition. Indeed, much of the potential value of the acquisition is transferred to the target firm’s shareholders in the form of the acquisition premium (Barney, 1986). For example, acquiring firms often pay large premiums for target firms, potentially nullifying the returns available to the acquiring firm’s shareholders (Barney, 1986; Hayward & Hambrick, 1997). This problem is exacerbated when the firm competes in a densely populated market, both because more firms are available to bid up the acquisition price and because there are more rivals competing over the same returns. Moreover, the expense and difficulty of managing acquisitions can result in the allocation of fewer funds to other profitable investments (i.e., opportunity costs) (Lei & Hitt, 1995; Hitt et al., 1996). These opportunity costs limit the combined firm’s ability to create and utilize novel resource and capability configurations in the future. Thus, due diligence is necessary to reduce information asymmetries about the assets being acquired and to carefully evaluate potential synergies (Hitt et al., 2001). Nonetheless, acquisitions can be viable when managed properly (Hayward, 2002). Indeed, firms must properly integrate acquisitions if synergy is to be created and competitive advantage and superior performance are to result. The existence of complementary resources contributes to synergy when acquisitions are successfully integrated. Synergy that cannot be anticipated or duplicated by rivals is called private synergy (Mahoney & Pandian, 1992; Hitt et al., 2001). Private synergy results from combining complementary resources and is important because it is inimitable and is unlikely to be accounted for fully in the acquisition price (Harrison et al., 1991). Similarly, research suggests that the bidding process is unlikely to allocate all of the potential gains into the acquisition price when the ultimate source of value lies with the acquiring firm’s resources because rival firms do not have the same resources and cannot create the same synergies (Capron & Pistre, 2002). Additionally, compatibility between the parent and the target increases the likelihood of success. Compatibility is enhanced by the presence of strategic and organizational fit. Strategic fit refers to the matching of capabilities related to strategy formulation and implementation. Organizational fit refers to the similarity in management processes, cultures, structures, and systems between the firms (Harrison & St. John, 1998). The
Diversification to Achieve Scale and Scope
573
existence of strategic and organizational fit suggests that integration processes can be expected to enhance knowledge transfer and facilitate resource sharing between the firms. Thus, by providing a more seamless transfer of knowledge and resources, compatibility can improve bundling and leveraging processes. Acquisitions are also likely to be more successful when acquiring firms follow an established pattern of acquisitions within their industry (Haleblian & Finkelstein, 1999). Pursuing a consistent pattern of acquisitions fosters learning, allowing insights from prior acquisition experiences to be utilized (Amburgey & Miner, 1992; Baum et al., 2000). Such experience can improve an acquiring firm’s ability to evaluate targets and can enhance the development of skills necessary to bundle the target’s resources with those of the acquirer. Exploiting the target’s institutional memory can also improve success (Hitt et al., 2001) because institutional memory, which is embedded in managers’ knowledge sets, provides information about idiosyncrasies unique to the firm (Cannella & Holcomb, 2005). Such knowledge facilitates the identification of valuable resources and capabilities in markets served by the target.
Strategic Alliances Engaging in strategic alliances provides firms with access to, but not control of, additional resources from their partners (Das et al., 1998; Hitt et al., 2000). Strategic alliances are cooperative arrangements that utilize resources from two or more autonomous firms and are intended to improve each firm’s performance (Ireland et al., 2002). Diversification into new product or geographic markets that is achieved through alliances offers several advantages. Alliances often are used to access the resources of other firms when firms cannot or choose not to accumulate those resources internally or acquire them permanently. Thus, alliances allow partner firms to create resource combinations that single firms cannot (Das & Teng, 2000). Alliances may also allow firms to more effectively leverage existing resources by bundling them with resources accessed from partners (Ireland et al., 2002). The ability to gain access to another firm’s resources without controlling them (Koh & Venkataraman, 1991) generally makes alliances less capital intensive and therefore more flexible than actions involving internal development or acquisitions (Inkpen, 2001). Further, alliances are often utilized when firms lack the necessary resources to enter a new market with sufficient scale (e.g., Shan, Walker, &
574
TIM R. HOLCOMB ET AL.
Kogut, 1994). Alliances are also common when firms confront uncertainty, such as when entering an unfamiliar market for the first time (Oliver, 1990; Rodriguez, Uhlenbruck, & Eden, 2005). Scholars have also identified several disadvantages in alliances. For example, the lack of control over most of the accessed resources can limit the formation of unique resource combinations. When opportunities are identified, attempts to transfer and combine resources may result in higher transactions costs (Robins, Tallman, & Fladmoe-Lindquist, 2002). Further, alliances carry a risk of dependency. Firms that are overly dependent on their alliance partners may become vulnerable to the whims and fates of such partners (Miner et al., 1990). Similarly, because many alliances include elements of both cooperation and competition (Yan & Gray, 1994; Amburgey, Dacin, & Singh, 1996), alliance partners often must closely guard proprietary resources to prevent their expropriation by partners (Hamel, 1991). By hindering cooperation, such behavior can limit a firm’s ability to effectively bundle resources accessed from partners into valuable capabilities. Leveraging resources accessed from alliance partners can also be problematic. A lack of control over accessed resources and the need to collaborate with a partner on key decisions can slow decision-making. Delayed responsiveness can result in lost flexibility and difficulty in responding to changing market conditions. Resources accessed through alliances are also not entirely transparent to managers, which may further limit adaptability. Additionally, shifts in inter-partner dependence can similarly shift each partner’s bargaining power, causing the alliance to become unstable (Inkpen & Beamish, 1997). Opportunistic expropriation of alliance profits by one partner can inflict severe harm on the exploited partner (Alvarez & Barney, 2001), resulting in lost opportunities and forfeiture of valuable competitive positions gained through the alliance. Overcoming the challenges of alliances often begins with partner selection. Partner selection is important both structurally (i.e., the overall composition of the network) and relationally (i.e., interactions with specific partners). Indeed, evidence suggests that the structure of interfirm networks influence alliance formation patterns (Gulati, 1995). Firms that establish a position in a heterogeneous network of alliances may have access to diverse information and other resources that improve competitiveness (Baum, Calabrese, & Silverman, 2000). Selection of specific partners is also important. Ideally, a partner is willing to share knowledge with the focal firm (Hitt et al., 2000). Partner-specific absorptive capacity is also needed so that exchanged knowledge can be readily understood and applied by each partner
Diversification to Achieve Scale and Scope
575
(Dyer & Singh, 1998; Lane & Lubatkin, 1998). Further, selecting a partner with similar intentions that adopts a benevolent stance toward the partnership reduces the likelihood of exploitation (Yan & Gray, 1994), thereby allowing the partners to develop a trusting relationship. Trust increases each partner’s confidence that sharing resources will not expose the partner to the possibility of exploitation and therefore encourages the partners to share resources (Ireland et al., 2002). The free exchange of knowledge and other resources increases the likelihood of identifying, developing, and leveraging new capabilities. Alliances should also be embedded within an appropriate governance structure (Poppo & Zenger, 2002). Combining well-specified contracts that delimit partners’ control over alliance activities with informal mechanisms (such as trust) that institutionalize common goals help protect the alliance partners while promoting an environment in which cooperation can thrive (Yan & Gray, 1994; Luo, 2002). After the partnership is formed, extensive cooperation between the partners remains important (Dyer & Singh, 1998). Cooperation facilitates the ongoing process of responding to changing customer demands with new capabilities. These quality relationships are also often specific to the partnership (Kale, Dyer, & Singh, 2002) so alliance partners that maintain such relationships can have a competitive advantage over other firms (Dyer & Singh, 1998). Finally, although managing alliances is challenging, firms with dedicated alliance functions often fare better than firms without such functions (Kale et al., 2002). Dedicated alliance functions help firms identify partners with the desired resources, monitor the partnership’s governance structure, manage the configuration of the two firms’ resources, and promote successful leveraging of the joint firms’ capabilities.
DISCUSSION An important conclusion in previous research on product and international diversification is that such strategies may not achieve the performance outcomes desired by firms. Indeed, equivocal research findings may be the outcome of a research tradition that has largely neglected the implementation of diversification strategies. However, as previous scholars have made clear, understanding how to implement these strategies effectively is as important as understanding how to formulate the appropriate strategic actions (Ramanujam & Varadarajan, 1989; Hoskisson & Hitt, 1990).
576
TIM R. HOLCOMB ET AL.
Although firms often create scale and scope advantages by diversifying into new product or geographic markets, achieving consistent and beneficial performance outcomes from such actions remain an illusive goal. Moreover, even if such advantages make firms more efficient, scale and scope advantages may not result in superior performance if rivals are able to easily replicate similar advantages (Gimeno & Woo, 1999). Indeed, the performance implications of a given diversification strategy depend on the way that strategy is implemented. This chapter represents an attempt to extend theory by considering how firms implement their diversification strategies to derive the benefits of organizational scale and scope. Our work suggests that integrating the effects of internal firm-specific factors (i.e., strategic actions and resource management) with external conditions (i.e., environmental uncertainty, environmental munificence, and competitive rivalry) has the potential to improve future efforts to explain variation in performance outcomes across diversification strategies. This work has important implications for framing the evaluation of strategic actions that firms take to diversify and develop the scale and scope of their businesses. As noted, diversification strategies expressly develop organizational scale and scope but may fail to generate the desired returns. Further, performance advantages achieved may be short-lived. Competitors can change their offerings and threaten or eliminate a firm’s competitive advantage. Attacks by rivals may require, for example, recombining existing resources in new ways to extend established competitive positions or acquiring new resources from strategic factor markets to create capabilities more applicable to new competitive environments. Additionally, changes in macro-environmental conditions or government regulations may alter the competitive environment by affecting the supply of resources and/or the way in which resources may be used. Thus, firms must continuously scan their environment to identify emerging opportunities and threats. Effective monitoring helps identify segments and niches in the market where existing and new resources and bundled capabilities can be leveraged to create value.
Future Extensions With this in mind, we propose several avenues for future research. First, there is a need to develop better measures of resources and capabilities as well as the underlying processes used to structure the firm’s resource portfolio, bundle its resources to form new capabilities, and leverage the capabilities created.
Diversification to Achieve Scale and Scope
577
Measures must identify and capture the processes consistently over time and also distinguish among them. Further, measures must be applicable in multiple organizational and environmental contexts to facilitate greater comparability across studies. Second, we support Hrebiniak and Joyce’s (2001) call for more research on strategy implementation. Comprehensive and theoretically sound case studies are needed to develop our knowledge of implementation and facilitate further theory development. Resource management may be instrumental in improving our understanding of strategy implementation, thereby taking on added importance. For example, synchronizing resource management is particularly difficult (Sirmon et al., 2007), especially when firms conduct business across multiple product and geographic markets. Thus, we also need work that identifies the interdependencies among structuring, bundling, and leveraging and how firms can exploit such interdependencies to develop a competitive advantage. Third, we need to examine how firms can continuously upgrade valuable and difficult-to-imitate capabilities. For example, to what extent are patterns of organizational competitiveness consistent with Miller’s (2003) suggestion that organizational performance can exhibit path dependent effects? Are any of the three strategic actions especially advantageous in preserving or enhancing a firm’s position in this cycle? Finally, how does the nature of the firm’s environment influence which strategic action is more beneficial and how it is implemented? Fourth, further theory development and empirical investigation of the internal and external factors that moderate the relationships between strategy and firm performance are needed. For example, future research on product and international diversification should more closely examine the contingent effects of environmental factors on both the strategic actions firms pursue (i.e., internal development, strategic alliances, and acquisitions) and their routines through which firms manage the resources and the new resource combinations these actions generate. As strategic management scholars examine environmental factors, insights from the evolutionary perspectives on organizational change may prove especially valuable (e.g., Barnett & Burgelman, 1996). Integrating these insights with the research on resource management may help us better understand how firms implement diversification strategies to realize the benefits of changes to organizational scale and scope. Fifth, we need more research to understand the information-processing demands of diversification and how firms respond to these demands. For example, the difficulty of managing diversified firms may be compounded
578
TIM R. HOLCOMB ET AL.
when firms diversify into highly heterogeneous market segments and/or employ a diverse mix of strategic actions. How does the mix of environments into which the firm expands and the mix of strategic actions it employs influence its ability to succeed through diversification? Theory and empirical evidence suggesting a curvilinear relationship between diversification and performance (e.g., Hitt et al., 1997; Palich et al., 2000) and the potential moderating influence of intangible assets (Lu & Beamish, 2004) suggest the importance of organizational learning. However, we still do not fully understand the underlying processes that give rise to the performance effects found in prior work. For instance, do firms pursuing similar acquisition targets experience more success (Haleblian & Finkelstein, 1999) because learning effects allow them to purchase resources below their future value or is superior performance due to bundling and leveraging processes? Similar inquiries are needed on diversification strategies involving internal development and strategic alliances.
CONCLUSION This chapter has suggested that research on the performance effects of diversification should examine both the diversification strategies that firms pursue and the processes they follow to implement such strategies. Resource management may be especially critical in exploiting the advantages of scale and scope to create value. Resource management has been conceptualized as an integrated ‘‘closed-loop’’ cycle in which the experience gained from diversifying actions over time makes new resources available and provides ongoing learning opportunities that may enhance the firm’s competitiveness and performance prospects. In particular, firms need to accumulate, acquire, and access resources to establish and maintain an effective resource portfolio. Structuring the resource portfolio through strategic actions – specifically, internal development, acquisitions, and strategic alliances – hold the potential for achieving the benefits of organizational scale and scope. However, because the performance effect of diversification is contingent on firmspecific routines and environmental conditions, firms need to coordinate their activities to achieve alignment with both their internal and external contexts. Through bundling, a firm can differentiate itself from other firms. Leveraging processes help convert such differentiation into profits. By examining these processes, and how they operate under varying organizational and environmental conditions, we hope that we have contributed to the growing literature on strategy implementation.
Diversification to Achieve Scale and Scope
579
ACKNOWLEDGMENTS We are grateful to the editors of this volume, Joel A. C. Baum, Stanislav Dobrev, and Arjen van Witteloostuijn, for their insightful comments on earlier versions of this paper.
REFERENCES Alchian, A. A., & Demsetz, H. (1972). Production, information costs, and economic organization. American Economic Review, 62, 777–795. Alvarez, S. A., & Barney, J. B. (2001). How entrepreneurial firms can benefit from alliances with large partners. Academy of Management Executive, 15, 139–148. Amburgey, T. L., Dacin, T., & Singh, J. V. (1996). Learning races, patent races, and capital races: Strategic interaction and embeddedness within organizational fields. In: P. Shrivastava, A. S. Huff, & J. E. Dutton (Eds), Advances in strategic management (Vol. 13, pp., 303–322). Greenwich, CT: JAI Press, Inc. Amburgey, T. L., Kelly, D., & Barnett, W. P. (1993). Resetting the clock: The dynamics of organizational change and failure. Administrative Science Quarterly, 38, 51–73. Amburgey, T. L., & Miner, A. S. (1992). Strategic momentum: The effects of repetitive, positional, and contextual momentum on merger activity. Strategic Management Journal, 13, 335–348. Argote, L., Beckman, S. L., & Epple, D. (1990). The persistence and transfer of learning in industrial settings. Management Science, 36, 140–154. Barkema, H. G., & Vermeulen, F. (1998). International expansion through start-up or acquisition: A learning perspective. Academy of Management Journal, 41, 7–26. Barnett, W. P., & Burgelman, R. A. (1996). Evolutionary perspectives on strategy. Strategic Management Journal, 17, 5–19. Barney, J. B. (1986). Strategic factor markets: Expectations, luck and business strategy. Management Science, 32, 1230–1241. Barney, J. B. (1991). Firm resources and sustained competitive advantage. Journal of Management, 17, 99–120. Barney, J. B. & Arikan, A. M. (2001). The resource-based view: Origins and implications. In: M. A. Hitt, R. E. Freeman, & J. S. Harrison (Eds), The Blackwell handbook of strategic management (pp. 124–188). Malden, MA: Blackwell Publishers Inc. Bartlett, C. A., & Ghoshal, S. (1989). Managing across borders: The transnational solution. Boston: Harvard Business School Press. Baum, J. A. C. (1996). Organizational ecology. In: S. R. Clegg, C. Hardy, & W. Nord (Eds), Handbook of organizational studies, London: Sage. Baum, J. A. C., Calabrese, T., & Silverman, B. S. (2000). Don’t go it alone: Alliance network composition and startups’ performance in Canadian biotechnology. Strategic Management Journal, 21, 267–294. Baum, J. A. C., Li, S. X., & Usher, J. M. (2000). Making the next move: How experiential and vicarious learning shape the locations of chains’ acquisitions. Administrative Science Quarterly, 45, 766–801.
580
TIM R. HOLCOMB ET AL.
Baum, J. A. C., & Oliver, C. (1991). Institutional linkages and organizational mortality. Administrative Science Quarterly, 36, 187–218. Baum, J. A. C., & Oliver, C. (1996). Toward an institutional ecology of organizational founding. Academy of Management Journal, 39, 1378–1427. Baum, J. A. C., & Powell, W. W. (1995). Cultivating an institutional ecology of organizations: Comment. American Sociological Review, 60, 529–538. Baysinger, B., & Hoskisson, R. E. (1989). Diversification strategy and R&D intensity in multiproduct firms. Academy of Management Journal, 32, 310–332. Bergh, D. D. (2001). Diversification strategy research at a crossroads: Established, emerging and anticipated paths. In: M. A. Hitt, R. E. Freeman, & J. S. Harrison (Eds), The Blackwell handbook of strategic management(pp. 362–383). Malden, MA: Blackwell Publishers Inc. Black, J. A., & Boal, K. B. (1994). Strategic resources: Traits, configurations and paths to sustainable competitive advantage. Strategic Management Journal, 15, 131–148. Blyler, M., & Coff, R. W. (2003). Dynamic capabilities, social capital, and rent appropriation: Ties that split pies. Strategic Management Journal, 24, 677–686. Bourgeois, L. J. (1980). Performance and consensus. Strategic Management Journal, 1, 227–248. Bowman, E. H., & Helfat, C. E. (2001). Does corporate strategy matter? Strategic Management Journal, 22, 1–23. Brush, C. G., Greene, P. G., & Hart, M. M. (2001). From initial idea to unique advantage: The entrepreneurial challenge of constructing a resource base. Academy of Management Executive, 15, 64–78. Bryce, D. J., & Singh, J. V. (2001). The future of the firm from the evolutionary perspective. In: P. J. DiMaggio (Ed.), The twenty-first century firm: Changing economic organization in international perspective (pp. 161–185). Princeton, NJ: Princeton University Press. Bulow, J. I., Geanakoplos, J. D., & Klemperer, P. D. (1985). Multimarket oligopoly: Strategic substitutes and complements. Journal of Political Economy, 93, 488–511. Cannella, A. A., & Holcomb, T. R. (2005). A multi-level analysis of the upper-echelons model. In: F. J. Yammarino, & F. Dansereau (Eds), Research in multi-level issues: The many faces of multi-level issues (Vol. 4, pp. 197–238). Oxford, UK: JAI/Elsevier Science. Cantwell, J. (2001). Innovation and information technology. In: A. M. Rugman, & T. L. Brewer (Eds), The Oxford handbook of international business (pp. 431–456). Oxford: Oxford University Press. Capron, L., & Pistre, N. (2002). When do acquirers earn abnormal returns? Strategic Management Journal, 23, 781–794. Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in populations of organizations. American Journal of Sociology, 90, 1262–1283. Caves, R. E. (1971). International corporations: Industrial economics of foreign investment. Economica, 38, 1–27. Caves, R. E. (1996). MNEs and economic analysis. Cambridge: Cambridge University Press. Chandler, A. D. (1962). Strategy and structure: Chapters in the history of the American industrial enterprise (1st ed.). Cambridge, MA: The MIT Press. Chen, M.-J. (1996). Competitor analysis and interfirm rivalry: Toward a theoretical integration. Academy of Management Review, 21, 100–134. Cohen, M. D., & Bacdayan, P. (1994). Organizational routines are stored as procedural memory: Evidence from a laboratory study. Organization Science, 5, 554–568.
Diversification to Achieve Scale and Scope
581
Conner, K. R., & Prahalad, C. K. (1996). A resource-based theory of the firm: Knowledge versus opportunism. Organization Science, 7, 477–501. Darr, E. D., Argote, L., & Epple, D. (1995). The acquisition, transfer, and depreciation of knowledge in service organizations: Productivity in franchises. Management Science, 41, 1750–1762. Das, S., Sen, P. K., & Sengupta, S. (1998). Impact of strategic alliances on firm valuation. Academy of Management Journal, 41, 27–41. Das, T. K., & Teng, B. S. (2000). Instabilities of strategic alliances. An internal tensions perspective. Organization Science, 11, 77–101. Delios, A., & Beamish, P. W. (1999). Geographic scope, product diversification, and the corporate performance of Japanese firms. Strategic Management Journal, 20, 711–727. Denis, D. J., Denis, D. K., & Yost, K. (2002). Global diversification, industrial diversification, and firm value. Journal of Finance, 57, 1951–1979. Denrell, J., Fang, C., & Winter, S. G. (2003). The economics of strategic opportunity. Strategic Management Journal, 24, 977–990. Dierickx, I., & Cool, K. (1989). Asset stock accumulation and sustainability of competitive advantage. Management Science, 35, 1504–1511. Dyer, J. H., & Singh, H. (1998). The relational view: Cooperative strategy and sources of interorganizational competitive advantage. Academy of Management Review, 23, 660–679. Echols, A., & Tsai, W. (2005). Niche and performance: The moderating role of network embeddedness. Strategic Management Journal, 26, 219–238. Eden, L., & Miller, S. (2004). Distance matters: Liability of foreignness, institutional distance and ownership strategy. In: M. A. Hitt, & J. L. C. Cheng (Eds), The evolving theory of the multinational firm: Advances in international management (Vol. 16). Amsterdam: Elsevier. Edwards, C. D. (1955). Conglomerate bigness as a source of power. In: Business concentration and price policy (National Bureau of Economic Research conference report) (pp. 331–352). Princeton, NJ: Princeton University Press. Eisenhardt, K. M. (1999). Strategy as strategic decision making. Sloan Management Review, 40, 65–72. Eisenhardt, K. M., & Martin, J. A. (2000). Dynamic capabilities: What are they? Strategic Management Journal, 21, 1105–1121. Errunza, V. R., & Senbet, L. W. (1984). International corporate diversification, market valuation, and size-adjusted evidence. Journal of Finance, 39, 727–743. Fiol, C. M., & Lyles, M. A. (1985). Organizational learning. Academy of Management Review, 10, 803–813. Franko, L. G. (1989). Global corporate competition: Who’s winning, who’s losing, and the R&D factor as one reason why. Strategic Management Journal, 10, 449–474. Galbraith, J. (1973). Designing complex organizations. Reading, MA: Addison-Wesley. Galunic, D. C., & Rodan, S. (1998). Resource recombinations in the firm: Knowledge structures and the potential for Schumpeterian innovation. Strategic Management Journal, 19, 1193–1201. Gimeno, J. (1999). Reciprocal threats in multimarket rivalry: Staking out ‘spheres of influence’ in the US airline industry. Strategic Management Journal, 20, 101–128. Gimeno, J., Hoskisson, R. E., Beal, B. D., & Wan, W. P. (2005). Explaining the clustering of international expansion moves: A critical test in the US telecommunications industry. Academy of Management Journal, 48, 297–319.
582
TIM R. HOLCOMB ET AL.
Gimeno, J., & Woo, C. Y. (1999). Multimarket contact, economies of scope, and firm performance. Academy of Management Journal, 42, 239–259. Grant, R. M. (1991). The resource-based theory of competitive advantage: Implications for strategy formulation. California Management Review, 33, 114–135. Grant, R. M. (1996). Toward a knowledge-based theory of the firm. Strategic Management Journal, 17, 109–122. Greve, H. R., & Baum, J. A. C. (2001). Introduction: A multiunit, multimarket world. In: J. A. C. Baum, H. R. Greve (Eds), Advances in strategic management: Multiunit organization and multimarket strategy (Vol. 18, pp. 1–28). Greenwich, CT: JAI Press. Gupta, A. K., & Govindarajan, V. (2000). Knowledge flows within multinational corporations. Strategic Management Journal, 21, 473–496. Gulati, R. (1995). Social structure and alliance formation patterns: A longitudinal analysis. Administrative Science Quarterly, 40, 619–652. Haleblian, J., & Finkelstein, S. (1999). The influence of organizational acquisition experience on acquisition performance: A behavioral learning perspective. Administrative Science Quarterly, 44, 29–56. Hamel, G. (1991). Competition for competence and inter-partner learning within international strategic alliances. Strategic Management Journal, 12, 83–103. Hamel, G., & Prahalad, C. K. (1994). Competing for the future (1st ed.). Cambridge, MA: Harvard Business School Press. Hannan, M. T., & Freeman, J. (1977). The population ecology of organizations. American Journal of Sociology, 82, 929–964. Harrison, J. S., Hitt, M. A., Hoskisson, R. E., & Ireland, R. D. (1991). Synergies and postacquisition performance: Differences versus similarities in resource allocations. Journal of Management, 17, 173–190. Harrison, J. S., & St. John, C. (1998). Strategic management of organizations and stakeholders (2nd ed.). Cincinnati: SouthWestern College Publishing. Haveman, H. A. (1993). Organizational size and change: Diversification in the savings and loan industry after deregulation. Administrative Science Quarterly, 38, 20–50. Hayward, M. L. A. (2002). When do firms learn from their acquisition experience? Evidence from 1990–1995. Strategic Management Journal, 23, 21–39. Hayward, M. L. A., & Hambrick, D. C. (1997). Explaining the premiums paid for large acquisitions: Evidence of CEO hubris. Administrative Science Quarterly, 42, 103–127. Helfat, C. E., & Peteraf, M. A. (2003). The dynamic resource-based view: Capability lifecycles. Strategic Management Journal, 24, 997–1010. Henderson, R., & Mitchell, W. (1997). The interactions of organizational and competitive influences on strategy and performance. Strategic Management Journal, 18, 5–14. Hennart, J. F., & Park, Y. R. (1993). Greenfield vs. acquisition: The strategy of Japanese investors in the United States. Management Science, 39, 1054–1070. Hill, C. W. L., Hitt, M. A., & Hoskisson, R. E. (1992). Cooperative versus competitive structures in related and unrelated diversified firms. Organization Science, 3, 501–521. Hitt, M. A., Bierman, L., Uhlenbruck, K., & Shimizu, K. (2006). The importance of resources for the internationalization of professional service firms: The good, the bad and the ugly. Academy of Management Journal, in press. Hitt, M. A., Dacin, M. T., Levitas, E., Arregle, J.-L., & Borza, A. (2000). Partner selection in emerging and developed market contexts: Resource-based and organizational learning perspectives. Academy of Management Journal, 43, 449–467.
Diversification to Achieve Scale and Scope
583
Hitt, M. A., Harrison, J. S., & Ireland, R. D. (2001). Mergers and acquisitions: Creating value for stakeholders. New York: Oxford University Press. Hitt, M. A., Hoskisson, R. E., & Ireland, R. D. (1994). A midrange theory of the interactive effects of international and product diversification on innovation and performance. Journal of Management, 20, 297–326. Hitt, M. A., Hoskisson, R. E., Johnson, R. A., & Moesel, D. D. (1996). The market for corporate control and firm innovation. Academy of Management Journal, 39, 1084–1119. Hitt, M. A., Hoskisson, R. E., & Kim, H. (1997). International diversification: Effects on innovation and firm performance in product diversified firms. Academy of Management Journal, 40, 767–798. Hoskisson, R. E., & Hitt, M. A. (1988). Strategic control systems and relative R&D investment in large multiproduct firms. Strategic Management Journal, 9, 605–621. Hoskisson, R. E., & Hitt, M. A. (1990). Antecedents and performance outcomes of diversification: A review and critique of theoretical perspectives. Journal of Management, 16, 461–509. Hrebiniak, L. G., & Joyce, W. F. (2001). Implementing strategy: An appraisal and agenda for future research. In: M. A. Hitt, R. E. Freeman, & J. S. Harrison (Eds), The Blackwell handbook of strategic management (pp. 602–626). Malden, MA: Blackwell Publishers Inc. Inkpen, A. C. (2001). Strategic alliances. In: M. A. Hitt, R. E. Freeman, & J. S. Harrison (Eds), The Blackwell handbook of strategic management (pp. 409–432). Malden, MA: Blackwell Publishers Inc. Inkpen, A. C., & Beamish, P. W. (1997). Knowledge, bargaining power, and the instability of international joint ventures. Academy of Management Review, 22, 177–202. Ireland, R. D., Hitt, M. A., & Vaidyanath, D. (2002). Alliance management as a source of competitive advantage. Journal of Management, 28, 413–446. Jin, C., & Holmes, R. M. (2006). The evolution of global R&D: Theory and evidence from China. In: H. Li (Ed.), Growth strategies of new technology ventures in China. Northhampton, MA: Edward Elgar Publishing, in press. Jones, G. R., & Hill, C. W. L. (1988). Transaction cost analysis of strategy-structure choice. Strategic Management Journal, 9, 159–172. Kale, P., Dyer, J. H., & Singh, H. (2002). Alliance capability, stock market response, and longterm alliance success: The role of the alliance function. Strategic Management Journal, 23, 747–767. Karim, S., & Mitchell, W. (2000). Path-dependent and path-breaking change: Reconfiguring business resources following acquisitions in the US medical sector, 1978–1995. Strategic Management Journal, 21, 1061–1081. Khanna, T., & Palepu, K. (1997). Why focused strategies may be wrong for emerging markets. Harvard Business Review, 75, 41–51. Kim, W. C., Hwang, P., & Burgers, W. P. (1989). Global diversification strategy and corporate profit performance. Strategic Management Journal, 10, 45–57. Knott, A. M. (2003). Persistent heterogeneity and sustainable innovation. Strategic Management Journal, 24, 687–705. Kogut, B. (1984). Normative observations on the international value-added chain and strategic groups. Journal of International Business Studies, 15, 151–167. Kogut, B., & Zander, U. (1992). Knowledge of the firm, combinative capabilities, and the replication of technology. Organization Science, 3, 383–397.
584
TIM R. HOLCOMB ET AL.
Koh, J., & Venkataraman, N. (1991). Joint venture formations and stock market reactions: An assessment in the information technology sector. Academy of Management Journal, 34, 869–892. Kotabe, M., Srinivasan, S. S., & Aulakh, P. S. (2002). Multinationality and firm performance: The moderating role of R&D and marketing capabilities. Journal of International Business Studies, 33, 79–97. Lane, P. J., & Lubatkin, M. H. (1998). Relative absorptive capacity and interorganizational learning. Strategic Management Journal, 19, 461–477. Lawrence, P., & Lorsch, J. (1967). Organization and Environment. Homewood, IL: Irwin Publishing. Lei, D., & Hitt, M. A. (1995). Strategic restructuring and outsourcing: The effect of mergers and acquisitions and LBOs on building firm skills and capabilities. Journal of Management, 21, 835–859. Lei, D., Hitt, M. A., & Bettis, R. (1996). Dynamic core competences through meta-learning and strategic context. Journal of Management, 22, 549–569. Lu, J. W., & Beamish, P. W. (2001). The internationalization and performance of SMEs. Strategic Management Journal, 22, 565–586. Lu, J. W., & Beamish, P. W. (2004). International diversification and firm performance: The S-curve hypothesis. Academy of Management Journal, 47, 598–609. Luo, Y. D. (2002). Contract, cooperation, and performance in international joint ventures. Strategic Management Journal, 23, 903–919. Mahoney, J. T., & Pandian, J. R. (1992). The resource-based view within the conversation of strategic management. Strategic Management Journal, 13, 363–380. Makadok, R. (2001). Toward a synthesis of the resource-based and dynamic-capability views of rent creation. Strategic Management Journal, 22, 387–401. March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2, 71–87. Markides, C. C., & Williamson, P. J. (1996). Corporate diversification and organizational structure: A resource-based view. Academy of Management Journal, 39, 340–367. Miller, D. (1993). The architecture of simplicity. Academy of Management Review, 18, 116–138. Miller, D. (2003). An asymmetry-based view of advantage: Towards an attainable sustainability. Strategic Management Journal, 24, 961–976. Miller, D., Eisenstat, R., & Foote, N. (2002). Strategy from the inside out: Building capability creating organizations. California Management Review, 44, 37–54. Miner, A. S., Amburgey, T. L., & Stearns, T. M. (1990). Interorganizational linkages and population-dynamics: Buffering and transformational shields. Administrative Science Quarterly, 35, 689–713. Morck, R., & Yeung, B. (1991). Why investors value multinationality. Journal of Business, 64, 165–187. Morrow, J. L., Sirmon, D. G., Hitt, M. A., & Holcomb, T. R. (2005). Creating value in the face of declining performance: Designing actions to effect organizational recovery. Paper presented at the 2005 Strategic Management Society, Orlando, FL. Murtha, T. P., Lenway, S. A., & Bagozzi, R. P. (1998). Global mind-sets and cognitive shift in a complex multinational corporation. Strategic Management Journal, 19, 97–114. Nelson, R. R., & Winter, S. G. (1982). An evolutionary theory of economic change. Cambridge, MA: The Belknap Press of Harvard University Press.
Diversification to Achieve Scale and Scope
585
Newman, K. L. (2000). Organizational transformation during institutional upheaval. Academy of Management Review, 25, 602–619. Oliver, C. (1990). Determinants of interorganizational relationships: Integration and future directions. Academy of Management Review, 15, 241–265. Pablo, A. L., Sitkin, S. B., & Jemison, D. B. (1996). Acquisition decision-making processes: The central role of risk. Journal of Management, 22, 723–746. Palich, L. E., Cardinal, L. B., & Miller, C. C. (2000). Curvilinearity in the diversificationperformance linkage: An examination of over three decades of research. Strategic Management Journal, 21, 155–174. Panzar, J. C., & Willig, R. D. (1981). Economies of scope. American Economic Review, 71, 268–272. Peng, M. W. (2001). How entrepreneurs create wealth in transition economies. Academy of Management Executive, 15, 95–108. Penrose, E. (1959). The theory and growth of the firm (3rd ed.). New York, NY: Oxford University Press Inc. Peteraf, M. A. (1993). The cornerstones of competitive advantage: A resource-based view. Strategic Management Journal, 14, 179–191. Peteraf, M. A., & Bergen, M. E. (2003). Scanning dynamic competitive landscapes: A marketbased and resource-based framework. Strategic Management Journal, 24, 1027–1041. Pitts, R. A., & Hopkins, H. D. (1982). Firm diversity: Conceptualization and measurement. Academy of Management Review, 7, 620–629. Poppo, L., & Zenger, T. (2002). Do formal contracts and relational governance function as substitutes or complements? Strategic Management Journal, 23, 707–725. Porter, M. E. (1980). Competitive strategy: Techniques for analyzing industries and competitors (1st ed.). New York, NY: The Free Press. Porter, M. E. (1985). Competitive advantage: Creating and sustaining superior performance (1st ed.). New York, NY: The Free Press. Porter, M. E. (1991). Towards a dynamic theory of strategy. Strategic Management Journal, 12, 95–117. Porter, M. E. (1996). What is strategy? Harvard Business Review, 74, 61–78. Powell, T. C. (2001). Competitive advantage: Logical and philosophical considerations. Strategic Management Journal, 22, 875–888. Prahalad, C. K., & Hamel, G. (1990). The core competence of the corporation. Harvard Business Review, 68, 79–90. Priem, R. L., & Butler, J. E. (2001). Is the resource-based view a useful perspective for strategic management research. Academy of Management Journal, 26, 22–40. Ramanujam, V., & Varadarajan, P. (1989). Research on corporate diversification: A synthesis. Strategic Management Journal, 10, 523–551. Robins, J. A., Tallman, S., & Fladmoe-Lindquist, K. (2002). Autonomy and dependence of international cooperative ventures: An exploration of the strategic performance of US ventures in Mexico. Strategic Management Journal, 23, 881–901. Robins, J., & Wiersema, M. F. (1995). A resource-based approach to the multibusiness firm: Empirical analysis of portfolio interrelationships and corporate financial performance. Strategic Management Journal, 16, 277–299. Rodriguez, P., Uhlenbruck, K., & Eden, L. (2005). Government corruption and the entry strategies of multinationals. Academy of Management Review, 30, 383–396.
586
TIM R. HOLCOMB ET AL.
Romanelli, E., & Tushman, M. L. (1994). Organizational transformation as punctuated equilibrium: An empirical test. Academy of Management Journal, 37, 1141–1166. Rowley, T., Behrens, D., & Krackhardt, D. (2000). Redundant governance structures: An analysis of structural and relational embeddedness in the steel and semiconductor industries. Strategic Management Journal, 21, 369–386. Rugman, A. M. (1981). Inside the multinationals: The economics of international markets. London: Croom Helm. Rumelt, R. P. (1974). Strategy, structure and economic performance. Cambridge, MA: Harvard University Press. Shan, W. J., Walker, G., & Kogut, B. (1994). Interfirm cooperation and startup innovation in the biotechnology industry. Strategic Management Journal, 15, 387–394. Singh, J. V., House, R. J., & Tucker, D. J. (1986). Organizational change and organizational mortality. Administrative Science Quarterly, 31, 587–611. Singh, J. V., & Lumsden, C. J. (1990). Theory and research in organizational ecology. Annual Review of Sociology, 16, 161–195. Sirmon, D. G., Hitt, M. A., & Ireland, R. D. (2007). Managing firm resources in dynamic environments to create value: Looking inside the black box. Academy of Management Review, in press. Smith, K. G., Grimm, C., & Gannon, M. (1992). Dynamics of competitive strategy. London: Sage. Sorenson, O. (2000). Letting the market work for you: An evolutionary perspective on product strategy. Strategic Management Journal, 21, 577–592. Stinchcombe, A. L. (1965). Social structure and organizations. In: J. G. March (Ed.), Handbook of organizations (pp. 153–193). Chicago, IL: Rand McNally. Tallman, S. B., & Yip, G. S. (2001). Strategy and the multinational enterprise. In: A. M. Rugman, & T. L. Brewer (Eds), The Oxford handbook of international business. Oxford: Oxford University Press. Teece, D. J. (1982). Towards an economic theory of the multiproduct firm. Journal of Economic Behavior and Organization, 3, 39–63. Teece, D. J. (1986). Profiting from technological innovation: Implications for integration, collaboration, licensing and public policy. Research Policy, 15, 285–305. Teece, D. J., Pisano, G., & Shuen, A. (1997). Dynamic capabilities and strategic management. Strategic Management Journal, 18, 509–533. Thomke, S., & Kuemmerle, W. (2002). Asset accumulation, interdependence and technological change: Evidence from pharmaceutical drug discovery. Strategic Management Journal, 23, 619–635. Vermeulen, F., & Barkema, H. G. (2001). Learning through acquisitions. Academy of Management Journal, 44, 457–476. Vermeulen, F., & Barkema, H. (2002). Pace, rhythm, and scope: Process dependence in building a profitable multinational corporation. Strategic Management Journal, 23, 637–653. Wan, W. P., & Hoskisson, R. E. (2003). Home country environments, corporate diversification strategies, and firm performance. Academy of Management Journal, 46, 27–45. Wezel, F. C., & Witteloostuijn, A. v. (2006). Scale and scope economies in the British motorcycle industry, 1899–1993. In: J. A. C. Baum, S. D. Dobrev & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 521–548). Oxford, UK: JAI/Elsevier. Williamson, O. E. (1975). Markets and hierarchies. New York, NY: Free Press.
Diversification to Achieve Scale and Scope
587
Williamson, O. E. (1985). The economic institutions of capitalism. New York, NY: Free Press. Winter, S. G. (2000). The satisficing principle in capability learning. Strategic Management Journal, 21, 981–996. Yan, A. M., & Gray, B. (1994). Bargaining power, management control, and performance in United States–China joint ventures: A comparative case study. Academy of Management Journal, 37, 1478–1517. Yin, X. L., & Zajac, E. J. (2004). The strategy/governance structure fit relationship: Theory and evidence in franchising arrangements. Strategic Management Journal, 25, 365–383. Zaheer, S. (1995). Overcoming the liability of foreignness. Academy of Management Journal, 38, 341–363. Zahra, S. A., Ireland, R. D., & Hitt, M. A. (2000). International expansion by new venture firms: International diversity, mode of market entry, technological learning, and performance. Academy of Management Journal, 43, 925–950. Zollo, M., & Winter, S. G. (2002). Deliberate learning and the evolution of dynamic capabilities. Organization Science, 13, 339–351.
This page intentionally left blank
588
PART IX: INDUSTRY EVOLUTION
589
590
DIRECT AND INDIRECT EFFECTS OF PRODUCT PORTFOLIO ON FIRM SURVIVAL IN THE WORLDWIDE OPTICAL DISK DRIVE INDUSTRY, 1983–1999 Olga M. Khessina ABSTRACT This paper explores how two understudied characteristics of a firm’s product portfolio, namely, aging of products and (non)innovativeness of products, affect firm survival. The influence of these product portfolio characteristics on organizational mortality can be observed both at the firm and at the industry levels. Paradoxically, the portfolio’s influence at the firm and at the industry levels may go in opposite directions. Specifically, I predict that portfolios with aging products make their firms weaker competitors and survivors. However by weakening these firms, ‘‘aging’’ portfolios reduce competitive pressures at the industry level and, therefore, improve firm survival indirectly by changing industry vital rates. In contrast, firms with innovative product portfolios should be stronger survivors. At the same time, they are likely to intensify competition in the industry and, as a result, diminish survival chances of all firms, including those with innovative products. The analyses of all Ecology and Strategy Advances in Strategic Management, Volume 23, 591–630 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23018-3
591
592
OLGA M. KHESSINA
firms’ product portfolios in the worldwide optical disk drive industry, 1983–1999, support these predictions.
INTRODUCTION Scholars have long recognized that the characteristics of a firm’s product portfolio in a given industry significantly affect its performance and survival chances in this industry (Clark & Wheelwright, 1993; Lancaster, 1990; Sorenson, 2000). The literature has focused on two characteristics of the product portfolio that may influence organizational outcomes. First, scholars have explored how the breadth of a product portfolio (also called product variety and product differentiation) impacts organizational performance and survival (Barnett & Freeman, 2001; Gilbert & Matutes, 1993; Lancaster, 1990; Sorenson, 2000). It was established that firms with larger portfolios tend to perform better (Kerke & Srinivasan, 1990; Lancaster, 1990; Swaminathan & Delacroix, 1991). This effect was especially pronounced in uncertain environments (Sorenson, 2000). Second, scholars have studied how the number of new products in existing portfolios affects organizational outcomes (Barnett & Freeman, 2001; Burton, 1994; Chaney, Devinney, & Winer, 1991; Dowell & Swaminathan, 2000; Stern & Henderson, 2004). It was found that while having new products in the portfolio is largely beneficial for a firm (Chaney et al., 1991; Dowell & Swaminathan, 2000), the process of adding many products simultaneously can be harmful (Barnett & Freeman, 2001). Although the literature agrees that characteristics of product portfolio exhibit significant influences on firm performance and survival, it tends to focus on immediate or direct effects of product portfolio on a firm’s outcomes. However, it is important to realize that a firm’s product portfolio may also generate ecological effects at the industry level that are felt both by other organizations in the industry and by the firm itself. Specifically, organizational ecology suggests that by influencing firm outcomes (e.g., making a firm a stronger or weaker competitor), the product portfolio may impact the environment (e.g., contribute to change in population vital rates) and, in this way, affect the firm and its competitors indirectly by changing the intensity of competition. The objective of this study is to understand the dynamics and consequences of direct and indirect effects of firm product portfolios. To achieve this goal, I look at two understudied characteristics of product portfolio: product portfolio age and product portfolio (non)innovativeness.
Direct and Indirect Effects of Product Portfolio
593
A firm’s product portfolio in a given industry is often composed of products of varying market age. Market age of a product may impact firm survival. On the one hand, products that stay on the market longer may generate greater financial returns. On the other hand, long-lived products are likely to become obsolete and drain firm resources. It is little known how combinations of products of different ages in a firm’s portfolio affect organizational outcomes in either direct or indirect way. One objective of this study is to understand how mixtures of products of different market longevity in a firm’s product portfolio affect its survival. The ‘‘innovativeness’’ of products in the portfolio may also have profound consequences for organizational outcomes. The literature showed that firms with more innovative products have an advantage over those with less innovative products (Carroll & Teo, 1996; Tushman & Anderson, 1986) and firms with a greater number of innovative products in their portfolios have an advantage over firms with a fewer number of such products (Barnett & Freeman, 2001; Khessina, 2003). However, not much attention has been paid to non-innovative products in firm portfolios. It is little known how the presence of such products in the portfolio affects firm survival. The last objective of this paper is to establish direct and indirect effects of innovative and non-innovative products in the firm’s portfolio on its survival and to understand what combination of innovative and non-innovative products can be beneficial or harmful. The rest of the paper proceeds as follows. First, I provide a brief review of the literature on effects of product portfolio on organizational outcomes. I argue that although profound characteristics of product portfolio for firm performance were identified, there are other important characteristics that have not been fully explored yet. Second, I develop the theoretical arguments regarding both direct and indirect effects of two understudied dimensions of product portfolios – product age and product (non)innovativeness. I propose hypotheses and describe the empirical setting of the worldwide optical disk drive industry during 1983 to 1999, on which I test these hypotheses. Then I explain the methodology. Finally, I describe empirical results and discuss them in the light of the existing literature and the developed arguments.
BACKGROUND The literature agrees that characteristics of a firm’s product portfolio affect its organizational outcomes. Two such characteristics have received
594
OLGA M. KHESSINA
thorough attention of scholars. First, a significant amount of research has been done to understand how the breadth of a product portfolio (also called product variety and product differentiation) affects firm performance and survival (for review, see Lancaster, 1990; Sorenson, 2000). Product variety may bring different advantages (for review, see Sorenson, 2000). First, firms with broader portfolios better meet customer preferences, which theoretically should allow them to charge higher prices and increase sales (Salop, 1979; Perloff & Salop, 1985). Supporting these ideas, empirical evidence shows that product portfolio breadth is positively associated with higher market share (Kerke & Srinivasan, 1990) and lower mortality rates (Barnett & Freeman, 2001; Dowell & Swaminathan, 2000; Greenstein & Wade, 1998; Khessina, 2003). Second, firms with broad product portfolios may create entry barriers for new firms (Gilbert & Matutes, 1993; Schmalensee, 1978; Stavins, 1995). By offering a variety of products, firms occupy niches that otherwise would be attractive to new entrants. Third, firms with multiple products may reduce competition through mutual forbearance (Barnett, 1993; Bernheim & Whinston, 1990; Haveman & Nonnemaker, 2000). Finally, firms that offer products in multiple market niches may fare better during bad times in overcrowded niches (Dobrev, Kim, & Hannan, 2001; Dobrev, Kim, & Carroll, 2002; Swaminathan & Delacroix, 1991). The advantages of a broad product portfolio are not universal. They can be overweighed by the loss in production efficiency due to diseconomies of scale created by manufacturing many different products in relatively small batches (Lancaster, 1990). Some empirical studies showed that product variety increases direct costs and inventories (Lubben, 1988). Other studies, however, found no evidence of these costs (Kerke & Srinivasan, 1990) and instead demonstrated beneficial effects of broad product portfolio on firm performance and survival (Barnett & Freeman, 2001; Dowell & Swaminathan, 2000; Kerke & Srinivasan, 1990; Khessina, 2003). These two opposite findings are not necessarily contradictory. For example, Sorenson (2000) suggested that broad product portfolios may be both beneficial and harmful for firm survival depending on the ecological dynamics of an industry. He proposed and demonstrated that product variety is less valuable for firm survival when the total number of products on the market increases. He also showed, however, that broad product portfolio becomes extremely beneficial in uncertain markets when the accurate prediction of demand is difficult and multiple products help firms detect shifts in customer preferences. Ecological literature on niche width (Baum & Singh, 1994; Freeman & Hannan, 1983) suggests another explanation as to why broad product
Direct and Indirect Effects of Product Portfolio
595
portfolios may have both beneficial and harmful effects on firm’s survival. Firms with broad product portfolios occupy broad resources niches that may affect firm survival in contradictory ways. On the one hand, firms operating in broad niches are able to spread risk and, as a result, improve their survival chances (Dobrev et al., 2001, 2002). On the other hand, firms occupying broad niches are more likely to experience resource overlap with niches of other firms and, consequently, suffer from more intense competition (Baum & Singh, 1994; Dobrev et al., 2001; McPherson, 1983; Podolny, Stuart, & Hannan, 1996). Based on the niche width research it is possible to suggest that because firms with broad product portfolios occupy broad niches, they can experience both favorable and harmful effects depending on the degree of their portfolio overlap with those of other firms. The second dimension of product portfolio, which received significant attention in the literature, is the number of new products in a firm’s portfolio. Scholars generally agree that the presence of new products in the firm portfolio is beneficial for firm performance and survival (Barnett & Freeman, 2001; Burton, 1994; Chaney et al., 1991; Dowell & Swaminathan, 2000; Khessina, 2003; Stern & Henderson, 2004). However, this positive effect of new products is not universal. It was found that ‘‘newness’’ of product portfolio is more beneficial for firms’ market value in hightechnology industries than the low-technology ones (Chaney et al., 1991). It was also shown that when there are fewer new product introductions at the industry level, the ‘‘newness’’ of a firm’s product portfolio is more beneficial for its survival than when the industry is overflowed with new products (Khessina, 2003; Stern & Henderson, 2004). Furthermore, when scholars distinguished between the ‘‘technology newness’’ and the ‘‘market newness’’ of product portfolio, they found that while the ‘‘technology newness’’ is largely beneficial, the ‘‘market newness’’ can be neutral or even harmful for firm growth rates (Meyer & Roberts, 1986). Additionally, even the ‘‘technology newness’’ is not always beneficial for organizational outcomes. It was shown that firms that stop manufacturing old products at the time or soon after they introduced new products, as well as firms that manufacture both old and new product generations for a long time, fail at a higher rate than firms that have a moderate temporal overlap between manufacturing of old and new products (Dowell & Swaminathan, 2000). Finally, a dynamic modeling of product ‘‘newness’’ in the portfolio revealed that while having many new products in the portfolio is largely beneficial for firm survival, adding such products simultaneously momentarily increases failure rates of firms and significantly decreases or even eliminates the advantage of having new products (Barnett & Freeman, 2001; Dowell
596
OLGA M. KHESSINA
& Swaminathan, 2000). This effect is observed because although new products may be very attractive to customers, the process of adding these products to the existing portfolios disrupts organizational routines and causes problems in organizational operation (Barnett & Carroll, 1995; Hannan, Po´los, & Carroll, 2003). In sum, the literature provides significant theoretical and empirical evidence about how the characteristics of firm product portfolios influence organizational outcomes. This literature also shows that effects of product portfolio on firm performance and survival are dynamic and complicated, because they may depend on firm and industry characteristics. Therefore, the issue of product portfolio warrants further investigation. In this study, I focus on two understudied characteristics of product portfolio. Specifically, I aim to understand how ‘‘aging’’ and ‘‘(non)innovativeness’’ of product portfolio affect organizational survival. I take an ecological approach to this issue and look not only at direct effects of product portfolio on firm survival, but also at indirect effects of product portfolio on organizational hazard through its influence on industrial ecology.
THEORETICAL ARGUMENTS AND HYPOTHESES Direct vs. Indirect Effects of Product Portfolio The literature on product portfolios focuses on immediate or direct effects of characteristics of a firm’s product portfolio on its performance and survival. It studies how specific portfolio features influence firm outcomes. Both conceptual and empirical analyses of firm portfolios are usually limited to the firm level. The literature tends to ignore that a firm’s product portfolio may have broader ecological effects on all firms participating in the industry, and, as a result, influence a focal firm not only in a direct immediate way, but also indirectly through change in the ecological dynamics of the industry. The tendency to focus on direct effects of firm characteristics and actions on organizational outcomes is not unique to the product portfolio literature. Most strategy research favors this approach as well. This is unfortunate. While direct effects of a firm’s actions may be stronger and more profound than indirect ecological effects, it is necessary to consider both in order to get a more precise picture of consequences of the firm’s strategic actions. For example, the ecological theory of Red Queen competition (Barnett & Hansen, 1996; Barnett & Pontikes, 2005) shows that firms facing competition look
597
Direct and Indirect Effects of Product Portfolio
for ways of improving their performance and when successful become stronger competitors as a result of organizational learning. These firms intensify competition in the industry and, consequently, trigger search for performance improvement in their competitors, which if successful become more formidable and intensify competition in the industry even further triggering farther learning in the first firms. This self-reinforcing process of competition and organizational learning shows that a focal firm’s actions aimed toward performance improvement affect this firm both directly by making it a stronger competitor and indirectly by prompting other firms to become stronger competitors. Whereas the direct effect of Red Queen competition is beneficial for the firm’s performance and survival, the indirect effect is harmful. The analysis of both is necessary to understand full consequences of firm strategic actions. I argue that a firm’s product portfolio exhibits both direct and indirect effects on organizational performance and survival. These effects are likely to work in opposite directions. Fig. 1 shows how direct effects of product portfolios at the firm level turn into indirect effects at the industry level. As firms develop product portfolios with beneficial features they become more formidable competitors and stronger survivors. Increasing number of firms that are stronger survivors because of advantages accrued in their product portfolios changes vital dynamics at the industry level. Specifically, more surviving firms, and especially more surviving firms that are stronger competitors, intensify competition for scarce resources at the industry level. More intense competition leads to increase in failure rates of firms populating the industry, which includes firms with advantageous product portfolios as well. Thus, while the direct effect of superior product portfolios on firm survival is positive, it can be significantly reduced by indirect harmful influences from intense competition generated by superior portfolios at the industry level. A different scenario should be observed when the industry gets populated by the increasing number of firms with product portfolios with inferior features. Such firms being weaker competitors and survivors decrease the intensity of competition in the industry, which in turn results in higher Direct Effect Features of Product Portfolio
Fig. 1.
Firm Survival Chances
Indirect Effect Industry Vital Rates
Intensity of Competition
Firm Survival Chances
Direct and Indirect Effects of Product Portfolio on Firm Survival.
598
OLGA M. KHESSINA
survival rates of industry participants, including those with inferior product portfolios. Thus, while firms with mediocre product portfolios suffer from higher failure rates, this direct effect of inferior product portfolio on survival is mitigated by indirect favorable dynamics generated from weakening competition at the industry level. I extend and test the developed arguments about direct and indirect effects of firm product portfolio on organizational survival to two understanding portfolio characteristics: aging of product portfolio and (non)innovativeness of product portfolio. I argue that the net effect of these portfolio features on firm survival can be evaluated only when both direct and indirect influences are taken into account. Aging of Product Portfolio Although the literature established significant relationships between the number of new products in a firm’s portfolio and organizational outcomes, it also showed that these relationships are complex and demand further investigation. An important question that has not received much attention yet is whether a portfolio with greater number of new products is always more beneficial than a portfolio with fewer new products. The management of innovation literature implies that the more new products in the portfolio, the better the firm performance (Henderson, 1993; Tripsas, 1997; Tushman & Anderson, 1986). However, other literatures suggest that a balanced portfolio of both new and old products is necessary to achieve a long-term profitability of the firm. Thus, Dowell and Swaminathan (2000) showed that in the U.S. early bicycle industry, firms had higher survival chances when they manufactured both old and new products. Production of old products helped firms to rely on old operational routines while new routines were being created. It is interesting that practitioners also recommend a balance between relatively mature products that can provide a stable current cash flow and new products that may generate cash flow in the future (e.g., the Boston Consulting Group’s growth share matrix). Thus, the issue of new products in portfolio is not as straightforward, as many assume. To understand how the combination of mature and new products in a firm’s portfolio affects organizational outcomes, I explore effects of ‘‘aging’’ of a firm’s product portfolio on its survival. I conceptualize product portfolio ‘‘aging’’ in two complementary ways. First, I look at the average age of all products currently present in the firm’s portfolio. Then, I explore the effects of spread of product ages, i.e., age range between the newest and the oldest products in the firm’s portfolio.
Direct and Indirect Effects of Product Portfolio
599
The average age of portfolio products describes how young or how old on average products in the firm’s portfolio are. Although no research has investigated directly how the average age of products affects firm survival, valuable insights can be derived from studies that explored how an age of a specific product affects organizational outcomes. There are two opposing views with regard to effects of product market longevity on firm performance and survival. One group of researchers assumes that product longevity brings benefits to a firm. They believe that longer lived products generate higher financial returns than products that disappear from the market quickly (Reinganum, 1983). Firms with longlived products are also thought to create and maintain entry barriers discouraging potential competitors from entering into less crowded niches (Gilbert & Matutes, 1993; Schmalensee, 1978; Stavins, 1995). Finally, products that stay on the market longer are believed to help a firm better detect shifts in customer preferences (Sorenson, 2000). According to these views, a firm with long-lived products should have higher performance and survival than a firm with short-lived products. However, another group of scholars suggests that long-lived products are more likely to be a liability than a benefit for a firm. The longer products stay on the market, the more obsolete and out-of-the-date they become (Greenstein & Wade, 1998; Tushman & Anderson, 1986). Such products generate hardly any financial returns, whereas their manufacturing and marketing drain firm resources (Bayus, Erikson, & Jacobson, 2003; Dowell & Swaminathan, 2000). Furthermore, old products embody organizational learning from competitors from the distant past, which is usually less beneficial and more maladaptive than learning from competitors from the recent past (Barnett & Hansen, 1996). It seems that these two standpoints about the effects of product longevity on firm performance are not contradictory but rather complementary. A product has to stay on the market for a certain period of time in order to justify the development costs and generate revenues. However, a product that stays on the market too long is likely to become obsolete and unpopular among customers and as a result drain firm resources. Thus, the likely relationships between product longevity and firm performance and survival are not linear, but rather curvilinear. I suggest that average age of products in a firm’s portfolio should affect organizational survival in a non-linear fashion as well. Firms with portfolios filled with products that are mostly very young and new to the market, either because the firms are new to the market or because they constantly introduce new products and quickly withdraw the old ones, may have
600
OLGA M. KHESSINA
difficulties to generate financial returns when their products do not stay long enough on the market to justify developmental and promotional costs. In contrast, firms with portfolios containing on average rather old products are at risk of becoming obsolete and not meeting customer demands. Therefore, firms should derive the largest survival advantage when average age of products in their portfolios is neither extremely young nor old: Hypothesis 1. Firm failure rates have a U-shape relation with average age of firm products: first firm failure rates decrease with increasing average age of its products, then firm failure rates increase with increasing average age of its products. It is important to emphasize that the strength of the U-shaped dependence between average age of products in the portfolio and firm failure rates should vary with changes in the environment. In slow-paced industries, where products become obsolete very slowly, the U-shaped dependence should be strong and well pronounced, with an inflection point further on the right. In contrast, in high-velocity environments, where products become obsolete very fast, the U-shaped dependence should be weak, with an inflection point further on the left. The average age of products in the portfolio may exert not only a direct impact on organizational survival by affecting manufacturing costs and firm profits. I suggest that age of product portfolio may also affect firm survival indirectly through changing the vital rates of firms in the industry. If Hypothesis 1 is valid, then firms with old product portfolios are weaker competitors, because their products are obsolete and not attractive to customers. Weaker competitors do not exert strong pressures on other firms in the industry and so make competition less intense (Barnett, 1997). Therefore, it is possible to infer that as average product age in the industry increases, competitive pressures exerted both by old products and by firms producing mostly old products decrease. As a result, survival rates in the industry increase with increasing age of firm product portfolios. Hypothesis 2. The greater the average age of products in the industry, the lower failure rates of firms in the industry. The average age of products indicates how young or old on average products in the firm’s portfolio are. The average product longevity, however, fails to distinguish between portfolios filled with many medium-aged products and portfolios containing the mixture of very old and very new products. Therefore, in order to estimate effects of ‘‘aging’’ of product portfolio more precisely, it is also necessary to look at how the age spread
Direct and Indirect Effects of Product Portfolio
601
between the newest and the oldest products in the firm’s portfolio affects its survival. The spread of product age in the portfolio may range from very narrow to very wide. Portfolios containing products from the same generation only have very narrow age spread; portfolios filled with products from distant generations have wide age spread. Drawing an analogy from the ecological niche theory (Freeman & Hannan, 1983), it is possible to describe firms that offer products from one generation as having narrow ‘‘temporal’’ scope and firms shipping products from multiple generations as occupying broad ‘‘temporal’’ scope. Although hardly any empirical research on relationships between temporal scope in firm product portfolios and organizational outcomes was done (for exception see Dowell & Swaminathan, 2000), the ecological literature on niche width and Red Queen competition and the economics literature on product cannibalization offer some valuable insights about how product age spread may affect firm survival. It seems that temporal scope of product portfolio may influence firms in contradictory ways. On the one hand, organizations offering products only from one generation, because they discard old products as soon as the new ones are introduced, may suffer from the discontinuity in production and promotion routines (Dowell & Swaminathan, 2000). They also run into a risk of failure if the new product generation does not succeed on the market. In contrast, firms with broad temporal scope portfolios spread risk of failure of new products by backing up their market position with older proven products. Such firms also enjoy continuity in routines between product generations which helps a smoother transition from one generation to another (Hannan & Freeman, 1984; Hannan et al., 2003). Additionally, firms with broad temporal scope possess greater absorptive capacity which facilitates innovation process (Cohen & Levinthal, 1990). On the other hand, firms with narrow temporal scope in their portfolios can be at a greater advantage than firms with broad temporal scope. Firms with narrow temporal scope can focus their resources and attention on manufacturing and promoting products from the same generation. As a result, they are more likely to enjoy greater specialization and efficiency due to economies of scale. In contrast, firms with a broad temporal scope have to divide their resources and attention between different generations of products that can be quite distant from each other in terms of technology, manufacturing, and marketing (Bayus et al., 2003; Dowell & Swaminathan, 2000). Additionally, by offering products from both old and new generations, such firms run into risk of cannibalizing their products (Greenstein
602
OLGA M. KHESSINA
& Wade, 1998; Reinganum, 1983; Stavins, 1995). Furthermore, firms occupying multiple temporal niches face greater competition (Baum & Singh, 1994; Dobrev et al., 2001). Although a positive side-effect of exposure to intense competition is organizational learning (Barnett, Greve, & Park, 1994), firms with broad temporal scopes may fall victims to maladaptive learning because they have to compete with varied cohorts of rivals (Barnett & Hansen, 1996). In sum, exposure to greater competition and inferior efficiency are likely to cause firms with broad temporal scope product portfolios fail at a higher rate than firms with narrow ones. Since the literature assumes that both narrow and broad temporal scopes of product portfolios can be both beneficial and harmful, it is reasonable to suggest a curvilinear relationship between temporal scope of product portfolio and firm survival. A firm’s survival chances should improve as its temporal scope increases, because it helps to mitigate a risk of introducing a new product to the market, on the one hand, and ensure continuity in production and marketing routines, on the other. However, as the temporal scope of a firm’s product portfolio becomes too broad, its survival chances decrease, because the firm spreads its resources too thin, becomes inefficient, and faces greater competition but does not learn from it. Hypothesis 3. Firm failure rates have a U-shape relation with age spread of firm products: first firm failure rates decrease with increasing age spread of its products, then firm failure rates increase with increasing age spread of its products. As in the case with an effect of average age of product portfolio on firm survival, the strength of the U-shaped dependence between the temporal scope of firm product portfolio and firm failure rate should vary with change in the environment. In slow-paced industries, where new generations of products are rare and take a long time to develop, the U-shaped dependence should be strong and well pronounced, with an inflection point further on the right. In contrast, in high-velocity environments, where new generations of products are introduced very often, the U-shaped dependence should be weak, with an inflection point further on the left. The temporal scope of a firm’s product portfolio affects a firm’s survival not only directly but also indirectly by changing the intensity of competition in the industry. As argued above, firms with very broad temporal scope are weaker competitors, both because they are less efficient (Barnett, 1997; Bayus et al., 2003; Dowell & Swaminathan, 2000) and because they do not learn as adaptively as firms with narrower temporal scope (Barnett & Hansen, 1996). As the industry gets filled with weaker competitors in the
Direct and Indirect Effects of Product Portfolio
603
form of firms with broad temporal scope in their product portfolios, the intensity of competition should decrease, and as a result, survival rates of firms should increase. Additionally, broad temporal scope at the industry level indicates the presence of multiple niches that address different types of customers: customers that prefer to pay more for newer products and customers that prefer to buy older products if the price is significantly lower than that of new ones (de Figueiredo & Kyle, 2006; Podolny & Stuart, 1995). Such market partitioning may reduce competition as well, and consequently improve firm survival chances: Hypothesis 4. The greater the age spread of products in the industry, the lower failure rates of firms in the industry. (Non)Innovativeness of Product Portfolio How does technological advancement of products in a firm’s portfolio affect its survival? The management of innovation literature suggests that more innovative firms experience better performance and survival chances, because they create operational processes that allow them to be more efficient and products that generate greater consumer demand (Christensen & Rosenbloom, 1995; Henderson & Clark, 1990; Tripsas, 1997; Tushman & Anderson, 1986). Following this literature, it makes sense to suggest that the more innovative products in a firm’s portfolio, the higher its performance and survival chances, because innovative products are more attractive to customers and are likely to generate positive financial returns. On the other hand, having too many innovative products may come at a cost. Developing such products requires significant investment of resources. Furthermore, creating and promoting innovative products generate structural strains in firms. They have to change their internal routines of production and labor relationships, and links with outside actors (Christensen & Rosenbloom, 1995; Khessina, 2003; Tushman & Anderson, 1986). Such structural changes create disruptions in innovating organizations and cause distraction of managerial attention and drainage of resources, and as a result, temporarily increase organizational risk of failure (Barnett & Freeman, 2001; Hannan & Freeman, 1984; Hannan et al., 2003). Additionally, firms that offer many innovative products experience niche overlap with more and stronger competitors and, as a result, suffer from more intense competition (Barnett & McKendrick, 2004; Dobrev et al., 2001). Therefore, whereas having innovative products in the portfolio is largely beneficial for a firm, offering too many of these products may become harmful:
604
OLGA M. KHESSINA
Hypothesis 5. Firm failure rates have a U-shape relation with firm number of innovative products: first firm failure rates decrease with increasing number of innovative products, then firm failure rates increase with increasing number of innovative products. Innovativeness of a firm’s product portfolio may affect its survival also in an indirect way by changing the competitive intensity in the industry. The literature shows that innovative products are often more attractive to customers (Christensen & Bower, 1996; Henderson & Clark, 1990). They do not require as much market promotion as obsolete products (Bayus et al., 2003). Consequently, innovative products are likely to generate higher sales and profit rates, and to obtain greater market shares (Bayus et al., 2003; Roberts, 1999; Tushman & Anderson, 1986). By these reasons, it is logical to suggest that innovative products generate stronger competitive pressures in the industry than non-innovative ones. Increasing intensity of competition affects negatively survival rates of all firms, including those with innovative products. Therefore, although a firm’s innovative products increase its survival directly, the same products decrease this firm survival indirectly, by contributing to intense competition at the industry level. Hypothesis 6. The greater the number of innovative products in the industry, the higher failure rates of firms in the industry. The literature says that it is beneficial to be innovative. Specifically, firms that innovate at the technological frontier generate higher sales (Tushman & Anderson, 1986), get higher market shares (Henderson, 1993), acquire market leadership (Christensen & Rosenbloom, 1995; Henderson & Clark, 1990), and survive longer (Barnett & Freeman, 2001; Carroll & Teo, 1996). However, it is less understood what it means for a firm to be non-innovative and what the consequences of non-innovativeness are. I address this issue again from the perspective of a firm’s product portfolio. Podolny and Stuart (1995) and de Figueiredo and Kyle (2006) suggest that more advanced and less advanced technologies compete on difference bases and address different types of customers. Customers who prefer more advanced technology are willing to pay more. There are, however, customers, who are ready to sacrifice superiority in technology for a substantially lower price. This research suggests that non-innovative products do not directly compete with innovative and can be quite successful in their respective niche. Therefore, it is possible to predict that the presence of noninnovative products in the portfolio is not always harmful, as the management of innovation literature implies, but can be beneficial when the
Direct and Indirect Effects of Product Portfolio
605
demand for such products exists. The danger with less advanced technology, however, is that it becomes obsolete even faster than more advanced technology. Thus, in conditions of high technological pace a firm risks market failure if it focuses too much on manufacturing of technologically nonadvanced products. Therefore, it is possible to predict that firm failure rates will have a U-shaped relation with the firm number of non-innovative products: first firm failure rates decrease with increasing number of noninnovative products, then firm failure rates increase with increasing number of non-innovative products. Although, having both innovative and non-innovative products in the firm portfolio is predicted to have a U-shaped relation with firm failure rates, the U-shaped for non-innovative products should be more pronounced and with an inflection point further on the left than for innovative products. Even if a niche for non-innovative products exists, a firm likely derives fewer benefits from such products than from innovative, because it has to charge lower prices and compete with other multiple technological laggards. These benefits also last for a shorter period of time, since non-innovative products become obsolete faster. Finally, indirect effects of non-innovative products at the industry level are unlikely to harm all firm survival, because they are weaker competitors than innovative products. Hence, under certain conditions, the presence of non-innovative products in a firm’s portfolio can be beneficial both at the firm and at the industry level.
METHODOLOGY Research Setting of Worldwide Optical Disk Drive Industry I test the developed hypotheses on the population of all optical disk drive producers that operated in the worldwide market from the beginning of the industry in 1983 through 1999, which is the last year of full coverage in the most comprehensive data source available. The key product of the optical disk drive industry is the optical disk drive. An optical disk drive is one of the many devices (e.g., hard drives, floppies, tapes, disk arrays) used for the storage and retrieval of information. The most familiar examples of optical disk drives are CD-drives and DVD-drives. The optical method for data storage is based on the recording and retrieval of information with the help of a laser. The optical disk drive industry is an appropriate setting for testing the arguments regarding effects of firms’ product portfolios on firm survival for
606
OLGA M. KHESSINA
several reasons. First, the data on this industry cover every product shipped by every producer during 1983 to 1999 making it possible to avoid sample selection biases. Second, this industry has experienced high rates of product innovation that makes it possible to test hypotheses about effects of innovative product portfolios. Finally, the optical disk drive industry is worldwide, insuring that the results of this study are generalizable beyond the national borders of a particular country. Using the optical disk drive industry as a research setting also has several limitations coming from the specific nature of this industry. The industry is highvelocity; it is capital intensive; it experienced many format wars but very few accepted standards. Consequently, the results of this study can be generalized only to the similar classes of industries to that of the optical disk drive industry. Background on Optical Disk Drives Technology Optical storage systems consist of two main components: a storage medium in a rotating disk form and a drive unit for recording, retrieval, and output (Purcell, 2000). An optical disk consists of four layers: a polycarbonate substrate layer, a reflective layer, a protective layer, and the label. Optical storage media use the intensity of reflected laser light as an information source. In the polycarbonate substrate layer, a laser beam encounters holes that correspond to the coded data, which are called pits. The areas between these pits are called lands. The substrate layer is covered with a thin reflective layer. The laser beam is focused on the reflective layer from the substrate layer. The reflected beam has a strong intensity at the lands and a weak intensity at the pits. An optical disk consists of a sequential order of pits and lands allocated in one spiral track. The process of optical recording and the retrieval of information can be described as follows. Information is stored on a polycarbonate disk in the form of pits. During recording, pits are generated by a laser beam. The stored digital information can later be retrieved by an optical disk drive. The drive’s optical pickup creates a laser beam directed at the spinning disk. Logic timing circuits can register the difference between distance the light travels when it strikes lands and distance the light travels when it strikes pits. The pattern composed of pits and lands corresponds to the coding of 1s and 0s. The reflected signals are directed to a processor that reads the reflection and converts it into a stream of digital pulses, which in turn are converted into text, pictures, or sounds. The entire system is controlled via a microprocessor-based central processing unit.
Direct and Indirect Effects of Product Portfolio
607
Brief History of the Optical Data Storage Technology In 1972, Philips Corporation announced a method of optical storage of audio content based on analog modulation techniques. The analog modulation approach was soon abandoned in favor of more promising digital signal encoding methods. During the same period, Sony Corporation was engaged in research to perfect error-correction methods that could be applied to digitally encoded audio. Collaboration between Sony and Philips resulted in the merging of Philips’s signal format with Sony’s errorcorrection method, and in June of 1980, the two companies introduced their proposal for the Compact Disc Digital Audio system. The proposed standard was adopted by 25 manufacturers and efforts shifted toward retooling the industry to support manufacturing products incorporating the new standard. Adoption of the optical method for audio storage was paralleled by efforts of Philips, Sony, NEC, and other companies to develop techniques for storing data on disk. The result of these efforts was the CD-ROM (Compact Disk-Read Only Memory) format, which was introduced in 1985. Initially the costs and dismal performance of the first optical disk drives discouraged many potential users. However, further development drove costs down and improved performance. In 1986, a number of industry representatives agreed upon a common file system structure named the High Sierra format, which was formalized as ISO 9660 standard in 1988. The next-generation device, which was introduced in the mid-1980s, provided a flexible write-once, read-many (WORM) capability. This enabled end-users to record and playback computer data from the same drive. The third-generation optical disks, which are today’s rewritable systems, were introduced in 1988. They offer record, playback, and erase capabilities. Two different digital videodisk formats emerged in January 1995. One camp, led by Toshiba, introduced the Super Density format. Sony and Philips devised their own approach – the Multi Media Compact Disc. In December 1995, the charter for the DVD Consortium was drawn up and dissension among the industry leaders diminished as the standard for the Digital Versatile Disk (DVD) was formalized. The first DVD players were shipped in 1996. The industry has been always characterized by format wars. The Optical Storage Technology Association (OSTA) was established in 1992 with a goal to end format wars by promoting industry standards that would allow compatibility across different types of drives and manufacturers. In 1997, OSTA developed MultiRead specification that enables all classes of CD disks to be read on current and future CD and DVD devices. The efforts of
608
OLGA M. KHESSINA
OSTA to promote the common standard succeeded in 2000 when 17 CD drive manufacturers, representing over 90% of all CD optical drive shipments worldwide, have achieved compliance with MultiRead specification. Operationalization of Variables Starting Events of Production I defined a firm’s entry into the optical disk drive industry as occurring when the firm shipped its first optical disk drive product to the customer market.1 The Disk/Trend report provides information on the first customer shipment in varying degrees of precision. Disk/Trend gives some dates with precision to the month, others with precision to the quarter, and still others with precision to the year. To make the analysis tractable, all the information about timing was converted to decimal years. Dates given to the month were coded as occurring at the beginning of the month. Following Petersen’s (1991) recommendations for dealing with time aggregation, dates given to only the quarter were coded as occurring at the midpoint of the quarter. Dates given to only the year were coded as occurring at the midpoint of the year. Ending Events of Production I defined a firm’s exit from the optical disk drive industry as occurring when the firm stopped shipping its optical disk drive products. The Disk/Trend report does not provide exact information on the last customer shipment of the product. The report comes out in the third quarter of each year. It covers revenues and unit shipment for the previous calendar year, but it covers firms and products for the current year. Based on this information, I assumed that the last shipment of the product happens in the third quarter of the year the product was last mentioned in a Disk/Trend report and coded a firm’s exit as occurring at the midpoint of the third quarter of the last year the firm shipped its last product. From 1983 to 1999, 109 firms entered the worldwide optical disk drive industry, and 66 exited, of which 59 disbanded and 7 exited via either merger or acquisition. The data include 635 firm-year observations. These firms shipped 1,323 products on the worldwide optical disk drive market, of which 1,019 products exited the market. Dependent Variable The ‘‘dependent variable’’ in this paper is firm failure in the optical disk drive industry. Firm failure includes firm disbanding from the optical disk
Direct and Indirect Effects of Product Portfolio
609
drive industry. Firm disbanding is considered to happen in year t, if the firm did not ship any optical disk drive product in year t+1 and the exit did not happen by either a merger or acquisition.2 Independent Variables There are primary independent variables in this study. Several variables measure aging of firm product portfolio and aging of products in the whole industry. The others measure ‘‘innovativeness’’ of product portfolios. Unless otherwise noted, all the variables are updated annually. Product Age. Firm’s average product age is the mean age of all products shipped by a firm to the market in a given year. It is measured in years. A squared form of this variable is used for testing the predicted U-shaped relationships between portfolio average product age and firm failure rates. To test for indirect effects of firms’ product portfolio average product age on organizational survival, I constructed the variable industry average product age, which is the mean age of all products shipped in the industry in a given year. It is also measured in years. Firm’s spread of product ages is the difference between market ages of the firm’s youngest and oldest products in a given year. A squared form of this variable is used for testing the predicted U-shaped relationships between a firm’s portfolio age spread and its failure rates. Industry spread of product ages, measured as the difference between market ages of the oldest and the youngest products in the industry in a given year, is used to test for an indirect effect of portfolio temporal scope on organizational survival. Both variables are measured in years. Product Portfolio Innovativeness. Innovativeness of a product portfolio is defined by technological advancement of its products. A product is technologically advanced or innovative if it has technical performance close to the technological frontier. The technological frontier might be defined in several ways. I constructed an endogenous measure of the technological frontier using the product with the best performance in the industry in a given year as a point of reference for this year (cf. Tushman & Anderson, 1986). I use an optical disk drive’s data access time as a performance parameter. Data access time is the physical operation associated with positioning the read/write head of a storage device in the proper location to read or write a particular piece of data. It is one of few important indicators of optical disk drive performance (Disk/Trend, 1999; Merrill Lynch & Co., & McKinsey
610
OLGA M. KHESSINA
and Company, 2001; Purcell, 2000). Smaller (i.e., faster) data access time signifies better optical disk drive performance. To make the effects of product data access time across different years easily interpretable, I standardize its measure by dividing a product’s data access time in each year by the industry’s mean data access time in the year. This standardized measure of data access time is used in all analyses. I define a product as innovative if its data access time is near the technological frontier, i.e., it is among the best in the industry. Products near the technological frontier in a given year are defined as those with data access time in the top 15% of the industry’s performance in a given year. The (non)innovativeness of product portfolio is measured by two variables. Firm number of top 15% products is a number of a firm’s products with data access time among top 15% in the industry in a given year. Firm number of bottom 84% products is a firm’s number of products with data access time among bottom 84% in the industry in a given year. Squared terms of these two variables are used for testing the predicted U-shaped relationships between innovative and non-innovative products in a firm’s portfolio and its failure rates. Product Densities. To look at indirect effects of product portfolio (non) innovativeness on firm survival, I constructed two variables. Density of top 15% products is the number of products with top 15% data access time in the industry in a given year. Density of bottom 84% products is the number of products with bottom 84% data access time in the industry in a given year. Organizational Controls Organizational characteristics other than innovation outcomes affect a firm’s probability to fail. Several organizational controls are used to account for these influences. Unless otherwise noted, all controls are updated annually. Age-dependence makes it necessary to control for organizational tenure in the industry while estimating failure rates (Hannan, 1998). Firm tenure is measured as the number of years a firm has operated in the optical disk drive industry. Larger firms tend to be stronger survivors than smaller firms (Barnett, 1997; Freeman, Carroll, & Hannan, 1983). To account for these differences organizational size is controlled. I constructed a measure of the firm’s size as scale of operations, specifically, as a firm’s annual revenue in millions of U.S. dollars from its sale of optical disk drives. The measure is logged to reduce skewness. Previous literature established that firms entering an industry de novo have higher mortality rates than diversified entrants, because they have
Direct and Indirect Effects of Product Portfolio
611
inferior resources and experience compared to de alio entrants (Carroll, Bigelow, Seidel, & Tsai, 1996; Mitchell, 1994). To control for differences in survival due to entry mode, I constructed the variable de novo firm dummy that takes the value of one if a firm entered the worldwide optical disk drive industry as a start-up, and the value of zero if a firm diversified into this industry from another market. This variable is time-invariant. The literature also found that de alio firms may differ in their pre-entry experience and resources which may affect their survival chances in the industry of new entry (Carroll et al., 1996; Dobrev, Kim, & Solari, 2004; Klepper & Simons, 2000). To control for differences in the amount of pre-entry experience, I constructed a variable firm’s age at the time of entry into the industry which is the number of years elapsed since the year a firm was first founded in any industry until the year a firm entered into the optical disk drive industry. The variable is time-invariant. Firms with larger product portfolios tend to have higher survival rates (Barnett & Freeman, 2001; Dowell & Swaminathan, 2000). To control for this influence, I constructed the variable firm’s total number of products on the market, which is the total number of products that a firm ships to the market in a given year. The optical disk drive industry was always dominated by large Japanese producers. The Japanese headquarters dummy takes the value of one if a firm has headquarters in Japan and zero if otherwise and is controlled to account for distinctiveness attributed to Japanese manufacturers (Gerlach & Lincoln, 2000; West, 2002) Industry Level Controls Several variables are used to control for industry processes. Unless otherwise noted, all variables are updated annually. Three variables are used to control for environmental conditions that may affect firm failure rates. The variable worldwide industry revenues measured in millions of U.S. dollars describes the realized demand for the product. The variable PC unit shipments measured in millions of units provides information about the size of the key (largest) market for optical disk drives. Number of patents granted in ODD technology, which is the number of distinct patents granted in a given year in any of the eight patent subclasses that cover different components of an optical disk drive system (for the list of patent subclasses see Rosenkopf & Nerkar, 2001), measures technological crowding of the industry. Density of all products measured as the number of products on the market in a given year is created to control for competitive pressures generated by product proliferation on the market (Greenstein & Wade, 1998).
612
OLGA M. KHESSINA
Organizational density was found to have a profound effect on firm survival (Hannan & Carroll, 1992; Carroll & Hannan, 2000). Density all firms is defined as the number of firms alive in a given year. Square density term is not significant and is not included in the reported analyses. This variable is lagged one year. Aside from contemporaneous density, density at a firm’s entry into the industry may affect subsequent failure rates. It was found that organizational density at a firm’s entry may have a competitive effect on the firm throughout its life leading to increasing exit rates over time (Carroll & Hannan, 1989). The variable density at entry all firms is defined as organizational density in the year that a firm shipped its first optical disk drive product to the market. Its value does not vary with time. The presence or absence of technological standards may affect firm longevity on the market (Baum, Korn, & Kotha, 1995; Suarez & Utterback, 1995). The optical disk drive industry has been always populated by multiple incompatible formats that generated format wars and sometimes reduced industry growth (Disk/Trend, 1999; Purcell, 2000). Different formats were introduced regularly across the history of the industry. However, only the High Sierra format became a widely adopted industry standard. It was formalized as ISO 9660 standard in 1988. So, the variable standard ISO 9660 period dummy, which takes a value of one for years 1988–1999, and zero otherwise, is created to control for influences of this standard. Model Specification To test the hypotheses that make predictions about effects of characteristics of product portfolio on firm survival, I use continuous-time event-history analysis. I treat a firm as the unit at risk, and the ‘‘dependent variable’’ is the instantaneous rate of a firm’s disbanding defined as: rðtÞ ¼ lim
Dt!0
Prob½toTot þ DtjT4t Dt
where T is a random variable for the time of the event of interest, t the time elapsed since the time a firm entered into the optical disk drive industry, and P(.) the conditional probability that the firm will disband over the interval [t, t+Dt] given that the firm did not disband at time t. I use a piecewise exponential function to represent variation in the timing of a firm’s failure rate to allow a flexible specification of duration-dependence.
613
Direct and Indirect Effects of Product Portfolio
"
rðuÞ ¼ exp
q X
mp Ap ðuÞ
p¼0
#
where Ap(u) ¼ 1 when u belongs to interval (up, up+1) and 0 if otherwise. A piecewise exponential model splits the time-axis into time pieces determined by an analyst (Carroll & Hannan, 2000, pp. 136–138). After examining life tables and exploring estimates of a variety of choices of the breakpoints, I decided to break the duration scale in years in four time pieces: 0–1, 1–4, 4–8 and 8, and greater. To test the hypotheses, I specified a firm’s disbanding rate r(u,t) as a function of firm tenure in the industry (u), firm size (S), the number of covariates measuring characteristics of firm product portfolio (P), the number of covariates measuring product characteristics at the industry level (I), and the other measured covariates (X). The general class of models I estimate has the form: X X X ln ri ðu; tÞ ¼ mp þ aS it þ gX nit dk I kit þ jl Plit þ l
k
n
where mp denotes tenure-specific effects, Sit size for firm i at time t, the l time-varying covariates measuring characteristics of firm product portfolio are summarized in Plit, the k time-varying covariates measuring product characteristics at the industry level are summarized in Ikit, and Xnit summarizes all other time-varying covariates. I estimated piecewise exponential models using the method of maximum likelihood as implemented with a user-defined routine in STATA (Sørensen, 1999). To estimate rate models with time-varying covariates, I constructed split-spell data breaking observed durations in year-long periods with the values of covariates updated every year.
FINDINGS Table 1a provides descriptive statistics of the variables. Table 1b is a correlation matrix between key variables. The file contains multiple spells for each firm, so the descriptive statistics does not always reflect intuitively the experiences of firms in the industry. Tables 2 and 3 present estimates of piecewise exponential hazard rate models designed to test the proposed hypotheses. Table 2 tests Hypotheses 1–4, the arguments about effects of product portfolio aging on firm survival. Model 2.1 provides a baseline for key covariates. Organizational-level variables exhibit common effects. Coefficients
614
OLGA M. KHESSINA
Table 1a.
Descriptive Statistics for Optical Disk Drives Producers: Split-Spell File. Mean
Firm tenure in the industry (in years) (t) Firm age at entry into ODD industry (in years) (u0) Firm size: revenue in the ODD industry (not logged, logged) (t) De novo firm dummy ¼ 1 Japanese HQ dummy ¼ 1 Density all firms (t–1) Density at entry all firms (u0) Worldwide industry revenues (t) PC unit shipments (in million units) (t) Number of patents granted in ODD technology (t) Standard ISO9660 period dummy ¼ 1 Firm average product age (t) Industry average product age (t) Firm spread of product ages (t) Industry spread of product ages (t) Firm number of all products (t) Firm # of top 15% products (t) Firm # of bottom 84% products (t) Density of all products (t) Density of top 15% products (t) Density of bottom 84% products (t)
Std. Dev. a
3.58 36.9
3.66 31.3
71.5 2.63 0.145 0.506 43.5 32.1 3,669 49.9 243
160.5 1.82 0.352 0.500 15.0 17.0 2,982 31.8 79.4
0.876 0.804 0.812 1.28 6.84 4.72 2.67 1.60 229.8 131.8 75.3
0.330 1.00 0.292 1.72 2.21 5.59 3.69 2.42 112.8 72.2 31.9
Min. 0 0 0.3 1.20 0 0 3 3 2 8 57 0 0 0 0 0 0 0 0 3 1 2
Max. 15.13 106 1,080 6.98 1 1 63 63 8,764 113 346 1 6.48 1.27 9.38 9.38 44 34 19 377 251 126
N of firms 109; N of exits 66 (7 mergers and acquisitions; 59 disbandings); N of firm-years 635. N of products 1,323; N of product-years 2,999. a Based both on uncensored and right-censored firms.
on the firm tenure pieces show positive age-dependence: the older the firm, the more likely it is to disband. This effect is consistent with the high-velocity nature of the optical disk drive industry. Larger firms have significantly lower failure rates than smaller firms. Firms with larger product portfolios are less likely to disband than those with small portfolios. Organizational failure rates are not significantly affected by firm entry mode, firm length of preentry experience and location of firm headquarters. Environmental controls are also largely in the expected direction. The effect of founding density is positive and significant. Worldwide revenues have positive effect on failure. The significant negative coefficient of PC shipments indicates that increasing demand for the optical disk drives
1 1 Firm tenure in the industry (in years) (t) 2 Firm age at entry into ODD industry (in years) (u0) 3 Ln Firm size: revenue in the ODD industry (t) 4 De novo firm dummy ¼ 1 5 Japanese HQ dummy ¼ 1 6 Density all firms (t–1) 7 Density at entry all firms (u0) 8 Worldwide industry revenues (t) 9 PC unit shipments (in million units) (t) 10 Number of patents granted in ODD technology (t) 11 Standard ISO9660 period dummy ¼ 1 12 Firm average product age (t) 13 Industry average product age (t) 14 Firm spread of product ages (t) 15 Industry spread of product ages (t) 16 Firm number of all products (t) 17 Firm # of top 15% products (t) 18 Firm # of bottom 84% products (t) 19 Density of all products (t) 20 Density of top 15% products (t) 21 Density of bottom 84% products (t)
2
Correlation Matrix for Key Variables.
3
4
5
6
7
8
9
10
11
0.64 0.00 0.08 0.14 0.88 0.27 0.25 0.11 0.90 0.82 0.78
0.11 0.49 0.22 0.79 0.20 0.19 0.10 0.63 0.59 0.63
12
13
0.26 0.46 0.03 0.00 0.05 0.03 0.07 0.08 0.02
0.22 0.22 0.02 0.04 0.09 0.17 0.22 0.21
14
15
16
0.16 0.53 0.49 0.39 0.08 0.07 0.13
0.24 0.23 0.12 0.88 0.82 0.88
0.87 0.66 0.28 0.26 0.20
17
18
19
20
0.22 0.58
0.19
0.18 0.28 0.37 0.48 0.40 0.42
0.43 0.51 0.06 0.27 0.09 0.08
0.27 0.19 0.51 0.02 0.56 0.57
0.42 0.02 0.14 0.02 0.03
0.11 0.39 0.14 0.13
0.56 0.85 0.83
0.50 0.49
0.99
0.41
0.05
0.53
0.05
0.10
0.85
0.50
0.82
0.84
0.29 0.41 0.01 0.62 0.36 0.57 0.54 0.32 0.39 0.36 0.29
0.01 0.01 0.10 0.12 0.05 0.19 0.13 0.24 0.07 0.07 0.06
0.32 0.02 0.10 0.37 0.48 0.65 0.59 0.40 0.55 0.50 0.42
0.05 0.01 0.05 0.13 0.04 0.15 0.09 0.17 0.03 0.02 0.02
0.02 0.03 0.11 0.21 0.10 0.23 0.20 0.22 0.13 0.12 0.10
0.69 0.04 0.08 0.10 0.93 0.26 0.26 0.08 0.96 0.95 0.74
0.38 0.41 0.05 0.49 0.53 0.26 0.26 0.19 0.56 0.54 0.45
0.44 0.11 0.40 0.02 0.72 0.27 0.25 0.07 0.92 0.86 0.61
0.46 0.10 0.35 0.04 0.72 0.28 0.26 0.07 0.90 0.85 0.59
Direct and Indirect Effects of Product Portfolio
Table 1b.
0.26 0.27 0.08 0.28 0.04 0.97 0.15 0.16 0.76 0.60
615
616
Table 2.
OLGA M. KHESSINA
Piecewise Exponential Models: Effects of Product Age on Firm Failure Ratesa.
Firm tenure: 0oup1 Firm tenure: 1oup4 Firm tenure: 4oup8 Firm tenure: u48 Density all firms (t–1) Density at entry all firms (u0) Worldwide industry revenues (t) PC unit shipments (in million units) (t) Number of patents granted in ODD technology (t) Standard ISO9660 period dummy ¼ 1 De novo firm dummy ¼ 1 Firm age at entry into ODD industry (u0) Japanese HQ dummy ¼ 1 Ln Firm size: Revenue in ODD industry (t) Firm number of all products (t) Firm average product age (t) Industry average product age (t) Firm spread of product ages (t) Firm spread of product ages2 (t) Industry spread of product ages (t)
Model (2.1)
Model (2.2)
Model (2.3)
Model (2.4)
Model (2.5)
4.31 (1.34) 3.31 (1.23) 1.92 (1.25) 1.55 (1.34) 0.008 (0.021) 0.078 (0.028) 0.001 (0.000) 0.069 (0.029) 0.007 (0.004)
4.93 (1.41) 4.09 (1.30) 2.85 (1.32) 2.05 (1.36) 0.009 (0.024) 0.115 (0.031) 0.001 (0.000) 0.080 (0.028) 0.006 (0.004)
3.66 (1.05) 2.84 (0.972) 1.49 (1.07) .618 (1.25) 0.039 (0.027) 0.123 (0.032) 0.001 (0.000) 0.079 (0.032) 0.014 (0.005)
4.32 (1.33) 3.25 (1.23) 1.81 (1.27) 1.44 (1.37) 0.002 (0.021) 0.086 (0.029) 0.001 (0.000) 0.071 (0.029) 0.006 (0.004)
4.54 (1.31) 3.59 (1.22) 2.07 (1.30) 1.66 (1.42) 0.071 (0.037) 0.089 (0.030) 0.001 (0.000) 0.108 (0.042) 0.016 (0.005)
1.48 (0.985) 0.145 (0.311) 0.007 (0.005) 0.319 (0.383) 0.447 (0.133) 0.489 (0.099)
2.01 (0.958) 0.119 (0.313) 0.009 (0.005) 0.633 (0.396) 0.550 (0.139) 0.403 (0.101) 0.561 (0.108)
0.318 (1.36) 0.164 (0.313) 0.009 (0.004) 0.726 (0.403) 0.568 (0.136) 0.390 (0.094) 0.620 (0.116) 3.70 (1.54)
1.52 (0.970) 0.175 (0.311) 0.008 (0.005) 0.436 (0.404) 0.467 (0.136) 0.501 (0.108)
0.784 (0.988) 0.201 (0.312) 0.007 (0.005) 0.413 (0.414) 0.394 (0.132) 0.497 (0.105)
0.198 (0.207) 0.075 (0.030)
0.192 (0.206) 0.075 (0.030) 0.671 (0.279)
617
Direct and Indirect Effects of Product Portfolio
Table 2. (Continued )
Number of firm-year spells Number of firms Number of events of firm disbanding (excludes merges and acquisitions events) Wald statistics Log-likelihood (d.f.)
Model (2.1)
Model (2.2)
Model (2.3)
Model (2.4)
Model (2.5)
632 109 59
632 109 59
632 109 59
632 109 59
632 109 59
173.43 91.40 (15)
202.26 83.69 (16)
213.70 81.26 (17)
178.39 90.02 (17)
195.05 87.58 (18)
po0.05. po0.01. po0.001. a
Robust standard errors shown in parentheses.
reduces failure rates of their producers. Technological crowding measured by number of patents in the optical disk drive technology, adoption of ISO 9660 standard, and density of all firms do not significantly affect firm disbanding rates. Model 2.2 builds on the baseline model by adding the variable of a firm’s average product age. The variable has a positive significant effect on firm mortality. It shows that the older the average age of products in a firm’s portfolio, the more likely the firm is to exit the optical disk drive market. The square term of a firm’s average product age was not significant and was omitted from the analyses. This results support Hypothesis 1 only partly indicating that firms’ failure rates increase linearly with increasing average age of products in their portfolios. Model 2.3 tests Hypothesis 2, which concerns the indirect effects of product portfolio average age on firm survival. It builds on Model 2.2 by adding the variable industry average product age. Estimates in Model 2.3 are as predicted. As the average age of products in the industry increases, firm failure rates significantly decrease. This finding supports Hypothesis 2 that the proliferation of old products in the industry weakens competitive pressures and stimulates firm survival. Figure 2 is based on Model 2.3 and provides a graphical depiction of these dynamics. It plots the effects of industry average product age on failure rates of firms, assuming different average age of their product portfolios. Values of firms’ average product age are fixed at the mean value of this variable (0.8) and mean value plus one standard deviation (1.8). Fig. 2 shows that the
618
OLGA M. KHESSINA
Table 3. Piecewise Exponential Models: Effects of Product (Non)Innovativeness on Firm Failure Ratesa.
Firm tenure: 0oup1 Firm tenure: 1oup4 Firm tenure: 4oup8 Firm tenure: u48 Density at entry all firms (u0) Worldwide industry revenues (t) PC unit shipments (in million units) (t) Number of patents granted in ODD technology (t) Standard ISO9660 period dummy ¼ 1 De novo firm dummy ¼ 1 Firm age at entry into ODD industry (u0) Japanese HQ dummy ¼ 1 Ln firm size: Revenue in ODD industry (t) Firm number of all products (t)
Model (3.1)
Model (3.2)
Model (3.3)
4.41
3.93
3.80 (1.29) 2.86 (1.19) 1.53 (1.26) 1.49 (1.40) 0.068 (0.029) 0.002 (0.001) 0.191 (0.078) 0.019 (0.008) 1.79 (1.10) 0.226 (0.324) 0.007 (0.005) 0.342 (0.397) 0.407 (0.120)
(1.44) 3.46 (1.33) 2.33 (1.35) 2.16 (1.41) 0.060 (0.029) 0.000 (0.000) 0.065 (0.041) 0.005 (0.004) 2.58 (1.20) 0.110 (0.315) 0.006 (0.005) 0.277 (0.375) 0.452 (0.131) 0.483 (0.097)
(1.29) 3.04 (1.19) 1.73 (1.25) 1.70 (1.39) 0.067 (0.029) 0.002 (0.001) 0.190 (0.078) 0.019 (0.008) 1.70 (1.09) 0.239 (0.320) 0.007 (0.005) 0.383 (0.401) 0.396 (0.121)
0.540 (0.149)
Firm # of top 15% products (t) Firm # of top 15% products2 (t)
0.358 (0.103)
Firm # of bottom 84% products (t) Firm # of bottom 84% products2 (t) Density of all products (t) Density of top 15% products (t) Density of bottom 84% products (t)
0.651 (0.156) 0.019 (0.004) 0.529 (0.147) 0.032 (0.014)
0.016 (0.006) 0.010 (0.005) 0.020 (0.015)
0.011 (0.005) 0.020 (0.015)
619
Direct and Indirect Effects of Product Portfolio
Table 3. (Continued )
Number of firm-year spells Number of firms Number of events of firm disbanding (excludes merges and acquisitions events) Wald statistics Log-likelihood (d.f.)
Model (3.1)
Model (3.2)
Model (3.3)
635 109 59
635 109 59
635 109 59
180.21 88.80 (15)
174.88 89.36 (17)
191.50 88.50 (19)
po0.05. po0.01. po0.001. a
Robust standard errors shown in parentheses.
2 firm average product age =0.8
Multiplier Rate of Firm Failure
1.8 1.6 1.4
firm average product age =1.8
1.2 1 0.8 0.6 0.4 0.2
1. 2
1. 1
1
0. 9
8 0.
0. 7
6 0.
5 0.
4 0.
3 0.
0. 2
0. 1
0
0
Industry Average Product Age (observed range)
Fig. 2.
Effect of Industry Average Product Age on Firm Mortality.
higher a firm’s average product age, the greater its mortality rate. Specifically, a firm with an average product age of 1.8 years old is more likely to exit the industry at any point of time than a firm with an average product ages of 0.8 years. Additionally, consistent with Hypothesis 2, Fig. 2
620
OLGA M. KHESSINA
demonstrates that as industry average product age increases, the mortality rates of all firms decrease. Models 2.4–2.5 test effects of temporal scope of firm product portfolio on firm failure rates. Model 2.4 shows that the effect is only weakly curvilinear: the linear coefficient is negative but not significant, and quadratic coefficient is positive and significant. Initially, firm mortality rates are not affected by increasing age spread of products in firm portfolios, but when the spread between the youngest and the oldest products in firm portfolio reaches 1 year and 4 months, firm failure rates start to significantly increase.3 In other words, failure rates of firms non-monotonically increase with increasing age spread of products in firm portfolio. Therefore, Hypothesis 3 is only partly supported. As discussed above, a failure to find a significant U-shaped effect can be attributed to the fast-velocity environment of the optical disk drive industry. Model 2.5 confirms Hypothesis 4 proposing that as the age spread of products at the industry level increases, competition weakens and failure rates of firms decrease. Table 3 tests Hypotheses 5 and 6 regarding the effects of product portfolio innovativeness on firm survival. Model 3.1 builds on the baseline Model 2.1 by substituting the variable density of all firms by a variable measuring the density of all products. The density of all products has a predicted significant competitive effect on firm survival. Model 3.2 builds on Model 3.1 such that the firm number of all products variable is broken down into the firm number of top 15% products and the firm number of bottom 84% products, and the density of all products variable is separated out into the variables of the density of top 15% products and the density of bottom 84% products. The results demonstrate that firms with many products, no matter near the top or bottom frontier, have better survival chances than firms with fewer products. However, the presence of the top 15% products in a firm’s portfolio decreases firm hazard rates more than the presence of the bottom 84% products. The indirect effects of portfolio innovativeness are also in the predicted direction. The greater the density of the top 15% products in the industry, the higher the mortality rates of firms. The effect of the density of the bottom 84% products on the hazard is not significant. Thus, Hypothesis 6, which predicts harmful effects of density of innovative products on firm survival, is supported. The results show that product innovativeness affects firm mortality both directly and indirectly. Firms with more innovative product portfolios have higher survival rates. However, innovative products intensify competition in the industry and in this way reduce all firms’ survival chances, including
621
Direct and Indirect Effects of Product Portfolio 70 firm number of top 15% products=3
Multiplier Rate of Firm Failure
60
50
firm number of top 15% products=6
40
30
20
10
24 0
22 0
20 0
18 0
16 0
14 0
12 0
10 0
80
60
40
20
1
0
Industry Density of Top 15% Products (observed range)
Fig. 3.
Effect of Density of Top 15% Products on Firm Mortality.
those with innovative product portfolios. Fig. 3 is based on Model 3.2 and demonstrates these dynamics. It plots the effects of the industry density of top 15% products on the failure rates of firms with a mean number of top 15% products (about three products) and of firms with a mean number of top 15% products plus one standard deviation (about six products). The figure shows that firms with a greater number of top 15% products (i.e., six products) have lower mortality rates than firms with fewer top 15% products (i.e., three products) at any point of industry development. However, the mortality rates of firms with both few and many innovative products significantly increase, as the total number of top 15% products in the industry increases. Model 3.3 tests for curvilinear effects of innovative and non-innovative products on firm failure rates. It shows that while having either innovative or non-innovative products in a portfolio is largely beneficial for firm survival, having too many of these products becomes harmful. Specifically, as the number of top 15% products in a firm’s portfolio increases, its failure rate significantly decreases. However, as the number of innovative products in the firm’s portfolio reaches 17, organizational failure rate starts to increase. As predicted, having bottom 84% products in the portfolio is less
622
OLGA M. KHESSINA
beneficial than having top 15% products. Although non-innovative products can decrease firm-disbanding rate as well, as their numbers approaches 8, organizational failure rate starts to increase.4 These results provide support for Hypothesis 5 predicting a non-monotonic effect of innovative (and non-innovative) products on firm survival. Sensitivity Analyses Several sensitivity analyses were conducted. First, based on the literature suggesting that de alio firms with different pre-entry experience should differ in their survival chances in the industry of new entry (Carroll et al., 1996; Dobrev et al., 2004), I created a set of dummy variables measuring whether a firm’s origin was in one of the following five industrial sectors: computers and computer peripherals, consumer electronics, electric and electronic components, optics, and unrelated industries. Product portfolio models where the variable de novo firm dummy was substituted with this set of de alio origin dummies did not show any significant differences in survival between de alio firms that came from different industries. Market concentration may have a profound effect on firm survival (Carroll, 1985; Carroll, Dobrev, & Swaminathan, 2002). However, the sensitivity analyses did not show either significant or consistent effect of the four-firm concentration ratio on failure rates in the optical disk drive industry. Third, using top 15% and bottom 84% technological performance threshold to define innovative and non-innovative products respectively is somewhat arbitrary. In sensitivity analyses, I tried different breakpoints: top 10% (bottom 89%), top 20% (bottom 79%), top 25% (bottom 74%). The hypotheses were confirmed when these alternative measures were used. The interesting additional result was that measures that defined fewer products as innovative (e.g., top 10% vs. top 15% and top 15% vs. top 20%), showed stronger beneficial effect of product portfolio innovativeness on organizational survival. Finally, few control variables, such as, worldwide industry revenues, PC unit shipments, and the number of patents, are highly correlated with each other. I ran models which excluded two of the three variables and still received consistent results.
DISCUSSION AND CONCLUSION Traditionally a domain of strategy, economics, and marketing literatures (e.g., Burton, 1994; Lancaster, 1990), the phenomenon of product portfolio
Direct and Indirect Effects of Product Portfolio
623
and its role in firm performance and survival have recently attracted attention of organizational ecologists (e.g., Barnett & Freeman, 2001; Dobrev et al., 2001; Dowell & Swaminathan, 2000; Greenstein & Wade, 1998; Sorenson, 2000). Early research established that breadth (e.g., Kerke & Srinivasan, 1990) and newness of product portfolio (e.g., Burton, 1994; Chaney et al., 1991) improve firm performance and survival. Organizational ecology offered theoretical insights that allowed researchers to advance these initial findings and establish, for example, that while having new products in the portfolio improves firm survival, adding too many new products at the same time is detrimental, because new product development disrupts organizational routines (Barnett & Freeman, 2001). While firms with broader portfolios benefit from risk spreading and economies of scale, firms with very broad portfolios suffer from exposure to more intense competition, because their resource niches overlap with those of competitors (Dobrev et al., 2001). Organizational ecology has more conceptual and empirical tools to offer to advance research on product portfolios. One of the most valuable insights of this theoretical tradition is that a firm’s individual actions may have ecological ramifications that are felt not only by competitors, but also by the very firm that initiated them. More specifically, by influencing firm outcomes directly, e.g., making a firm a stronger or weaker competitor, the firm’s actions may also have an impact the environment, e.g., contribute to change in population vital rates, and in this way affect this firm indirectly through changing the intensity of competition in the industry. Strategy research on product portfolios has been focused on their direct effects on firm performance and survival, and largely ignored that product portfolios may influence their firms indirectly by changing ecological dynamics of the industry. In this paper, I argued and demonstrated that in order to evaluate the net effect of product portfolio on an organization, it is necessary to take into account not only direct immediate impact observed at the firm level, but also indirect ecological influences generated at the industry level. I focused on two understudied characteristics of product portfolios: portfolio aging and portfolio (non)innovativeness and predicted that direct and indirect effects of these two portfolio features may work in opposite directions. Based on the empirical analyses of product portfolios of all firms ever participated in the worldwide optical disk drive industry from its beginning in 1983 through 1999, I found that while carrying aging products is harmful for a firm, this negative effect on firm survival is significantly reduced when the industry is filled with many aging products. By contrast, although firms with many innovative products in their portfolios enjoy
624
OLGA M. KHESSINA
greater survival advantages, these benefits significantly diminish when other firms in the industry offer innovative products as well. These findings imply that whether a product portfolio brings a firm the most or least advantages will depend on the degree of its similarity to product portfolios of other firms competing in the industry. Obviously, different industries are populated by firms with different degree of similarities and differences between their product portfolios. One limitation of this study is that it is based on a single industry. However, knowing about the effects of product portfolios on firm survival in the optical disk drive industry, which is high velocity and technology intensive, it is possible to speculate how product portfolios will affect firms in slow-paced and other types of contexts. The importance of similarities and differences in product portfolios of competitors for firm survival makes it interesting to compare this paper to the study by McGahan (2006), also published in this volume. McGahan seeks to establish whether diversity among competitors improves average profitability of an industry. Her study covers great variety of industries, but it is cross sectional and based on the data on public firms only. The sizebiased sample makes it somewhat difficult to interpret results of the study. Nevertheless, there is one strong finding: industries significantly differ in the degree of diversity of firms that inhabit them. This suggests that different industries are populated by firms with a greater or lesser degree of similarity in their product portfolios. When is similarity in competitors’ product portfolio beneficial and when is it harmful? The answer obviously depends on what characteristics of product portfolios are used for comparison. This paper suggests that firms with older portfolios fare better when their competitors have similarly aged portfolios, because increasing number of old products in the industry diminishes negative effects of a focal firm’s portfolio aging on survival. Thus, the similarity in this characteristic of product portfolio is beneficial. Since many aging products are usually present in slow-paced, technology non-intensive industries (Tushman & Anderson, 1986), it is reasonable to expect that firms with older portfolios fare better in these environments than in high-velocity, technology-intensive ones. By the same logic, mature markets provide a more favorable environment for firms with aging portfolios than emergent and young industries. It also seems that firms with aging portfolios will fare better in a given industry in underdeveloped economies, because they are more slow-paced than the developed ones. If this speculation is true, then multinational corporations while prompted by competition to renew their product portfolios
Direct and Indirect Effects of Product Portfolio
625
in business units located in developed countries may slow down this process in the divisions located in underdeveloped economies. Another implication from this study is that firms benefit from having innovative products in their portfolios the most when they are among very few in the industry that offer such products. Therefore, similarity in this product portfolio characteristic among competitors is harmful. Industries with very few innovative firms are those among slow-paced and technology non-intensive. Although in high-velocity environments firms have to be innovative in order to survive, somewhat paradoxically, the greatest relative benefit from innovative portfolios is likely to be achieved in slow-paced industries where innovation is rarer and advantages of a firm’s innovative portfolio are not significantly diminished by very few innovative products of other firms in the industry. The paper reveals interesting findings with respect to direct effects of innovative and non-innovative products on firm survival. Differently from the prevailing view in the management of innovation and strategy literatures, the study provides empirical evidence that having a great number of innovative products in a portfolio is not always beneficial for a firm, whereas having non-innovative products is not necessarily harmful. A firm may benefit from offering both innovative and non-innovative products. Certain balance is warranted. Too many innovative products may hurt firm survival chances, because their introduction causes disruptions in organizational operation. Too many non-innovative products may render a firm obsolete. The study shows that firms should have greater number of innovative products than non-innovative to attain better survival benefits. The optimal balance between innovative and non-innovative products in a firm’s portfolio is likely to be context-specific. In high-velocity environments, like the optical disk drive industry, firms should avoid having too many non-innovative products. In slow-paced environments, number of such products could be higher without harm to a firm. It also seems that firms can do better with noninnovative portfolios in underdeveloped economies than in developed ones. Thus, important differences in the power of product portfolios’ effects on firm survival can be discovered by studying industries that are different from the optical disk drives in terms of technological change, growth, and firm composition. Future research could help to establish what industry characteristics make certain features of product portfolio have a stronger or weaker effect on firm survival. To conclude, product portfolio may affect firms in complex and unexpected ways. Organizational ecology possesses tools that help to uncover this complexity. Specifically, the study generated two key conclusions. First,
626
OLGA M. KHESSINA
when establishing influence of product portfolios on organizational performance and survival, it is necessary to examine both direct effects that are manifested at the firm level and indirect ecological effects that are generated at the industry level. Second, it is important to remember that the same product portfolio characteristic may affect firm survival in both beneficial and harmful ways depending on the degree of presence of this characteristic in the portfolio. Future research in the area of product portfolio could further benefit from drawing on theoretical and empirical heritage of organizational ecology tradition.
NOTES 1. Disk/Trend reports three instances when a firm was acquired (or merged) and its products appeared under the name of the acquiring (or merged) organization preceding the date of the event. For these cases, I defined an acquiring/merging firm’s entry as occurring at the time of acquisition (or merger). 2. Since the meaning of mergers and acquisitions is ambiguous – possibly resulting from either firm success or failure – exits by either merger or acquisition are excluded from constructing the dependent variable and treated as non-informatively censored on the right. Findings are not affected by whether merges and acquisitions are treated as exit events or as censored observations. 3. The non-linear effect of firm spread of product ages on firm survival is within the observed range of the data. 4. The non-linear effects of the number of top 15% and bottom 84% products in firm portfolio on firm survival are within the observed range of the data.
ACKNOWLEDGMENTS This study is an extension from my dissertation. I would like to thank my dissertation chair Glenn Carroll for his guidance on this research. I am also indebted to my other dissertation committee members: John Freeman, Bronwyn Hall, David Mowery, and Trond Petersen. I appreciate comments from participants of Georgetown University – University of Maryland Brown Bag Strategy Seminar. Detailed comments and suggestions from Joel Baum, Stanislav Dobrev, and Arjen van Witteloostuijn have helped to significantly improve the paper. This research was supported by Jr. Faculty Research Fellowship provided by McDonough School of Business at Georgetown University. The study is a part of the research project directed by Glenn Carroll and David McKendrick, under the auspices of the Information Storage Industry Center, U.C. San Diego funded by the Alfred P. Sloan Foundation.
Direct and Indirect Effects of Product Portfolio
627
REFERENCES Barnett, W. P. (1993). Strategic deterrence among multipoint competitors. Industrial and Corporate Change, 2, 249–278. Barnett, W. P. (1997). The dynamics of competitive intensity. Administrative Science Quarterly, 42, 128–160. Barnett, W. P., & Carroll, G. R. (1995). Modeling internal organizational change. Annual Review of Sociology, 21, 217–236. Barnett, W. P., & Freeman, J. (2001). Too much of a good thing? Product proliferation and organizational failure. Organization Science, 5, 539–558. Barnett, W. P., Greve, H. R., & Park, D. Y. (1994). An evolutionary model of organizational performance. Strategic Management Journal, 15(Winter Special Issue), 11–28. Barnett, W. P., & Hansen, M. T. (1996). The red queen in organizational evolution. Strategic Management Journal, 17(Summer Special Issue), 139–157. Barnett, W. P., & McKendrick, D. G. (2004). Why are some organizations are more competitive than others? Evidence from a changing global market. Administrative Science Quarterly, 49, 535–571. Barnett, W. P., & Pontikes, E. G. (2005). The red queen: History-dependent competition among organizations. Research in Organizational Behavior, 26, 351–371. Baum, J. A. C., Korn, H. J., & Kotha, S. (1995). Dominant designs and population dynamics in telecommunications services: Founding and failure of Facsimile Transmission Service Organizations, 1965–1992. Social Science Research, 24, 97–135. Baum, J. A. C., & Singh, J. V. (1994). Organizational niches and the dynamics of organizational mortality. American Journal of Sociology, 100, 346–380. Bayus, B. L., Erikson, G., & Jacobson, R. (2003). The financial rewards of new product introductions in the personal computer industry. Management Science, 49, 197–210. Bernheim, B. D., & Whinston, M. D. (1990). Multi-market contact and collusive behavior. RAND Journal of Economics, 21, 1–26. Burton, P. S. (1994). Product portfolio and the introduction of new products: An example from the insecticide industry. The RAND Journal of Economics, 25, 128–140. Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in populations of organizations. American Journal of Sociology, 90, 1262–1283. Carroll, G. R., Bigelow, L. S., Seidel, M-D. L., & Tsai, L. B. (1996). The fates of de novo and de alio producers in the American automobile industry, 1885–1981. Strategic Management Journal, 17, 117–137. Carroll, G. R., Dobrev, S., & Swaminathan, A. (2002). Organizational processes of resource partitioning. Research in Organizational Behavior, 24, 1–40. Carroll, G. R., & Hannan, M. T. (1989). Density delay in the evolution of organizational populations: A model and five empirical tests. Administrative Science Quarterly, 34, 411–430. Carroll, G. R., & Hannan, M. T. (2000). The demography of corporations and industries. Princeton, NJ: Princeton Press. Carroll, G. R., & Teo, A. C. (1996). Creative self-destruction among organizations: An empirical study of technical innovation and organizational failure in the American automobile industry, 1885–1981. Industrial and Corporate Change, 5, 619–644. Chaney, P. K., Devinney, T. M., & Winer, R. S. (1991). The impact of new product introductions on the market value of firms. Journal of Business, 4, 573–610.
628
OLGA M. KHESSINA
Christensen, C. M., & Bower, J. L. (1996). Customer power, strategic investment, and the failure of leading firms. Strategic Management Journal, 17, 197–218. Christensen, C. M., & Rosenbloom, R. S. (1995). Explaining the attacker’s advantage: Technological paradigms, organizational dynamics, and the value networks. Research Policy, 24, 233–257. Clark, K. B., & Wheelwright, S. C. (1993). Managing new product and process development: Text and cases. New York: Free Press. Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35, 128–152. de Figueiredo, J. M., & Kyle, M. K. (2006). Surviving the gales of creative destruction: The determinates of product turnover. Strategic Management Journal, 27, 241–264. Disk/Trend. (1985–1999). Disk/trend report: Optical disk drives. Mountain View, CA: Disk/ Trend, Inc. Dobrev, S. D., Kim, T., & Carroll, G. R. (2002). The evolution of organizational niches: U.S. automobile manufacturers, 1885–1981. Administrative Science Quarterly, 47, 233–264. Dobrev, S. D., Kim, T., & Hannan, M. T. (2001). Dynamics of niche width and resource partitioning. American Journal of Sociology, 106, 1299–1337. Dobrev, S. D., Kim, T., & Solari, L. (2004). The two sides of the coin: Core competence as a capability and obsolescence. In: J. A. C. Baum, & A. M. McGahan (Eds), Business strategy over the industry life cycle: Advances in strategic management, (Vol. 21, pp. 255– 285). Oxford, UK: JAI. Dowell, G., & Swaminathan, A. (2000). Racing and back-pedalling into the future: New product introduction and organizational mortality in the US bicycle industry, 1880–1918. Organization Studies, 21, 405–431. Freeman, J., Carroll, G. R., & Hannan, M. T. (1983). The liability of newness: Age dependence in organizational death rates. American Sociological Review, 48, 692–710. Freeman, J., & Hannan, M. T. (1983). Niche width and the dynamics of organizational populations. American Journal of Sociology, 88, 1116–1145. Gerlach, M. L., & Lincoln, J. R. (2000). Economic organization and innovation in Japan: Networks, spin-offs and the creation of enterprise. In: G. von Krogh, I. Nonaka, & T. Nishiguchi (Eds), Knowledge creation: A new source of value (pp. 151–196). London: MacMillan Press. Gilbert, R. J., & Matutes, C. (1993). Product life rivalry with brand differentiation. Journal of Industrial Economics, 41, 223–240. Greenstein, S. M., & Wade, J. B. (1998). The product life cycle in the commercial mainframe computer market, 1968–1982. RAND Journal of Economics, 29, 772–789. Hannan, M. T. (1998). Rethinking age dependence in organizational mortality: Logical formalizations. American Journal of Sociology, 104, 85–123. Hannan, M. T., & Carroll, G. R. (1992). Dynamics of organizational populations: Density, legitimation, and competition. New York: Oxford University Press. Hannan, M. T., & Freeman, J. (1984). Structural inertia and organizational change. American Sociological Review, 49, 149–164. Hannan, M. T., Po´los, L., & Carroll, G. R. (2003). Cascading organizational change. Organization Science, 14, 463–482. Haveman, H. A., & Nonnemaker, L. (2000). Competition in multiple geographic markets: The impact on growth and market entry. Administrative Science Quarterly, 45, 232–267.
Direct and Indirect Effects of Product Portfolio
629
Henderson, R. (1993). Underinvestment and incompetence as responses to radical innovation: Evidence from photolithographic alignment equipment industry. RAND Journal of Economics, 24, 248–270. Henderson, R. M., & Clark, K. B. (1990). Architectural innovation: The reconfiguration of existing product technologies and the failure of established firms. Administrative Science Quarterly, 35, 9–30. Kerke, S., & Srinivasan, K. (1990). Broader product line: A necessity to achieve success? Management Science, 36, 1216–1231. Khessina, O. M. (2003). Entry mode, technological innovation and firm survival in the worldwide optical disk drive industry, 1983–1999. Unpublished doctoral dissertation. University of California, Berkeley, CA. Klepper, S., & Simons, K. L. (2000). Dominance by birthright: Entry of prior radio producers and competitive ramifications in the U.S. television receiver industry. Strategic Management Journal, 21, 997–1016. Lancaster, K. (1990). The economics of product variety: A survey. Marketing Science, 9, 189–206. Lubben, R. (1988). Just-in-time manufacturing. New York: McGraw-Hill. McGahan, A. M. (2006). Industry performance and changes in competitor characteristics: Evidence on isolationism versus mutual forbearance. In: J. A. C. Baum, S. D. Dobrev, & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 631–666). Oxford, UK: JAI. McPherson, M. (1983). An ecology of affiliation. American Sociological Review, 48, 519–532. Merrill Lynch & Co., & McKinsey and Company. (2001). The storage report – customer perspectives and industry evolution. Special report of a joint industry study. Meyer, M. H., & Roberts, E. B. (1986). New product strategy in small technology firms: A pilot study. Management Science, 7, 806–821. Mitchell, W. (1994). The dynamics of evolving markets: The effects of business sales and age on dissolutions and divestitures. Administrative Science Quarterly, 39, 575–602. Perloff, J. M., & Salop, S. C. (1985). Equilibrium with product differentiation. Review of Economic Studies, 52, 107–120. Petersen, T. (1991). Time-aggregation bias in continuous-time hazard-rate models. In: P. Masden (Ed.), Sociological Methodology (Vol. 21, pp. 263–290). Oxford: Basil Blackwell. Podolny, J., & Stuart, T. (1995). A role-based ecology of technological change. American Journal of Sociology, 100, 1224–1260. Podolny, J., Stuart, T., & Hannan, M. T. (1996). Networks, knowledge, and niches: Competition in the worldwide semiconductor industry, 1984–1991. American Journal of Sociology, 102, 659–689. Purcell, L. (2000). CD-R/DVD: Disc recording demystified. New York, NY: McGraw-Hill. Reinganum, J. F. (1983). Uncertain innovation and the persistence of monopoly. American Economic Review, 73, 141–148. Roberts, P. W. (1999). Product innovation, product-market competition and persistent profitability in the U.S. pharmaceutical industry. Strategic Management Journal, 20, 655–670. Rosenkopf, L., & Nerkar, A. (2001). Beyond local search: Boundary-spanning, exploration, and impact in the optical disk industry. Strategic Management Journal, 22, 287–306. Salop, S. C. (1979). Monopolistic competition with outside goods. Bell Journal of Economics, 10, 141–156.
630
OLGA M. KHESSINA
Schmalensee, R. (1978). Entry deterrence in the ready-to-eat breakfast cereal industry. Bell Journal of Economics, 9, 305–327. Sørensen, J. B. (1999). Stpiece: A program for the estimation of piecewise-constant rate hazard models in STATA 6.0. Unpublished ado-file. Graduate School of Business, University of Chicago. Sorenson, O. (2000). Letting the market work for you: An evolutionary perspective on product strategy. Strategic Management Journal, 21, 577–592. Stavins, J. (1995). Model entry and exit in a differentiated-product industry: The personal computer market. Review of Economics and Statistics, 77, 571–584. Stern, I., & Henderson, A. D. (2004). Within-business diversification in technology-intensive industries. Strategic Management Journal, 25, 487–505. Suarez, F. F., & Utterback, J. M. (1995). Dominant designs and survival of firms. Strategic Management Journal, 16, 415–430. Swaminathan, A., & Delacroix, J. (1991). Differentiation within an organizational population: Additional evidence from the wine industry. Academy of Management Journal, 34, 679–692. Tripsas, M. (1997). Unraveling the process of creative destruction: Complementary assets and incumbents survival in the typesetter industry. Strategic Management Journal, 18(Summer Special Issue), 119–142. Tushman, M. L., & Anderson, P. C. (1986). Technological discontinuities and organizational environments. Administrative Science Quarterly, 31, 439–465. West, J. (2002). Limits to globalization: Organizational homogeneity and diversity in the semiconductor industry. Industrial and Corporate Change, 11, 159–188.
INDUSTRY PERFORMANCE AND CHANGES IN COMPETITOR CHARACTERISTICS: EVIDENCE ON ISOLATIONISM VERSUS MUTUAL FORBEARANCE Anita M. McGahan ABSTRACT This paper describes how firm characteristics evolve in different industries. In particular, it reports on relationships between industry performance and competitor diversity in the American economy from 1981 to 1997. Industry performance is measured using a prospective measure of performance (Tobin’s q) and a measure of performance that reflects historical competence (accounting profitability). Competitor diversity is characterized by differences in size, operating margin, asset composition, and asset utilization. The results indicate significant diversity among competitors in both high- and low-performance industries. The study suggests that low industry performance may be associated with processes of transition in competitor characteristics.
Ecology and Strategy Advances in Strategic Management, Volume 23, 631–666 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-3322/doi:10.1016/S0742-3322(06)23019-5
631
632
ANITA M. MCGAHAN
This paper reports on a panel study of a broad cross section of firms in the American economy from 1981 to 1997 to examine whether diversity in firm characteristics is tied to higher industry performance. Industry performance is measured by average Tobin’s q and accounting profitability.1 Diversity among competitors within industries is characterized by size, operating margin, asset composition, asset utilization, and focus. A comparison of industry performance with diversity in competitor characteristics yields information about the co-evolution of industry structures and competitive advantages. The perspective and methodology of this paper is rooted in industrial economics. Several theoretical traditions in economics carry implications for the relationships between industry performance and variation in incumbent characteristics. One tradition is based on the idea that competitors perform better over the long run when they are unique and isolated (Bain, 1956; Mueller, 1986). This industrial-organization tradition goes further to stipulate that direct competitors tend to sustain advantaged positions for longer periods when their structural differences are slow to decay. A closely related tradition explains low-performance industries as ‘‘perfectly competitive,’’ where competitors are similar in their characteristics and structural barriers are not available to protect against rivalrous pricing. According to this view, imitation among firms with similar characteristics tends to lower industry performance. Thus, the industrial-organization view associates diversity in firm characteristics with higher industry performance. A second tradition – also in economics – associates industry performance with ‘‘mutual forbearance,’’ or cooperation among rivals (Axelrod, 1984; Abreu, 1986). According to the theory in this tradition, tacitly cooperative arrangements tend to arise when firms are similar (called ‘‘weakly symmetric’’) in their characteristics. Similarity in firm characteristics is important because it gives each competitor the ability to threaten the others with retaliation if one competitor defected from cooperating. Thus, the isolationist and ‘‘mutual forbearance’’ views predict different empirical relationships between industry performance and diversity in the characteristics of competitors. While the perspective of this paper reflects industrial economics, the issues studied here have also been addressed in other fields – and particularly in the field of organizational ecology. An earlier volume of Advances in Strategic Management (Volume 18 in 2001) dealt extensively with the relationships between firm performance (a prevailing interest in industrial economics) and firm survival (a prevailing interest in organizational ecology). Firm survival is both an element of ultimate performance and integral to diversity, and yet it is not part of this analysis in this paper because the broad dataset used here does not contain reliable information on firm entry and exit. The analysis
Industry Performance and Changes in Competitor Characteristics
633
concludes with suggestions for further research that complement the findings from this research with the perspective advanced in organizational ecology.
1. ANTECEDENTS The empirical antecedents from economics on the co-evolution of industry performance and competitor performance fall into several related categories. First, studies document that industry and competitor performance tends to persist over time. Mueller (1977, 1986) showed persistence in the ranking and profitability of very high and very low performers among Federal Trade Commission (FTC) manufacturers from 1971 to 1974. Cubbin and Geroski (1987) documented persistence in firm-specific effects and relationships between industry structure and persistence among 217 British manufacturers. Waring (1996) partly explained persistent differences in competitor accounting profitability using 2-digit SIC industry characteristics. McGahan and Porter (1997, 2002) showed accounting profitability that industry effects persist longer than corporate-parent and business-specific effects on the accounting profit of businesses covered in the Compustat Business-Segment Reports from 1982 to 1994. None of the authors in this line has examined the relationship between industry performance and erosion in differences between competitor characteristics, however. A second stream of empirical research has demonstrated that mutual forbearance sometimes arises among competitors with similar characteristics. Bresnahan (1987), Baker (1989), and McGahan (1995) find evidence of cooperation among automobile dealers during the 1950s, steel companies during the 1930s, and brewers during the 1930s, respectively. None of these authors evaluated the relationships between industry performance and changes in characteristics, however. A third set of studies establishes that both industry and idiosyncratic components of profitability are important, although the research in this line does not examine whether diversity in competitor characteristics is tied to industry effects. Schmalensee (1985) established that industry explains about 20% of variance in the business-unit accounting profits among U.S. manufacturers. Rumelt (1991) and McGahan and Porter (1997) showed that competitor-specific effects were more important than either industry or corporate-parent effects to variance in business-specific accounting profits among all U.S. companies. McGahan (1999b) showed that industry membership and competitive position accounted for 27.0% and 36.1%, respectively, of variance in Tobin’s q among corporations covered in Compustat
634
ANITA M. MCGAHAN
from 1982 to 1994. None of these authors examined relationships between changes in industry performance and competitor characteristics, however. The study differs from the previous empirical research in several ways. First, the dataset and measures employed differ. Industry performance is measured using both Tobin’s q and accounting profitability, which are prospective and historical measures, respectively. The data cover a broader variety of sectors and a longer period than many of the previous studies that focused on manufacturing. After screening for financial firms and missing observations, the data cover about two-thirds of non-financial corporate sales and 45% of nonfinancial corporate assets reported to the Internal Revenue Service from 1985 to 1992. Thus, the study deals with a significant portion of economic activity in the U.S. during this period. A disadvantage of the dataset is that it includes information on only publicly traded firms, and, as a result, the industry performance does not represent the constituent performance of all competitors. Second, the analysis involves assessing industry performance at the 4-digit SIC (small industries corporation) level. Although the SIC system defines 4-digit industries in some sectors more narrowly than in others, the definitions are more precise than the 2-digit categories used by some previous authors (e.g., Waring, 1996). Data by industry are obtained from the Compustat Business-Segment Reports, which contain information reported to the SEC by 4-digit category.2 Results are replicated by economic sector to examine their robustness. Third, corporations are included in the analysis even if Compustat does not contain information on them for the entire period from 1981 to 1997. After screening the dataset (as described in a subsequent section), over 80% of the corporations are represented by less than a full series. Several previous studies (i.e., Mueller, 1977, 1986; Waring, 1996) excluded corporations with less than a complete series. This exclusion may have distorted estimates of industry and firm-specific effects on performance because corporations with longer series tend to perform better than their industry average. Finally, differences in competitor characteristics are measured in a variety of ways: size, operating margin, asset composition, asset utilization, and focus. Diversity across competitors in the group of characteristics provides a broad indication of differences in strategy.
2. METHODS The analysis involves several steps. The first step is to determine industry performance. The second step is the identification of competitor characteristics.
Industry Performance and Changes in Competitor Characteristics
635
The third step is to assess diversity in competitor characteristics and to link diversity to industry performance. The first step – determining industry performance – involves calculating a value of Tobin’s q that represents performance at the industry level. If all firms were associated with just one industry, then these calculations would be straightforward as they would only involve calculating a weightedaverage q for each industry from the q’s of the firms within the industries. The weights would reflect the sizes of each of the firms in assets. However, a challenge arises because the firms in the dataset often participate in more than one industry. Because q is a firm-level variable, the calculation of the industry averages must account for this. The approach used here assumes that the contribution of each of a firm’s activities by industry to its overall value of q occurs in proportion to the amount of its assets in each of its industries. Thus, the q of a firm that is entirely devoted to one industry is weighted more in the calculation for the industry than the q of a competitor that is spread over many industries. The second step – identifying competitor characteristics – involves specifying a series of characteristics of firms that reflect firm strategy and that are measurable using accounting statistics. Because these characteristics fluctuate between years, a model is developed to isolate the elements of the characteristics that are fixed over time and also that are consistently building or declining over time. The theory is that these persistent elements of the characteristics reflect the firm’s strategy. An important caveat is warranted here. Ideally, much more detailed information would be available about the strategic intentions of the firms in the analysis, and yet information about intent is unlikely to be verifiable and comparable across firms. Even without information on intentions, better measures of characteristics such as the product-portfolio indicator in Khessina (2006) would facilitate robust inferences about differences in competitor strategies. Yet even this information is not available across a range of industries. As a result, the analysis relies on accounting measures as the best available information for crosssectional analysis. The third step – assessing diversity in competitor characteristics and linking diversity to firm performance – leads to the results. Diversity in competitor characteristics is assessed by identifying the degree of difference between competitors in their characteristics. The model for this identification evaluates the standard deviation of the differences after weighting each characteristic by the inverse of the variance of the estimated characteristic (i.e., more accurate characteristic assessments get more weight). Once diversity among competitor characteristics is established, the final step is to
636
ANITA M. MCGAHAN
associate the diversity with the measure of industry-average performance obtained earlier.
2.1. Assessing Industry Performance The analysis relies on two measures of industry performance. The first is developed from the values of Tobin’s q among incumbents using the following model:3 X qkt ¼ m þ gt þ ai ai;k;t þ k;t (1a) i
In Eq. (1a), qk,t represents Tobin’s q of corporation k at time t. The term m is the mean value of Tobin’s q among corporations in the dataset; gt is the common increment to Tobin’s q among corporations at time t; ai is the effect of membership in industry i; ai,k,t is the proportion of corporation k’s assets at time t in industry i; and ek,t, the residual, is the firm-specific component of the corporation’s value of q at time t. The ai,k,t serve to distribute the value of Tobin’s q for corporation k at time t across the industries in which the corporation participates. This distribution is necessary because a single value of Tobin’s q is available for each corporation at time t even when the firm participates in more than one industry. The effect of the distribution is to calibrate each competitor’s contribution to the industry average by a measure of the importance of the business to the corporate parent. The second measure of industry performance is developed from the accounting profit for incumbents in the following model:4 ri;k;t ¼ m þ gt þ ai þ i;k;t
(1b)
In Eq. (1b), ri,k,t represents accounting profit as the ratio of operating income to identifiable assets for corporation k in industry i at time t. The term m is the mean value of accounting profit among the business segments in the dataset; gt is the common increment to accounting profit among business segments at time t; ai is the effect of membership in industry i; and ei,kt, the residual, is the idiosyncratic component of accounting profit for the business segment of corporation k in industry i at time t. Eq. (1b) differs from Eq. (1a) because the Compustat Business-Segment Reports contain information on accounting profit by 4-digit SIC category. Thus, there is a trade-off in the two measures of performance: Accounting profitability represents differences across the businesses of a corporation more accurately, but Tobin’s q is less vulnerable to reporting anomalies.
637
Industry Performance and Changes in Competitor Characteristics
The purpose of Eqs. (1a) and (1b) is to obtain estimates of ai for each of the 639 industries covered in the screened Compustat data. The term ai is not subscripted by time; each of the industry effects reflects performance over the entire period under study. Eqs. (1a) and (1b) are estimated through ordinary-least-squares assumptions on the unbalanced panel.5 2.2. Measuring Competitors’ Strategies The next step is to measure diversity in competitor characteristics. For each competitor, a number of characteristics are evoked to reflect the inherent complexity in strategy. Table 1 summarizes the hypotheses about high- and low-performance industries. These hypotheses are based on the reducedform empirical implications of the isolationist and mutual-forbearance traditions and are not structural in nature. The table shows characteristics in the left column, with associated hypotheses and measures in the middle and right columns. The first hypothesized relationship covers diversity in competitor size. For each rival, size is measured by sales and by assets. The Compustat BusinessSegment Reports are used to obtain sales and assets by 4-digit SIC category. Under the isolationist view, diversity in size may occur in either highor low-performance industries. Under the mutual-forbearance view, however, high-performance industries require weak symmetry in size. Thus, the Table 1.
Hypotheses about Diversity in Competitors Characteristics. Hypothesesa
Feature
Measures
Isolationism
Mutual Forbearance
None +
+ –
(iii) Asset composition
+
–
(iv) Asset utilization
+
–
(v) Corporate focus
+
–
(i) Size (ii) Operating margin
a
Sales assets Ratio of operating income to sales Percent of assets in property, plant & equipment (PP&E) Asset turnover (ratio of sales to assets) Index taking on higher values for unrelated diversification
Hypotheses about the relationship between industry performance and competitor diversity in each of the measures as implied by the isolationist and mutual-forbearance views.
638
ANITA M. MCGAHAN
mutual-forbearance view associates diversity in size with low-performing competitive industries, while the isolationist literature stipulates no association between diversity in size and industry performance. The second hypothesized relationship associates diversity in operating margin with industry performance. Operating margins are the ratio of operating profit to sales. Differences in operating margin may arise because some rivals command a higher price for products of higher quality or better features without incurring proportionately higher costs. Differences also may arise because some competitors make products more efficiently than others. Under the isolationist view, diversity in operating margin reflects differences in strategy that do not arise in competitive industries. Thus, the isolationist view positively associates industry performance and diversity in operating margin. Under the mutual-forbearance view, diversity in operating margin indicates an important asymmetry between competitors that may induce defection from a tacit agreement. Thus, the mutual-forbearance view negatively associates industry performance and diversity in operating margin. The third hypothesized relationship involves the asset composition of competitors. Asset composition is measured as the percent of booked assets held as property, plant and equipment. Competitors with a low proportion of assets in property, plant and equipment tend to carry large cash and inventory. Diversity in asset composition therefore reflects differences in fixed and variable cost structures. Competitors with a high portion of assets in property, plant and equipment likely have higher fixed costs than competitors with a high portion of assets in working capital. By the isolationist view, diversity in asset composition should be positively associated with industry performance. By the mutual-forbearance view, competitors in highperformance industries should have some similarities in cost structures and asset composition. Thus, the mutual-forbearance view negatively associates industry performance with diversity in asset composition. The fourth hypothesized relationship covers asset utilization, which is related but somewhat different than asset composition. Asset utilization is measured by asset turnover (which is the ratio of sales to assets). A competitor’s asset utilization rate reflects the efficiency with which the competitor generates a specific level of revenue from all its assets. A competitor that generates a large volume of revenues from an asset stock may be pursuing a low-cost strategy for manufacturing and marketing a mass-produced good. In contrast, a competitor that generates a small volume of revenues from an asset stock may be expending capital to manufacture lower volumes of a specialized product that generates a higher gross margin per dollar of revenue. Theories of perfect competition in low-performance industries
Industry Performance and Changes in Competitor Characteristics
639
indicate little latitude for variation in efficiency among rivals. In high-performance industries, diversity in asset utilization may occur when competitors are differentiated. Thus, diversity in asset utilization should arise in high-performance industries according to the isolationist view. Under the mutual-forbearance view, diversity in investment profiles is associated with asymmetry in the efficiency of competitors. Efficiency differences create an incentive for the most efficient competitor to defect from cooperative regimes. Thus, diversity in asset utilization should tend to arise in lowerperformance industries according to the mutual-forbearance view. The fifth hypothesized relationship covers relatedness in the diversification of industry members. Focus in corporate diversification is assessed using an index that takes on larger values if the firm is diversified into unrelated industries (see Wernerfelt & Montgomery, 1988; Caves, Porter, & Spence, 1980; McGahan, 1999b). Specifically, the index is given by Dk;t ¼ P P ai;k;t aj;k;t d i;j where: i
j 8 9 0 if industries i and j have the same 4-digit SIC code > > > > > > > > > if industries i and j have different 4-digit SIC codes but the same-digit SIC codes > > > <1 = d i;j ¼ 2 if industries i and j have different 3-digit SIC codes but the same 2-digit SIC codes > > > > > 3 if industries i and j have different 2-digit SIC codes but the same 1-digit SIC codes > > > > > > : 4 if industries i and j have different 1-digit SIC codes ði:e:; belong to different sectorsÞ > ;
According to the mutual-forbearance literature, similarity in the degree of diversification across competitors facilitates cooperation. If firms are similarly focused, then each is committed to retaliation should one firm defect from cooperation in a high-performance industry. If firms are similarly unfocused in their diversification, then they may have multi-market contact that enhances the incentive to cooperate. Thus, differences in focus are negatively associated with industry performance under the mutualforbearance view. Under the isolationist view, differences in focus reflect diversity in strategy and thus are positively associated with industry performance. The analysis also includes evidence on the relationships between industry concentration and industry performance (no hypothesis appears in the table because concentration is not a firm characteristic). Concentration is measured by both the percent of revenue held by the largest four competitors and by the total number of competitors in the industry. Under the isolationist view, diversity in the strategies of leading competitors may lead to a variety of concentration profiles; thus, the traditional view carries no implications for the relationship between concentration and industry performance. Under the mutual-forbearance view, however, concentration
640
ANITA M. MCGAHAN
is associated with greater ability to observe and enforce cooperative agreements. Thus, concentration is positively associated with industry performance under the mutual-forbearance view. 2.3. Assessing Diversity in Competitor Characteristics The final step in the analysis is to relate industry performance to diversity across competitors in the characteristics listed in the prior section. For each of the characteristic, this requires an assessment of (i) the average across competitors of the stable, long-term difference from average and (ii) the rate at which the competitor converges toward the industry average during the period under study. These assessments are then aggregated across competitors into measures of diversity on each characteristic among competitors, as well as into measures of the rate of convergence in the characteristics across competitors. Each measure of diversity is then regressed on the estimated coefficients of industry performance obtained from the earlier analysis. 2.3.1. A Model of Diversity in Competitor Characteristics This section develops a model for assessing diversity in each of the characteristics in Table 1. Eq. (2) indicates that each characteristic at time t in industry i consists of a fixed and incremental component: (2)
fi;k;t ¼ f i;k þ si;k;t
This model, originally due to Mueller (1986), contains a number of important features. The fixed component, fi,k, is the average size of corporation k’s effect in industry i over the entire period. The incremental component, si,k,t, is the additional amount that arises in year t. Following Mueller (1986), Waring (1996), and McGahan and Porter (2002, 2003), the incremental component is stipulated to follow the firstorder autoregressive process given by si;k;t ¼ ri;k si;k;t
1
þ ui;k;t
(3)
where ri,k is the rate of persistence in the incremental components and ui,k,t is a random shock at time t with mean zero and constant but unknown variance. Algebraic substitution yields: fi;k;t ¼ ai;k þ ri;k fi;k;t
1
þ ui;k;t
(4)
This equation indicates that fi,k,t consists of a drift, ai,k ¼ (1 ri,k)fi,k; the
Industry Performance and Changes in Competitor Characteristics
641
persistence rate in incremental components, ri,k, times the prior year’s effect, fi,k,t 1; and the random shock ui,k,t. Any of the terms in Eqs. (2), (3), and (4) – including the fixed component, the incremental component, and the persistence rate – may take on positive or negative values. Note that the rate of persistence in Eqs. (3) and (4) is assumed to be constant for a business over the entire period. A negative persistence rate indicates oscillation in the continuing portion of the incremental components of a characteristic. A negative rate might arise if a business manipulates its accounting policies to elevate earnings in one year at the expense of earnings in the next year. The fixed component of a characteristic, fi,k, takes on a negative value when the average value of fi,k,t for the business is negative (i.e., when the business is below the industry in the characteristic on average over the period). In these cases, convergence toward the mean occurs when the average incremental component, si,k,t, is negative and the rate of persistence, ri,k, is between 0 and 1.6 When the fixed component of a characteristic is positive, then convergence occurs when the incremental component is positive and the rate of persistence is between 0 and 1. Ordinary-least-squares estimation of Eq. (4) gives biased estimates of ri,k because the errors on the observations for corporation k in industry i at time t are correlated with the errors in time t+1.7 This bias can be corrected using a formula developed by Nickell (1981).8 With the correction, the procedure gives estimates of persistence for each corporation k within each of its industries i. From the corrected persistence estimates, the fixed and incremental components of each characteristic may be imputed. Measures of diversity in the characteristics of direct competitors are obtained from estimates of fi,k and ri,k for each business (i.e., for each corporation k in industry i). The first summary measure for each industry, di, captures diversity in the fixed component of competitor characteristics:
di ¼
P
absðf i;k f i Þ=wi;k
k
P k
1=wi;k
!
(5)
with fi equal to the average of the estimated fi,k’s among members of the industry and wi,k equal to the variance of the estimate of the fi,k’s.9 Eq. (5) is a weighted average of the absolute value of fi,k among direct competitors in industry i, with the weights equal to the inverses of the variances of the estimates of the fi,k. The intuition for the weighting is that the average should be influenced by the precision of the estimates.
642
ANITA M. MCGAHAN
The second summary measure of diversity for each industry, ci, is the average rate of convergence in the incremental component of the characteristic among direct competitors:
ci ¼
P
avgi ðconvðri;k ÞÞ=vi;k
½ðconvðri;k ÞÞ
k
P
1=vi;k
(6)
k
with avgi(conv(ri,k)) equal to the average rate of convergence and vi,k equal to the variance of the estimate of ri,k.10 Again, the weights account for precision in the estimates. The function conv(ri,k) gives the value ri,k if the fixed and incremental components of fi,k,t are both positive or both negative, which indicates that ri,k captures the rate of convergence toward the mean. The function conv(ri,k) gives the value 1 ri,k if the fixed and incremental components of fi,k,t take different signs, so that conv(ri,k) captures the rate of convergence. These two measures of diversity are useful because the dataset that supports the analysis represents the large firms in each industry – that is, those that are publicly traded and meet the various screening criteria described below. It is useful to consider that, on average and in any particular year, only a handful of competitors are represented in an industry. One of the most serious limitations of the analysis in this paper is that the Compustat reports contain no comprehensive information on the population count in each industry. Thus, traditional measures of diversity such as the Blau index (see Boone, van Witteloostuijn, & Carroll, 2002) are not good representations of true diversity across industries. 2.3.2. Linking Diversity in Competitor Characteristics with Industry Performance Eq. (7) links industry performance with diversity in the characteristics of direct competitors over the period: a i ¼ b1 þ b2 d i þ o i
(7)
In Eq. (7), ai is estimated industry performance, and di is the summary measure of diversity in the characteristic among direct competitors. The hypothesis of a positive association between industry performance and diversity in the characteristic is given by b240; the hypothesis of a negative association is given by b2o0; and the hypothesis of no association is given by b2ffi0.
Industry Performance and Changes in Competitor Characteristics
643
Eq. (8) links industry performance with the average rate of convergence in the incremental components of the competitor characteristics: a i ¼ b3 þ b4 c i þ Z i
(8)
In this equation, ci is the summary measure of convergence toward the mean among direct competitors. Thus, Eq. (8) captures diversity in the characteristic at the second order, i.e., in the rate of change in the characteristic. The hypothesis of a positive association between industry performance and convergence in the characteristic among direct competitors is given by b440; the hypothesis of a negative association is given by b4o0; and the hypothesis of no association is given by b4ffi0.
3. DATA Data for the analysis is drawn from two Compustat files, which cover companies with equity that is publicly traded on American markets. All information necessary to assess performance is taken from the Compustat Basic file on active and research companies from 1970 to 1997. Information for years prior to 1981 is used only to calculate historical series relevant to Tobin’s q. The series ends in 1997 for several reasons. First, restatements of accounting results were so pervasive after 1997 that results for those years were deemed somewhat unreliable. Second, Compustat changed its conventions for reporting on Business Segments, which created misleading discontinuities in the series reported by many organizations. Tobin’s q is calculated using the procedure in Lang and Stulz (1994) and McGahan (1999b), which follows Lindenberg and Ross (1980) (also see Salinger, 1984).11 The Compustat Basic file is also the source for data on the corporation’s aggregate sales; assets; the replacement value of assets; the percent of assets in property, plant & equipment; and the percent of the replacement value of assets in property, plant & equipment. The second source of information is the Compustat Business-Segment Reports for 1981–1997. The Business-Segment Reports contain information on corporate activities by industry, which are defined by 4-digit SIC codes. (The activity of a corporation within a 4-digit SIC category is called a ‘‘business segment.’’) SEC guidelines require the reporting of information on 4-digit activity that comprises 10% or more of a company’s sales. As a consequence, the Business-Segment Reports contain information on up to 10 segments per corporation in each year. Segment data prior to 1981 is excluded because of inconsistent reporting. The Business-Segment Reports
644
ANITA M. MCGAHAN
are the source of data for each segment on sales, assets, and operating income. Both the Compustat Basic file and the Compustat Business-Segment Reports are screened for anomalies (in the same ways as in McGahan (1999b) and McGahan and Porter (1997, 1999, 2002). The original Basic file for 1981 to 1997 contains 119,008 observations on corporations by year. Three hundred and thirty-three observations are excluded on corporations with market values of zero. Another 2,514 observations are excluded for lack of information on accounting profits. Corporations with less than $50 million in assets – which are 55,059 in number – are excluded because their market values may be distorted by infrequent trading.12 Another 20,358 observations are excluded for lack of information on the distribution of assets by segment. After the screens, the Basic file contains 40,744 observations, each of which applies to a corporation in a specific year. The original Business-Segment Reports for 1981 to 1997 contain 186,396 records on segments by year. A total of 3,168 observations are excluded because they do not contain a primary SIC code. Another 26,187 observations are excluded because they are assigned to industries with the titles ‘‘not elsewhere classified,’’ ‘‘non-classifiable establishments,’’ and ‘‘government, excluding finance.’’ ‘‘Depository institutions’’ with SIC’s between 6,000 and 6,999 – which are 19,796 in number – are also excluded. Four thousand one hundred and twenty-three observations are excluded because they represent the single organization in an SIC category in a particular year. The effects of industry cannot be distinguished from the effects of competitive position for these segments. Observations on small segments with sales of less than $10 million (33,807 records) and with assets of less than $10 million (7,093 records) are excluded to follow precedent in the literature. Another 30,315 observations cannot be associated with a corporation in the screened Basic file and are eliminated. After the screens, the Business-Segment dataset contains 61,907 records, each of which applies to a corporation’s activities in an industry in a specific year. After the screening and merging of the files, the dataset contains information on 6,067 corporations that participate in a total of 612 industries defined by 4-digit SIC codes. Thirty percent of the corporations participate in more than one industry after application of the screens. On an average, industries are populated for 12.2 years, and corporations are represented in the dataset for 6.7 years. Because only 6.7 years of data are available on average for each firm, the analysis focuses on the fixed and intemporally persistent elements of each firm characteristic, and then associates competitor diversity with an industry average that holds over the whole period.
Industry Performance and Changes in Competitor Characteristics
645
Thus, the analysis is fundamentally cross-sectional rather than longitudinal. The principal reason for this restriction relates to available data: Compustat uses information reported to the SEC by a large number of companies, but because regulations do not require longitudinal comparisons, the robustness of the data in time series is lower than in cross section. The results of this analysis are therefore interpreted as cross sectional. Industries contain an average of 19.6 segments over the entire period, or 8.3 segments per year. The dataset contains a total of 11,192 different segments. The average corporation reports on 1.52 segments and the average segment has assets of $985 million, which suggests that a typical Compustat segment reflects actual operating activity in several related 4-digit SIC codes. This suggestion is supported by Montgomery (1994, p. 164), who reports that the Fortune 500 participated in an average of 10–11 different 4-digit SICs during the period from 1985 to 1992. If the industries in Compustat contain activity that should be assigned to a different 4-digit category, then the estimates of industry performance may be dampened. If the aggregation smoothes differences in the competitive positions of operating business units, then estimates of differences in characteristics also may be dampened. The aggregation essentially introduces noise into estimates of each type. As a result, the results must be interpreted at the segment level and not at the level of operating business units. The advantage of relying on the Compustat files for this analysis in its potential for supporting comprehensive measures of firm performance. The disadvantage is that it does not contain information about industry structure or even about firm survival. In general, only the largest firms within industries are represented. As a result, some processes related to isolationism and mutual forbearance that involve small or private firms are not represented here. Prior research in organizational ecology suggests that smaller firms may be particularly involved in isolationism (Dobrev & Carroll, 2003) or that mutual forbearance requires at least some competitor diversity (Baum & Mezias, 1992). Table 2 describes the screened data at the corporate level. Panel (a) describes the data by year, and panel (b) describes the data by economic sector. In panel (a), the first column shows the number of corporations in each year. The next two columns show the average values of Tobin’s q and of accounting profit by year. The average value of Tobin’s q is 1.252. Average accounting profit equals 8.82% for the entire sample. Tobin’s q and accounting profit are correlated at 0.174. The next column in panel (a) shows that corporations report on an average of 1.52 of the identified industries. On average, 88.6% of corporate assets are attributed to business segments in the screened dataset.
646
Table 2. Year
No. of Corporate Assets
Tobin’s q Accounting Profita (%)
Screened Corporate Data.
Segments Per Assets in Corporation Segments (%)
Year
No. of Corporates Assets
1990 1991 1992 1993 1994 1995 1996 1997 Total
2,340 2,442 2,603 2,207 2,463 3,170 2,913 2,825 6,067
Tobin’s q Accounting Profita (%)
Segments Per Corporation
Assets in Segments (%)
1.48 1.46 1.41 1.46 1.44 1.38 1.35 1.28 1.52
88.3 89.1 90.0 90.4 91.8 92.7 93.0 92.5 88.6
(a) By Year 1981 1982 1983 1984 1985 1986 1987 1988 1989
0.949 1.008 1.149 1.041 1.112 1.185 1.130 1.116 1.155
13.05 10.18 10.29 11.06 9.50 8.33 9.51 9.94 9.39
Brief Description of Sector
1.82 1.86 1.78 1.73 1.64 1.57 1.54 1.50 1.49
82.8 84.5 84.6 85.3 85.6 86.3 87.4 87.4 87.7
1.078 1.225 1.293 1.384 1.308 1.502 1.622 1.637 1.252
8.23 7.19 7.69 7.44 8.42 8.08 7.29 6.88 8.82
No. SICs
No. Corporate Assets
Tobin’s q
Accounting. Profita (%)
Segments Per Corporation
Assets in Segments (%)
195 216 36 90
1,688 1,935 1,006 991
1.194 1.277 0.983 1.253
9.34 8.75 8.63 8.81
1.70 1.59 1.49 1.27
86.3 84.5 93.9 93.5
(b) By Sector 0,1,2 3 4 5
Agriculture, mining Manufacturing Transportation Wholesale and retail trade
ANITA M. MCGAHAN
First Digit of SIC
1,835 1,983 2,097 2,138 2,226 2,316 2,416 2,444 2,326
8 Total
Lodging and entertainment Services
ROA
Q
51
639
1.898
8.27
1.24
91.3
24 612
297 6,067b
1.549 1.252
7.21 8.82
1.26 1.52
89.1 88.6
Assets
Sales
Operating Margin
PPE
Turnover
Focus
1.000 0.636 0.014 0.049 0.028 0.034
1.000 0.016 0.047 0.057 0.097
1.000 0.003 0.143 0.000
1.000 0.010 0.185
1.000 0.034
1.000
(c) Correlation Matrix ROA Q Assets Sales OpMargin PPE Turnover Focus a
1.000 0.256 0.010 0.004 0.073 0.006 0.029 0.011
1.000 0.040 0.024 0.007 0.085 0.013 0.012
Ratio of earnings before interest and taxes to the book value of assets for the corporation. The sum of the entries in the column is 6,556 rather than 6,067 because some corporations change sectoral affiliation during the period under study. b
Industry Performance and Changes in Competitor Characteristics
7
647
648
ANITA M. MCGAHAN
Panel (b) of Table 2 describes firm characteristics. Manufacturing includes the largest number of SICs and companies (for this summary, each corporation is assigned to a sector based on the SIC category of its largest segment). The dataset probably does not represent a cross section of actual economic activity in the service sector. Only 24 service industries are represented because Compustat does not include information on proprietorships, partnerships, privately held corporations, and smaller firms, all of which are significant in the sector. Sectors are similar in the average number of segments per corporation and the coverage of corporate assets in included segments. Panel (c) of Table 2 shows the correlation matrix for various characteristics of corporations (with data aggregated to industry averages to allow for comparability with q and ROA). Note that the two measures of performance – namely q and ROA – are correlated at 25.6%. No firm characteristic is highly correlated with q. The characteristic most highly related to ROA is operating margin, but even in this instance the rate of correlation is just 7.3%. None of the measures of firm strategy (i.e., of firm characteristics) are highly correlated except sales and assets, which are substitute measures of size. Thus, multicolinearity is not of concern.
4. EMPIRICAL RESULTS The empirical results are presented in several groups. First, the ANOVA generates information on the highest- and lowest-performance industries among the publicly traded companies from 1981 to 1997. Second, the assessments of competitor characteristics generate measures of diversity across competitor characteristics. Finally, an evaluation of the relationships between the industry coefficients and diversity measures addresses the relationship between diversity in competitor characteristics and industry performance. 4.1. Assessments of Industry Performance Table 3 shows the results of ANOVA estimation on Eqs. (1a) and (1b). Instead of reporting the 17-year estimates and the 613 industry estimates, the table shows the contribution to R2 of the year and industry effects. The table indicates that industry differences contribute significantly to both Tobin’s q and accounting profitability. Table 4 shows the average values of the industry coefficients by each sector. The entries describe the average premium or discount in performance among businesses in the sector. For example, the average Tobin’s q of
649
Industry Performance and Changes in Competitor Characteristics
Table 3.
a
Year
Industryc Modeld a
ANOVA on Eqs. (1a) and (1b). Tobin’s q
Accounting Profitability
0.041 (99%)b 0.238 (99%)
0.008 (99%) 0.104 (99%)
0.279
0.112
2
R in the model of year effects. Probability with which a Wald test rejects the hypothesis of equality among the class of effects. c Increment to R2 in the model of year and industry effects over the model of year effects. d Total R2 in the model of year and industry effects. b
Table 4.
Estimated Value of Industry Coefficients,a ai.
SIC
Sector Description
Tobin’s q
Accounting Profit
0,1,2 3 4 5 7 8
Agriculture and mining Manufactures Transportation Wholesale and retail trade Lodging/entertainment Services
0.063 0.108 0.207 0.072 0.242 0.646
1.90 1.40 1.75 1.52 0.52 1.52
a
Simple average: This measure does not account for the number or sizes of firms in each industry.
agriculture and mining businesses is lower than the economic mean by 0.063. The average accounting profitability of the agriculture and mining businesses is 1.90% above the overall economic mean, however. The difference may arise because investors anticipate less favorable conditions for the sector in the future than currently. The evidence shows the same type of relationship in the manufacturing sector, which trades at a discount on Tobin’s q despite higher than average accounting profitability. The transportation and wholesale/retail trade sectors are populated by businesses with both average Tobin’s q and average accounting profitability below the economic mean. The average lodging, entertainment, and service business has both Tobin’s q and accounting profitability above the economic mean. Note that the dataset may exclude a large proportion of service businesses because it captures only publicly traded firms. There is little evidence suggesting that privately held service businesses are less profitable on average than the publicly traded service businesses, however.
650
ANITA M. MCGAHAN
Table 5 shows the simple average of the estimated industry coefficients for selected industries. The results in panel (a) show that the industries with the highest Tobin’s q are not the same as those with the highest accounting profitability. Only industry 7,383, news syndicates, appears in both columns. Differences may arise either because investor expectations about future performance differ from current return or because the replacement values differ from the book values of assets.13 The industry with the highest Tobin’s q is 8,221, Professional Schools. This category includes just two publicly traded businesses in the dataset: The Devry Institute and the Apollo Group. Prepackaged software and chewing gum also appear in the list. Industry 2,050, bakery products, has the highest accounting profitability and includes grand metropolitan and P&C food products. The 10 industries listed in panel (a) with the highest accounting profitability contain an average of 6.1 competitors. Panel (b) reports the values for the lowest-performing industries. The industry with the lowest Tobin’s q, 7,353, is heavy construction equipment rental and leasing, which contains four incumbents. Other industries out of favor on the financial markets include grain and field beans wholesaling; folding paperboard boxes manufacture; and lead and zinc ore mining. None of these industries are ranked among the lowest in accounting profitability. The lowest-profit industries include professional sports clubs and promoters (an industry charged with the diversion of profitability into sister media businesses); meat wholesaling; and recreational vehicle parks and camping sites. The industries with the lowest accounting profitability as listed in panel (b) contain 4.2 competitors on average. Thus, their lower profitability is not directly associated with a radical difference in the number of incumbents. Panel (c) shows the industries with the least variability from the economic average of Tobin’s q and of accounting profitability. Again, there is no duplication in the lists of industries ranked by Tobin’s q and by accounting profitability. The industries with accounting profitability near the mean are significantly more populated than those in the previous panels, however. On average, the industries listed in the right-hand column of panel (c) contain 10.5 competitors. The pattern is somewhat different for Tobin’s q. One industry, 7,372 prepackaged software, drives the average number of competitors among high-performance industries above those of moderate-performance industries. Thus, for Tobin’s q, the highest-performance industries have significantly more competitors than moderateperformance industries, which have significantly more competitors than low-performance industries. This preliminary evidence runs contrary to the theoretical traditions that associate low industry performance with an excessive number of competitors.
SIC
Estimated Value of Industry Coefficients for Selected Industriesa.
Model of Tobin’s q Description
Model of Accounting Profit Estimated ai
SIC
Description
Estimated ai
(a) Industries with Highest Estimates of ai 8221 5251 8721 7375 8243 7372 7383 2067 2836 8741
Professional schools Hardware stores Accounting and auditing services Information retrieval services Data processing schools Prepackaged software News syndicates Chewing gum Biological products, exc. diagnostics Management and public relations svcs.
3.98 2.84 2.58 2.28 1.87 1.86 1.78 1.61 1.61 1.42
2050 2023 2131 7221 7383 2111 2047 3421 2032 4724
Bakery products Dry and condensed dairy products Chewing tobacco Portrait photographic studios News syndicates Cigarettes Dog and cat food manufacture Cutlery manufacture Canned specialties manufacture Travel agencies
43.16 35.60 35.13 33.83 27.44 26.58 23.42 21.02 20.80 20.76
0.86 0.84 0.75 0.74 0.71 0.68
7941 5147 1044 2092 3320 7033
Pro. sports clubs/promoters Meat wholesaling Silver ore mining Prepared or frozen fish/seafood Iron and steel foundries Rec. vehicle parks/campsites
25.93 25.06 22.91 19.52 19.34 18.40
Industry Performance and Changes in Competitor Characteristics
Table 5.
(b) Industries with Lowest Estimates of ai Heavy constr. equipment leasing Grain and field beans wholesaling Prepared or frozen fish/seafood Cordage and twine manufacturing Ornamental shrub and tree services Molded rubber goods
651
7353 5153 2092 2298 783 3061
652
Table 5. (Continued ) SIC
Model of Tobin’s q Description
5033 4822 2657 1031
Roofing, siding and insulation mat’ls Telegraph and other message commu. Folding paperboard boxes Lead and zinc ore mining
Model of Accounting Profit Estimated ai
SIC
Description
0.64 0.64 0.63 0.62
4822 3313 3330 5170
Telegraph and other message communication Electrometallurgical pdts., exc. steel Primary smelting – nonferrous mtl. Petroleum distribution
0.01 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00
3553 3844 3631 211 3554 7382 2678 4222 5984 3086
Woodworking machinery manuf. X-ray equipment manufacture Household cooking equipment Mfr. Beef cattle feedlots Paper industries machinery Mfr. Security systems services Stationery products manufacture Refrigerated warehousing and storage Bottled gas dealers Plastic foam product manufacture
Estimated ai 18.32 16.61 16.20 14.97
(c) Industries with Estimates of ai Closest to Zero 2111 2052 1791 1522 2510 2095 3822 7213 3570 2671
Simple average: This measure does not account for the number or sizes of firms in each industry.
0.10 0.07 0.04 0.02 0.01 0.03 0.06 0.09 0.11 0.16
ANITA M. MCGAHAN
a
Cigarettes Cookie and cracker manufacture Structural steel erection contracting General contractors – residential Household furniture manufacture Roasted coffee manufacture Heating/cooling controls Linen supply services Computer manufacture Coating packaging paper/film
Industry Performance and Changes in Competitor Characteristics
653
4.2. Measures of Competitors’ Characteristics Table 6 shows information on the average sizes, operating margins, asset competition, asset turnover, and focus of the businesses in the dataset. Businesses in agriculture, mining, wholesale trade, and retail trade are over three times as large on average as businesses in entertainment, lodging, and services. The measures of competitor size indicate a great deal of diversity. The standard deviations of sales and assets are so large that they suggest strong skewness in the distributions. Industries tend to be populated by some businesses with sales less than the average and a small number of very large businesses with sales substantially above the average. This skewness is particularly strong in agriculture, mining, and manufacturing. Note that even this descriptive finding points to the value of complementary findings from resource-partitioning theory, which stipulates a longitudinal process that may explain this cross-sectional result (see Carroll & Hannan, 1995; Boone et al., 2002) because industries have their inceptions at different times and evolve at different rates (see McGahan, 2004). The evidence also indicates substantial variability in the operating margins of competitors. The highest average margins by far arise in transportation businesses. In this sector, the variability of operating margins is lower than in other sectors, however. The standard deviation of operating margins in transportation is 146% of the average, whereas the standard deviation in all sectors is over 300% of the average. The lesser variation among transportation companies may reflect less scope for strategic pricing because of greater competitive pressures, for example. The high average ratio for transportation may occur because of high interest charges among members of the sector. The greatest variability in operating margin arises in agriculture and mining. This variation may occur because of broader differences in the markups among direct competitors or between different industries within the sector. The subsequent section develops industry-specific measures of variability to control for these possibilities. Table 6 also shows an average asset composition of 41% in property, plant & equipment. This measure, which reflects the capital intensity of the businesses, is also quite variable across all sectors of the economy. The ratio is highest in transportation and lowest in manufacturing. Of course, only the booked assets of firms are reflected in the denominator of the ratio; brand capital and the knowledge stock (from accumulated R&D) are not capitalized. If these intangibles were in the denominator, then the ratios for entertainment, lodging, and services would probably be significantly lower than for manufacturing.
654
Table 6. Feature
(i) Size
Measure
Sales ($millions) Assets ($millions)
(ii) Operating margin
a
All Sectors
Agriculture and Mining
Manufacture
Transport
Wholesale/ Retail
Entertainment/ Lodging
Services
1073 (3,957) 985 (3,465) 0.09 (0.28)
1383 (5,244) 1,189 (3,581) 0.08 (0.42)
904 (3,883) 730 (3,408) 0.07 (0.16)
915 (2,823) 1663 (4,919) 0.15 (0.22)
1364 (2,852) 590 (1,296) 0.05 (0.07)
430 (818) 489 (1,033) 0.09 (0.20)
419 (757) 406 (886) 0.08 (0.18)
0.41 (0.23) 1.46 (1.38) 0.52 (0.76)
0.46 (0.23) 1.36 (1.06) 0.66 (0.98)
0.30 (0.15) 1.46 (0.66) 0.51 (0.62)
0.63 (0.23) 0.80 (0.73) 0.41 (0.61)
0.36 (0.21) 2.57 (1.63) 0.42 (0.65)
0.33 (0.26) 1.25 (1.00) 0.39 (0.66)
0.34 (0.23) 1.73 (5.00) 0.39 (0.63)
Simple average of the values with standard deviation in parentheses.
ANITA M. MCGAHAN
(iii) Asset composition (iv) Asset utilization (v) Corporate focus
Ratio of operating income to sales (%) Assets in PP&E (%) Asset turnover (sales/assets) Diversification index
Competitor Characteristicsa.
Industry Performance and Changes in Competitor Characteristics
655
The asset turnover ratio can be interpreted as the amount of assets necessary to support a given level of sales. The measure differs from asset composition in that it does not reflect capital intensity. All booked assets – including inventory, accounts receivable, cash, and goodwill – are considered in the assessment. The statistics indicate that the wholesale- and retail-trade sector is by far the most asset-intensive. The average transportation business requires substantially fewer assets to generate a specific level of sales than the average business from any other sector. The final measure in Table 6, corporate focus, reflects the tendency of each competitor to belong to a diversified corporation. The index value is highest when a firm participates in a number of different sectors, and lowest when a firm is focused in a particular sector. The prevalence of diversification is greatest on average among businesses in the agriculture and mining sector, and least among businesses in the entertainment, lodging, and service sectors. 4.3. Assessments of Diversity in Competitor Characteristics The final step in the analysis is to develop measures of diversity in each of the characteristics of the direct competitors within an industry. These measures can then be linked to industry-average performance. Tables 7 and 8 summarize the diversity measures after controlling for industry differences. Table 6 reports on the enduring tendencies of competitors to differ in characteristics. Each entry in the table is the simple average of the absolute differences between each competitor’s fixed effect, fi,k , and the industry average, fi,k, for each of the characteristics. Table 8 reports on diversity among competitors in their tendencies to regress toward the industry mean in their characteristics. Each entry is the simple average of the absolute differences in convergence rates among direct competitors. Both tables indicate strong diversity in the characteristics of competitors. For example, the first row of Table 7 shows that the average business deviates from its industry average in sales by $1,373 million. A comparison with Table 6 indicates that the average deviation from the mean is even greater than the mean itself: businesses post average sales of $1,073 million (see Table 6). Since the dataset excludes all businesses with sales below $10 million, this regularity again indicates skewness in the size distribution of competitors. The diversity in size is especially great in the manufacturing and transportation sectors, where the average deviation is especially large as a proportion of the industry means. Table 7 also indicates substantial diversity among competitors in their operating margins, asset composition, asset utilization, and corporate focus.
656
Table 7.
Diversity in Competitor Characteristicsa.
Feature
Measure
All Sectors
Agriculture and Mining
Manufacture
Transport
Wholesale/ Retail
Entertainment/ Lodging
Services
(i) Size
Sales ($millions) Assets ($millions) Ratio of operating income to sales (%) Assets in PP&E (%) Asset turnover (sales/assets) Diversification index
1,373 1,306 0.28
1,671 1,539 0.38
1,334 1,042 0.21
1,207 2,166 0.37
1,545 678 0.08
451 592 0.27
461 588 0.45
0.29
0.34
0.25
0.31
0.27
0.27
0.29
1.30
1.17
1.32
0.75
2.11
0.98
3.11
0.91
0.98
1.15
0.59
0.76
0.67
0.71
(ii) Operating margin (iii) Asset composition (iv) Asset utilization (v) Corporate focus
Simple average abs(fi,k fi ) from Eq. (5).
ANITA M. MCGAHAN
a
Convergence in the Incremental Components of Competitor Characteristicsa.
Feature
Measure
All Sectors
Agriculture and Mining
Wholesale/ Manufacture
Entertainment/ Transport
Retail
Lodging
Services
(i) Size
Sales Assets Ratio of operating income to sales Assets in PP&E
0.51 0.53 2.05
0.47 0.50 2.40
0.48 0.49 2.85
0.52 0.53 0.92
0.55 0.58 1.02
0.64 0.67 1.28
0.64 0.58 2.97
0.26
0.26
0.27
0.28
0.37
0.22
0.27
Asset turnover (sales/assets) Diversification index
0.28
0.29
0.26
0.31
0.26
0.25
0.25
0.34
0.34
0.36
0.33
0.40
0.24
0.24
(ii) Operating margin (iii) Asset composition (iv) Asset utilization (v) Corporate focus a
Simple average values of conv(ri,k) avgi(conv(ri,k)) from Eq. (6).
Industry Performance and Changes in Competitor Characteristics
Table 8.
657
658
ANITA M. MCGAHAN
For example, consider the operating margins of competitors in the agriculture and mining sector. Table 6 indicates that the mean operating margin is 8% of sales. Table 7 shows that the standard departure from the industry average within the agriculture and mining sector is 38%, i.e., businesses in the sector typically show operating margins that differ from their industry means by 0.38. The results on asset composition are also dramatic. In agriculture and mining, the average asset composition is 46% in PP&E, with a typical absolute deviation equal to 34%. The average deviation on asset turnover is quite large, which again indicates skewness in the distribution. Typically, a small number of competitors are significantly more efficient than their rivals, and thereby drive up the average deviation in asset utilization. The average deviation in corporate focus is also great. Industries typically contain a number of single-segment businesses (for which the corporate-focus measure is held to zero) and a few substantially diversified businesses. The results on convergence also reveal significant diversity among rivals. For example, consider the first entry in the table, which reports an average deviation of 0.51 in the convergence rate of sales toward industry averages. This statistic indicates that, typically, the rates of convergence of direct competitors differ by 0.51. One competitor in an industry may have a rate of sales convergence of 75% while another competitor has a rate of 126%. (A rate of 126% indicates divergence rather than convergence toward the industry mean.) The operating income rates differ so much that on average it is inappropriate to consider that convergence occurs. Thus, across all sectors on all measures, the evidence suggests great diversity even in convergence rates. Table 9 reports the results of the regressions specified in Eqs. (7) and (8) relating diversity to industry-average performance. The first two columns, labeled ‘‘Tobin’s q,’’ report the results when industry-average performance is measured by Tobin’s q. The second two columns, labeled ‘‘Accounting Profitability,’’ report the outcomes when performance is measured by the ratio of operating income to assets. In all four columns, the industry-average performance is captured by ai as specified in either Eqs. (1a) or (1b). For each pair, the first column reports the estimate of b2 from Eq. (7) and the second column reports the estimate of b4 from Eq. (8). The figures in parentheses represent the probabilities that the estimated coefficients differ from zero. Every estimate in Table 8 except two yields estimates that do not differ significantly from zero at the 5% level. Thus, we cannot reject the null hypothesis of no association between diversity and industry-average performance. The two significant estimates have negative coefficients, i.e., they
659
Industry Performance and Changes in Competitor Characteristics
Table 9.
Regression of Diversity in Competitor Characteristics on Industry Performance.
Feature
(i) Size
Measure
Sales Assets
(ii) Operating margin
Ratio of operating income to sales
(iii) Asset composition
Assets in PP&E
(iv) Asset utilization
Asset turnover
(v) Corporate focus
Diversification index
Tobin’s q
Accounting Profit’y
Eq. (7) Fixed
Eq. (8) Conv.
Eq. (7) Fixed
Eq. (8) Conv.
0.000a (90.4%) 0.000 (93.8%) 0.000 (87.3%) 0.000 (1.4%) 0.000 (65.7%) 0.000 (16.4%)
0.001b (18.4%) 0.005 (33.9%) 0.000 (41.7%) 0.008 (92.9%) 0.007 (46.8%) 0.017 (27.5%)
0.000a (38.5%) 0.000 (18.4%) 0.000 (89.6) 0.000 (36.1%) 0.000 (85.8%) 0.000 (84.6%)
0.005b (93.8%) 0.046 (0.01%) 0.002 (82.4%) 0.022 (98.7%) 0.201 (13.0%) 0.087 (72.1%)
Significant at the 5% level. a
Estimates of b2 in Eq. (7) with Pr(b2) ¼ / ¼ 0 from t-test in parentheses. Estimates of b4 in Eq. (8) with Pr(b4) ¼ / ¼ 0 from t-test in parentheses.
b
associate greater diversity with lower industry-average performance. Even for these two cases, however, the estimated coefficient is quite low. Three tests on the specification broadly confirm the robustness of the results in Table 9. In the first test, results were obtained for measures of diversity that were not weighted by the inverses of the variances of the estimates. This test is important because the measures of diversity used in Eqs. (7) and (8) are constructed to dampen the influence of both businesses with high variance in their characteristics and businesses with a short time series. Under the specification test, all businesses in each industry received equal weight regardless of the length of their series and regardless of the variation in their characteristics. The test yielded four significant coefficients (at the 5% level) out of the 24 that were estimated. In all four cases, performance reflected Tobin’s q, and the coefficients were small but positive. Two of the four positive coefficients applied to convergence in the sizes of rivals. Thus, the test provides some evidence that investor expectations are greater when the high-variance competitors differ from their rivals in growth rates. The second test on specification involved obtaining results from generalized-least-squares (GLS) estimates rather than ordinary-least-squares estimates of Eqs. (7) and (8). In all of the regressions, there is evidence of
660
ANITA M. MCGAHAN
heteroskedasticity in the dependent variable, which justifies the application of GLS methods.14 Of course, the ordinary-least-squares estimates reported in Table 8 are unbiased, although the significance of the estimates may be biased downward. The test yields five significant coefficients out of the 24 estimates. Three of the significant coefficients involve accounting profitability and diversity in the fixed effects of asset utilization, the convergence rates in focus, and convergence in sales. In each case, the estimated coefficients are negative, which suggests that greater diversity is tied to lowperformance industries. The remaining two of the significant coefficients involve Tobin’s q and diversity in the fixed effects of assets and of asset composition. In these cases, diversity is associated with greater performance. Thus, the test again suggests that investors may use operating characteristics as an indicator of future returns. The inconsistent signs on the estimated coefficients as well as the low number of significant coefficients again confirms the robustness of the general result of little strong association between competitor diversity and industry-average performance. The third test retained the GLS estimation procedure (which tends to raise the significance of the estimates) and introduced the potential for a non-linear effect. Specifically, Eq. (7) was extended to ai ¼ b1+b2di+b5di2+oi and Eq. (8) was extended to ai ¼ b3+b4ci+b6ci2+Zi. Just 5 of the 24 estimated linear coefficients (i.e., b2 and b4) were significant. Only one of the 24 estimated non-linear coefficients (i.e., b5 and b6) was significant; in this regression, the linear coefficient was not significant. Thus, the analysis generated no evidence of a curvilinear relationship between industry-average performance and diversity. The results in Table 9 have several implications. First, the diversity among competitors in high-performance industries provides weak support for the isolationist view despite the absence of positive estimated coefficients in the regressions. It is the large general degree of diversity among members of the typical industry that supports the isolationist view. The diversity among competitors by industry is especially striking because the dataset contains only the largest of the firms in the economy. Table 10 shows the concentration of businesses by industry. On average, 91.5% of industry revenue is attributable to the four largest competitors in an industry. In each year, the typical industry contains an average of 8.3 competitors. Evidence from the Internal Revenue Service (which unfortunately does not contain information on firm characteristics) indicates that the dataset excludes many small, privately held firms.15 Thus, the dataset clearly excludes from the typical industry a variety of smaller competitors that would differ substantially from those included. Even with these businesses excluded, the typical
661
Industry Performance and Changes in Competitor Characteristics
Table 10. Measure Percentage of revenue held by four largest comp’rs (%) Average number of competitors per industry (whole period) Average number of competitors per industry (per year)
Industry Concentration.
Sectors
Mining
Manufacture
Transport
Retail
Lodging
Services
91.5
92.5
92.3
77.0
91.8
93.1
92.1
19.6
18.0
18.2
44.1
17.6
17.9
19.0
8.3
8.1
7.0
18.6
8.2
7.1
6.3
competitor’s size, operating margin, asset composition, asset utilization, and focus differ from the industry averages in amounts that roughly equal the industry means. Thus, there is no general evidence of the symmetry required to support mutual forbearance. Second, the results on low-performance industries do not conform to the predictions of either the perfect-competition or the mutual-forbearance views. The perfect-competition view suggests that low-performance industries contain competitors that are less diverse in characteristics than those in high-performance industries. None of the analyses provided significant support for this hypothesis. The mutual-forbearance view suggests that lowperformance industries contain competitors with similar asset utilization and asset composition. While perfect competition and cooperation may occur in various settings, the analysis suggests that perfect competition is not the norm in low-performance industries, nor is cooperation the norm in high-performance industries.
5. CONCLUSION This paper reports on the relationship between diversity in competitor characteristics and industry-average profitability among a broad cross section of the publicly traded firms in the American economy from 1981 to 1997. Performance is measured using Tobin’s q and accounting profitability.
662
ANITA M. MCGAHAN
The analysis considers differences across competitors in business size, operating margin, asset composition, asset utilization, and the degree of corporate focus. The results indicate a high level of diversity among competitors in a broad variety of industries. Neither the fixed differences nor the rates of convergence in competitor characteristics are associated with industry-average performance. The results provide weak support for an isolationist view that associates high industry-average performance with competitor diversity. The level of diversity is too great to support the idea that mutual forbearance among rivals is the norm in high-performance industries. While the results weakly support the isolationist view of high-performance industries, they do not provide evidence of widespread ‘‘perfect competition’’ in low-performance industries. Contrary to prediction, competitors in the low-performance industries differ significantly in the measures most important to the theory: asset composition and asset utilization. A number of candidate explanations for the diversity among competitors in the low-performance industries arise. First, some efficient businesses may have become locked into poor positions that had once delivered better results. Second, entry barriers in some low-performance industries may allow inefficient leaders to continue operations. Third, some businesses may be engaged in investment programs and may be weathering low returns for future reward. Further theoretical and empirical research in industrial economics is needed to understand the prevalence of these and other explanations for low industry performance. The results reported here raise many questions for subsequent research. One of the central limitations of the approach in this paper is that the analysis is cross sectional rather than longitudinal. McGahan and Porter (1999, 2003) studied the sustainability of returns among a similar group of firms and found that the fixed and persistent elements of industry and firm performance were quite significant. Yet the short representation of businesses for 6.7 years within the dataset suggests that firm survival is an important issue. Additional research – perhaps integrating insights from organizational ecology as in Khessina (2006) – is needed to tie firm inception and survival to the processes that lead to competitor diversity. A second area for further research is to investigate the competitive processes that lead to diversity directly. Barnett (1997) describes a ‘‘red queen effect’’ where firms may invest repeatedly for distinctiveness but ultimately only preserve survival and financial performance. This research suggests that isolationism and mutual forbearance may be sensitively interrelated in ways that the large-scale analysis in this paper cannot discern. Further
Industry Performance and Changes in Competitor Characteristics
663
research that evaluates the potential for mutual and interrelated causality between entry, exit, and competitive processes is essential for understanding these relationships (see Kirzner (1997) for a discussion of some possibilities). Better information about industry age, life-cycle stage, and other characteristics would yield important insights along these lines. A third area for research involves understanding the particular circumstances in which isolationism and mutual forbearance each dominates. Again, insights from organizational ecology may shed light on the processes suggested by theory from industrial economics. Baum and Singh (1994) and Baum and Haveman (1997) provide surveys of research on niche overlap and identify population processes that are consistent with isolationism and mutual forbearance. In particular, they investigate prior research that shows how overlapping groups of competitors may elect either to compete or to cooperate based on long-term incentives for survival. Khessina (2006) suggests that detailed studies of particular settings using a single measure of overall diversity (such as the product-portfolio characteristic) can lead to insights about the intertemporal relationships between diversity, isolationism, and mutual forbearance. The findings in this study point in particular to one overriding conclusion: That large-scale statistical analysis in cross section yields only the broadest of insights about competitive processes. Detailed analysis of particular industry settings is crucial for identifying the relationships between return on investment, survival, distinctiveness, isolationism, and mutual forbearance.
NOTES 1. Tobin’s q is the ratio of the value of financial claims on a corporation to the replacement value of its assets, and therefore reflects a firm’s prospects for future return given its current stock of assets. Accounting profitability is measured as the ratio of income before interest and taxes to the book value of assets, and reflects the return on previous investments made by the firm. 2. The average business segment reported in the dataset probably covers the activity of several business units. This issue is explored in the discussion of data. 3. This specification follows Wernerfelt and Montgomery (1988) and McGahan (1999b). 4. This specification follows McGahan and Porter (1997, 2002). 5. McGahan and Porter (2002) and McGahan (1999b) show that the results of the ANOVA for year and industry effects are not significantly affected when the model is corrected for serial correlation or estimated on first differences. 6. In theory, convergence also may occur under some conditions when the incremental component is positive and the rate of persistence is negative. Although theoretically possible, this combination of estimates does not arise in the data.
664
ANITA M. MCGAHAN
7. Eq. (4) stipulates that fi,k,t is a function of ki,k,t. Now consider Eq. (4) for the same corporation in the same industry in the subsequent year: fi,k,t+1 ¼ ai,k+ri,kfi,k,t+ki,k,t+1. For this observation, fi,k,t+1 is a function of fi,k,t. The errors on the two observations are not independent because fi,k,t in the record for t+1 is correlated with ki,k,t in the record for t. This correlation introduces a systematic bias in the OLS estimate of ri,k. 8. See Waring (1996) and McGahan and Porter (2003, 2002) as well as Nickell (1981). Simulations verified that Nickell’s correction performs well except for extreme OLS estimates. The correction on each estimate of persistence was therefore limited to take on an absolute value less than or equal to 1. Thanks to Geoff Waring for discussions about this approach. 9. The variance of fi,k is wi,k ¼ [var(fi,k,t) est(ri,k)2var(fi,k,t 1)]/[1 est(ri,k)]. 10. The variance of each of the OLS estimates of persistence is given by vi,k ¼ var(ri,k) ¼ [var(fi,k,t) est(ri,k)2var(fi,k,t 1)]/[(n 2)var(fi,k,t 1)], where n represents the number of years of data for the business. Thanks to Arthur Schleifer for this derivation. 11. The numerator in Tobin’s q equals the sum of (i) the market value of outstanding equity at year end, and (ii) the book value of preferred stock and debt. The denominator in Tobin’s q is the sum of (i) the replacement value of inventory (which is assessed by recreating the schedule of inventory acquisition given the method of inventory accounting; and by applying the rate of inflation to the schedule); (ii) the replacement value of property, plant and equipment (which is assessed by recreating the schedule of acquisition; by applying a rate of depreciation of 5%; and by estimating price changes from the GDP deflator for non-residential fixed investments), and (iii) the book value of assets other than inventory and property, plant & equipment. The replacement value of inventory and of PP&E are calculated on the assumption that the replacement value equaled the book value in 1970. 12. The $50 million threshold is conservative with respect to precedents by Lang and Stulz (1994), which excludes corporations with less than $100 million in assets and Wernerfelt and Montgomery (1988), which includes just 247 large manufacturers. 13. McGahan (1999a, 1999b) suggests that differences between investor expectations and concurrent return are substantially more important than differences between the replacement values and booked values of assets. 14. Table 8 reports the ordinary-least-squares results rather than the GLS results because corrections for differences in the diversity of firm size, for example, could suppress the true differences of interest. Furthermore, the results of the specification test showed that the argument for generalized-least-squares techniques was immaterial. 15. Leo Troy (1981–1998) Handbook of business and financial ratios, Rutgers University Annual.
ACKNOWLEDGMENTS Thanks to the editors, Rich Makadok, Arthur Schleifer, Rogerio Victer, Geoff Waring, and colleagues in the strategy groups at Boston University,
Industry Performance and Changes in Competitor Characteristics
665
Harvard Business School and at Stanford’s Graduate School of Business for suggestions and discussions related to this paper. Thanks to James Schorr and Sarah Woolverton for help in extracting and assimilating data. The Systems Research Center, the Everett Lord Fund, and BUILDE at Boston University provided generous financial support for this project.
REFERENCES Abreu, D. (1986). Extremal equilibria of oligopolistic supergames. Journal of Economic Theory, 39, 191–225. Axelrod, R. (1984). The evolution of cooperation. New York: Basic Books. Bain, J. S. (1956). Barriers to new competition. Cambridge, MA: Harvard University Press. Baker, J. B. (1989). Identifying cartel policing under uncertainty: The U.S. steel industry, 1933–1939. Journal of Law and Economics, 32, S47–S76. Barnett, W. P. (1997). The dynamics of competitive intensity. Administrative Science Quarterly, 42(1), 128–160. Baum, J. A. C., & Haveman, H. A. (1997). Love thy neighbor? Differentiation and agglomeration in the Manhattan hotel industry, 1898–1990. Administrative Science Quarterly, 42(2), 304–338. Baum, J. A. C., & Mezias, S. J. (1992). Localized competition and organizational failure in the Manhattan hotel industry, 1898–1990. Administrative Science Quarterly, 37(2), 580–604. Baum, J. A. C., & Singh, J. V. (1994). Organizational niches and the dynamics of organizational mortality. The American Journal of Sociology, 100(2), 346–380. Boone, C., van Witteloostuijn, A., & Carroll, G. R. (2002). Resource distributions and the market partitioning: Dutch daily newspapers, 1968–1994. American Sociological Review, 67(3), 408–431. Bresnahan, T. F. (1987). Competition and collusion in the American automobile market: The 1955 Price War. Journal of Industrial Economics, 35(June), 457–482 Carroll, G. R., & Hannan, M. T. (1955). Resource partitioning. In: G. R. Carroll, & M. T. Hannan (Eds), Organizations in industry: Strategy, structure and selection (pp. 215–221). New York: Oxford University Press. Caves, R. E., Porter, M. E., & Spence, A. M. (1980). Competition in the open economy: A model applied to Canada. Cambridge, MA: Harvard University Press. Cubbin, J., & Geroski, P. (1987). The convergence of profits in the long run: Inter-firm and inter-industry comparisons. Journal of Industrial Economics, 35(June), 427–442. Dobrev, S. D., & Carroll, G. R. (2003). Size (and competition) among organizations: Modeling scale-based selection among automobile producers in four major countries, 1885–1981. Strategic Management Journal, 24, 541–558. Khessina, O. M. (2006). Direct and indirect effects of product portfolio on firm survival. In: J.A.C. Baum, S. D. Dobrev, & A. van Witteloostuijn (Eds), Strategy and ecology: Advances in strategic management (Vol. 23, pp. 589–630). Oxford, UK: JAI. Kirzner, I. (1997). Entrepreneurial discovery and the competitive market process: An Austrian approach. Journal of Economic Literature, 35(1), 60–85. Lang, L. H. P., & Stulz, R. M. (1994). Tobin’s q, corporate diversification, and firm performance. Journal of Political Economy, 102(6), 1248–1280.
666
ANITA M. MCGAHAN
McGahan, A. M. (1999a). Competition, strategy, and business performance. California Management Review, 41(3), 74–101. McGahan, A. M. (1999b). The performance of U.S. corporations: 1981–1994. Journal of Industrial Economics, 47(4), 373–398. McGahan, A. M. (1995). Cooperation in prices and capacities: Trade associations in brewing after repeal. Journal of Law and Economics, 38(2), 521–559. McGahan, A. M. (2004). How industries evolve. Boston, MA: Harvard Business School Press. McGahan, A. M., & Porter, M. E. (1997). How much does industry matter, really? Strategic Management Journal (Summer Special Issue), 15–30. McGahan, A. M., & Porter, M. E. (1999). The persistence of shocks to profitability. Review of Economics and Statistics, 81(1), 143–153. McGahan, A. M., & Porter, M. E. (2002). What do we know about variance in accounting profitability. Management Science, 48(7), 834–851. McGahan, A. M., & Porter, M. E. (2003). The emergence and sustainability of abnormal profits. Strategic Organization, 1(1), 79–108. Mueller, D. C. (1977). The persistence of profits above the norm. Economica, 44(November), 369–380. Mueller, D. C. (1986). Profits in the long run. Cambridge, UK: Cambridge University Press. Nickell, S. (1981). Biases in dynamic models with fixed effects. Econometrica, 49(6), 1417–1426. Rumelt, R. P. (1991). How much does industry matter? Strategic Management Journal, 12, 167–185. Schmalensee, R. (1985). Do markets differ much? American Economic Review, 75, 341–351. Troy, L. (1981–1998). Handbook of business and financial ratios. Rutgers University Annual. Waring, G. F. (1996). Industry differences in the persistence of firm-specific returns. American Economic Review (September), 1253–1265. Wernerfelt, B., & Montgomery, C. A. (1988). Tobin’s q and the importance of focus in firm performance. American Economic Review, 78, 246–250.
