Emerging Clusters: Theoretical, Empirical and Politcal Perspectives on the Initial Stage of Cluster Evolution (Industrial Dynamics, Entrepreneurship and Innovation)

Emerging Clusters

INDUSTRIAL DYNAMICS, ENTREPRENEURSHIP AND INNOVATION Series Editors: David B. Audretsch, Ameritech Chair of Economic Development, Indiana University, Bloomington, USA, Uwe Cantner, Friedrich Schiller University Jena, Germany and Dirk Fornahl, BAW Institute for Regional Economic Research, Bremen, Germany This series aims to discover important new insights in the relationship between the three cornerstones of economic development: industrial dynamics, entrepreneurship and innovation. In particular, the series will focus on the critical linkages between these three foundations. For example, the entry and exit of firms with differentiated growth processes can influence industrial development, but at the same time can also reflect the current industrial context shaping the entrepreneurial activities of single firms or individuals. A similar interaction linking industrial dynamics to entrepreneurship and innovation can also be identified. For instance, the particular technological regimes of industries may influence innovative activities, but the technological trajectory and type of innovative activity can, in turn, have a positive or negative influence on industry development. Innovation and entrepreneurship are also closely linked, since many types of entrepreneurial activities are barely distinguishable from similar innovative endeavors. Hence, the series addresses the linkages among the three fields in order to gain new findings concerning the nature of economic change. Theoretical, empirical as well as policy-oriented contributions are welcome. Titles in the series include: Innovation in Low-Tech Firms and Industries Edited by Hartmut Hirsch-Kreinsen and David Jacobsen Entrepreneurship and Openness Theory and Evidence Edited by David B. Audretsch, Robert E. Litan and Robert J. Strom Emerging Clusters Theoretical, Empirical and Political Perspectives on the Initial Stage of Cluster Evolution Edited by Dirk Fornahl, Sebastian Henn and Max-Peter Menzel

Emerging Clusters Theoretical, Empirical and Political Perspectives on the Initial Stage of Cluster Evolution

Edited by

Dirk Fornahl BAW Institute for Regional Economic Research, Bremen, Germany

Sebastian Henn Martin Luther University Halle-Wittenberg, Germany

Max-Peter Menzel University of Hamburg, Germany

INDUSTRIAL DYNAMICS, ENTREPRENEURSHIP AND INNOVATION

Edward Elgar Cheltenham, UK • Northampton, MA, USA

© Dirk Fornahl, Sebastian Henn and Max-Peter Menzel 2010 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 or photocopying, recording, or otherwise without the prior permission of the publisher. Published by Edward Elgar Publishing Limited The Lypiatts 15 Lansdown Road Cheltenham Glos GL50 2JA UK Edward Elgar Publishing, Inc. William Pratt House 9 Dewey Court Northampton Massachusetts 01060 USA

A catalogue record for this book is available from the British Library Library of Congress Control Number: 2009938416

ISBN 978 1 84844 522 2

02

Printed and bound by MPG Books Group, UK

Contents List of contributors Acknowledgements 1.

Emerging clusters: a conceptual overview Max-Peter Menzel, Sebastian Henn and Dirk Fornahl

PART I

2.

3.

4.

6.

7.

1

ACCIDENTS, PATH DEPENDENCY AND STRATEGIC ACTION

Jacobian cluster emergence: wider insights from ‘green innovation’ convergence on a Schumpeterian ‘failure’ Philip Cooke Economic policy and its impact on the evolution of clusters and spatial systems exemplified by German TV programme production Ansgar Dorenkamp and Ivo Mossig Bridging ruptures: the re-emergence of the Antwerp diamond district after World War II and the role of strategic action Sebastian Henn and Eric Laureys

PART II

5.

vii ix

17

43

74

INSTITUTIONS AND ENDOGENOUS DYNAMICS

Origins of human capital in clusters: regional, industrial and academic transitions in media clusters in Germany Anne Otto and Dirk Fornahl The co-evolution of ICT, VC and policy in Israel during the 1990s Gil Avnimelech and Morris Teubal Standards as institutions supporting the cluster emergence process: the case of aquaculture in Chile Paola Perez-Aleman

v

99

140

165

vi

Emerging clusters

PART III 8.

9.

10.

11.

The evolution of the banking cluster in Amsterdam, 1850–1993: a survival analysis Ron Boschma and Floris Ledder The role of the university in the genesis and evolution of research-based clusters Donald Patton and Martin Kenney Sources of ‘second generation growth’: spin-off processes in the emerging biochip industries in Jena and Berlin Max-Peter Menzel The emergence and development of the Cambridge ink jet printing industry Elizabeth Garnsey, Erik Stam and Brychan Thomas

PART IV

12.

13.

PATTERNS OF EMERGENCE AND GROWTH

214

239

265

CLUSTER EMERGENCE AND EMERGENCE OF CLUSTER POLITICS

Neither planned nor by chance: how knowledge-intensive clusters emerge Rolf Sternberg Policy transfer and institutional learning: an evolutionary perspective on regional cluster policies in Germany Matthias Kiese

Index

191

295

324

355

Contributors Gil Avnimelech, Faculty of Business Administration, Ono Academic College, Kiryat Ono, Israel Ron Boschma, Urban and Regional Research Centre Utrecht (URU), Utrecht, Netherlands Philip Cooke, Centre for Advanced Studies in Social Sciences, Cardiff School of City and Regional Planning, Cardiff University, Cardiff, United Kingdom, and Adjunct Professor, School of Development Studies, Aalborg University Ansgar Dorenkamp, Department of Geography, Philipps-University Marburg, Marburg, Germany Dirk Fornahl, BAW Institute for Regional Economic Research GmbH, Bremen, Germany Elizabeth Garnsey, Institute for Manufacturing, Centre for Technology Management, University of Cambridge, Cambridge, United Kingdom Sebastian Henn, Institute for Geosciences, Martin Luther University Halle-Wittenberg, Halle (Saale), Germany Martin Kenney, Department of Human Community Development, University of California, Davis, USA Matthias Kiese, Institute for Competitiveness and Communication, School of Business, University of Applied Sciences Northwestern Switzerland, Olten, Switzerland Eric Laureys, Scientific and Technical Information Service (STIS) of the Belgian Federal Science Policy Office, Brussels, Belgium Floris Ledder, Real Estate, Rosmalen, Netherlands Max-Peter Menzel, Institut for Geography, University of Hamburg, Hamburg, Germany Ivo Mossig, Institut for Geography, University of Bremen, Bremen, Germany

vii

viii

Emerging clusters

Anne Otto, Institute for Employment Research (IAB) RhinelandPalatinate-Saarland, Saarbrücken, Germany Donald Patton, Department of Human and Community Development, University of California, Davis, USA Paola Perez-Aleman, Desautels Faculty of Management, McGill University, Montreal, Canada Erik Stam, Utrecht School of Economics, Utrecht University, Utrecht, Netherlands Rolf Sternberg, Institute of Economic and Cultural Geography, Leibniz University of Hanover, Hanover, Germany Morris Teubal, Department of Economics, Hebrew University, Jerusalem, Israel Brychan Thomas, Welsh Enterprise Institute, Business School, University of Glamorgan, Pontypridd, United Kingdom

Acknowledgements This book is the outcome of the workshop ‘Emerging Clusters. Theoretical, Empirical and Political Aspects of the First Stage of Cluster Evolution’, held at the Max Planck Institute of Economics, Jena (Germany) in June 2008. We are indebted to the Max Planck Society for providing us with the opportunity to organize the workshop and for bringing together this group of scientists in order to discuss ideas and produce this book. Furthermore, the editors would like to thank all the authors for contributing their papers and taking part in the refereeing process.

ix

1.

Emerging clusters: a conceptual overview Max-Peter Menzel, Sebastian Henn and Dirk Fornahl

Economic activity is unevenly distributed in space. This fact has often been associated with spatial concentrations of firms in related fields commonly termed as ‘clusters’ (Malmberg and Maskell 2002). As such clusters are considered centers of economic activity and important elements in economic development in general and in regional development in particular (Porter 1998) it is not surprising that there have been many efforts to better understand in detail the benefits of regional clustering and the processes occurring in functioning clusters (Martin and Sunley 2003). While classical explanations refer to Marshall’s (1917) ideas of agglomeration externalities like a common regional labor pool, specialized suppliers, a shared infrastructure and knowledge spillovers (see also Gordon and McCann 2000), some additional factors beneficial for geographically concentrated firms have been identified in the past two decades (Armington and Acs 2002). Besides others they include the access to networks (OwenSmith and Powell 2004), to a local science base (Zucker et al. 1998) and/ or to local knowledge in general (Malmberg and Maskell 2006), but also ‘buzz’ in the sense of a diffuse and pervasive sharing of information (Bathelt et al. 2004), the co-ordination of complex tasks (Torre and Rallet 2005), local competition (Porter 1998), supportive institutions (Kenney and von Burg 1999) and the characteristics of regional cultures (Saxenian 1994). Though the strong research focus on the functionality of clusters has without doubt resulted in a profound knowledge about the processes occurring within regional clusters, it has largely involved a disregard of the questions how spatial concentrations actually come into being and how they gradually develop (Lorenzen 2005; Frenken and Boschma 2007). Knowing more about the drivers behind cluster formation and change, however, is not only of academic interest but rather a necessary precondition for plausible policy advice aiming at influencing the established

1

2

Emerging clusters

patterns of economic activity, for example by generating new growth centers in peripheral regions.

1

SOME FACTS ON THE EMERGENCE OF CLUSTERS

It is only recently that several contributions have addressed the existing research gap on cluster emergence (see for example Braunerhjelm and Feldman 2006). A short review of the most influential results of these studies will be given in the following. To start with, a differentiation has to be made between the features leading to the genesis of a cluster and those guaranteeing its functioning as is suggested by the studies by Orsenigo (2001) and Breshnahan et al. (2001). While a functioning cluster is marked by a critical mass of companies necessary for maintaining its endogenous dynamics, we use the notion ‘emergence’ when referring to the first stage of cluster development, that is the evolution of firms and institutions until their number reaches this critical mass. Emergence thus describes a broad continuum ranging from ‘normal’ economic activities to the stage of firms becoming ‘geographic concentrations of interconnected companies and institutions in a particular field’ (Porter 1998, 78). As the evolution of clusters must be considered a complex process it is neither possible to reduce their formation to only one single process nor to identify and analyse all factors which might be of importance in that context. The following survey therefore inevitably only covers those parts of the current discussion which we believe to be the most important research areas: the link between new clusters and established regional paths, the generation of institutions and endogenous dynamics as well as the patterns of emergence of successful clusters. 1.1

Accidents, Path Dependency and Strategic Action

Several concepts like the ‘window of locational opportunity’ approach (Scott 1988; Storper and Walker 1989), the core–periphery model by Krugman (1991) or the stochastic approaches based on the models by Arthur (1994) explain the emergence and growth of clusters as two distinct processes. The common rationale of these approaches is simply put as follows: the emergence of a cluster can be traced to ‘some seemingly historical accident’ (Krugman 1991, 35) leading to a number of new firms at a certain location. Once a certain threshold has been reached, agglomeration economies in the form of a regional labor-market

Emerging clusters: a conceptual overview

3

pooling, specialized suppliers and knowledge spillovers will occur (Arthur 1994). In general, these approaches suggest not to regard clusters as a logical outcome of a purposeful process but rather as a result of industrial dynamics which as such is difficult to assess. Apart from the so-called generic factors like a strong science base, an entrepreneurial culture and supportive institutions, the requirements of the future industries are unknown a priori. From this it follows that it is difficult to predict where a new industry or a new cluster, respectively, will start to form (Storper and Walker 1989). The explanatory value of these rather general approaches for understanding the emergence of single clusters has been discussed quite intensely. With special consideration of the core–periphery model, Pinch and Henry (1999, 819), criticize their ‘spot the accident’ logics. According to them, ‘accidents’ could be explained when considering the specifics of the given regional context. Martin and Sunley’s (2006, 418) critique points in the same direction when they argue that events would lose their ‘accidental character’ by thoroughly controlling the specific spatial and temporal context in which they occurred. In addition, they stress that ‘accidental events’ can be both provoked and exploited by strategic actions. According to this line of argumentation, ‘accidents’ simply make up events outside the explanatory range of any approach. Then, however, the question still remains how and why certain events trigger the emergence of a cluster in one region but not in another. When looking into the factors exerting influence on cluster formation, usually firm formations stand out as the most essential components (Feldman et al. 2005; Menzel and Fornahl 2005). This, however, should not hide the fact that firm formations themselves are affected by certain processes which also have been addressed in literature: Bathelt and Boggs (2003) for example highlight the rebundling of different technological trajectories by local actors as one major factor for the emergence of the media cluster in Leipzig. Referring to the example of the British motorcar sports industry in Oxfordshire, Pinch and Henry (1999) in contrast illustrate how new clusters are based upon technologies which are deeply embedded in the regional context. Owen-Smith and Powell (2006) again show that although network structures might differ during the emergence of the biotechnology clusters analysed, they become more similar during their development. All in all, the studies suggest that even though successful clusters might resemble each other, their origin still depends on a particular local context (Feldman et al. 2005).

4

1.2

Emerging clusters

Institutions and Endogenous Dynamics

When a new industry comes into being many regions will usually host the new firms while clusters will emerge only in some of them (Romanelli and Feldman 2006; Klepper 2007). Whatever is the logic behind the initial formations of companies, firms have to exhibit growth rates above average, at least for a certain time, in order to contribute to the formation of a cluster (Jacobs 1969). Therefore, the second field, which will be outlined here, deals with the factors and processes leading from single firm formations to aggregated cluster dynamics, that is the emergence of the endogenous cluster dynamics (Scott 2006). There are several examples showing that clusters did not emerge in those regions which exhibited most firms right from the start (Romanelli and Feldman 2006, Scott 2006, Klepper 2007). Rather, the dynamics which resulted in the formation of clusters started earlier or more strongly in the smaller concentrations. This suggests that the genesis of cluster dynamics does not only depend on the number and size of the firms as well as on generic assets but also on the local capability to form specific assets, even at a very early stage. However, assets such as specialized institutions and infrastructures known from successful clusters like Silicon Valley often form after the cluster in question has already emerged (see for example Longhi 1999 and Feldman 2001). The same holds true for a local labor pool. Following the ‘window of locational opportunity’ approach (Storper and Walker 1989), a specialized pool of labor does not exist due to the newness of an emerging industry. Rather, the growth of firms in the industry is said to attract people from other places. In contrast, Zucker et al. (1998), however, show that new firms in the biotechnology industry have emerged mostly at places where star scientists were located thereby implying that high-technology industries locate at those places, where a labor pool already had been established due to the research programs and postgraduate education in scientific organizations. Though the establishment of regional institutions and a local labor force must be regarded as an important part of cluster emergence, it obviously is not the only element responsible for the dynamic processes. Also associated with cluster emergence is the localization of knowledge generation. Empirical studies show that both knowledge exchange and collaborations are highly localized at the beginning of an industry life cycle (Audretsch and Feldman 1996) and that those industries form clusters that exhibit a large degree of localized collaborations (Brenner 2005). On the whole, these results point out that a connection between the localization of knowledge generation and the collective establishment of cluster-specific institutions seems to be responsible for successful cluster emergence.

Emerging clusters: a conceptual overview

1.3

5

Patterns of Emergence and Growth

A third, and probably the most recent, strand of literature has developed focusing on the patterns of emergence of clusters and industries. This literature indicates that the successful emergence of a cluster is often marked by a particular kind of firm formation, namely spin-offs (Romanelli and Feldman 2006; Klepper 2007). Just to give some examples for this: Klepper (2007) describes how the automobile cluster in Detroit emerged as a result of spin-off activities which can be traced to three incubators: Olds, Cadillac and Buick. Additionally, other types of formations in the cluster did not perform better than comparable firms in other regions and thus did not seem to benefit from agglomeration economies. Similar results were found for the tire cluster in Akron by Bünstorf and Klepper (2009). Romanelli and Feldman (2006) again identify that the largest clusters in the bio-therapeutics industry in the US formed not only where new firms were founded by entrepreneurs from the local science base, but also by the already established firms. They refer to the process by which early firm formations generate spin-offs as to ‘second-generation growth’ and argue that this would be an essential factor for the emergence of clusters. The obvious importance of spin-offs for the formation of clusters has been explained by Klepper (2007) with a heredity theory stating that firms with better routines grow stronger and generate more and better spin-offs. As spin-offs tend to locate close to their incubators, large concentrations may emerge without any necessity of positive externalities between the firms in question. The observation that spin-offs usually perform above average led to the assumption that even if there are some kinds of positive regional externalities, they might be concentrated on only some spinoffs. This argument of course challenges the commonly accepted view on agglomeration economies as a pervasive force which reinforces regional concentrations (Harrison et al. 1996; Pinch and Henry 1999; Martin and Sunley 2006).

2

ORGANIZATION OF THE EDITED VOLUME

Even though some valuable insights have recently been gained about how clusters actually emerge, the novelty of this aspect still implies that answering one question leads to several new ones. Therefore, this edited volume does not claim to exhaust the entire research field. The aim of this book is rather to elaborate and structure the questions which need to be asked, to give some indications as to how they could be answered, and to shed some light on possible directions for future research. In doing so, it tries to cover

6

Emerging clusters

some glimpses of the entire bandwidth of research on cluster emergence by gathering contributions from different research fields dealing with different aspects of cluster emergence by addressing different objects of analysis and different theoretical approaches. In detail, the book is structured around three core fields of research complemented by a policy section. The first part deals with the question how the seed for the cluster is planted and how the cluster is linked to established regional paths. It discusses cluster emergence as a result of regional learning, accidental events and strategic action. The second part deals with the question how the positive externalities and cluster-specific institutions which feed endogenous cluster dynamics come into being. The third part moves to the firm level and investigates the growth patterns during cluster emergence. Finally, the fourth part concludes with the question how policy can influence these processes. The first part of the book, Accidents, Path Dependency and Strategic Action, contains three chapters. In Chapter 2, Philip Cooke presents a new theoretical approach on cluster emergence based upon two fundamental innovation processes. The first one is ‘railroadization’ which is Schumpeter’s fourth form of innovation and describes how externalized organizational innovations like infrastructural investments in railroads, freeways, universities or wireless broadband open up agricultural lands allowing early clusters to form in proximity to key-network nodes or ‘hubs’ of these investments. The second innovation process is based upon railroadization. Where the regional context gives opportunities for Schumpeterian ‘new combinations’ from regionalized ‘related variety’, Jacobian clusters emerge through technological convergence. Cooke exemplifies his approach by studying the green technology clusters in California, North Jutland, Wales and Norway. In Chapter 3, Ansgar Dorenkamp and Ivo Mossig analyse the emergence of regional clusters in the broadcasting industry in Germany. They show that the current locational pattern is the outcome of past decisions and accidental events. For a considerable time, broadcasters under public law acted as monopolistic market pioneers. In 1984, when private companies were allowed to enter the market, their choice of location was primarily influenced by both public granting policies and the necessity to make use of qualified personnel. As a consequence, the locations of the public broadcasters were strengthened and most of them provided seedbeds for further clustering. Using the example of the Antwerp diamond cluster, Sebastian Henn and Eric Laureys consider the role of strategic action during the emergence of a cluster in Chapter 4. Antwerp, which had served as the center of diamond trading and polishing for hundreds of years, re-emerged after

Emerging clusters: a conceptual overview

7

World War II even though the industry-related activities and infrastructure to a large extent had been discontinued, respectively destroyed. In contrast to the literature which regards chance as the major determinant of cluster emergence, Henn and Laureys argue that deliberate actions to establish Antwerp as the leading center again were the main drivers for its resurrection. Each of the chapters elaborates different influences on the location of clusters: infrastructure building and regional innovations, accidents and policy decisions as well as strategic action. While general patterns of cluster emergence are observable, the results indicate that the reasons for the exact locations often are idiosyncratic. Part II, Institutions and Endogenous Dynamics, consists of three chapters focussing on both the externalities which are typical for successful clusters and on how they are created. In Chapter 5, Anne Otto and Dirk Fornahl deal with the interrelation between the development of clusters and labor markets. Although it is common sense that successful clusters rely on a strong local human capital base, there is little evidence on the emergence of such local labor market pools. By analysing the mobility in emerging and growing clusters in Germany’s audiovisual media industry, the authors find that especially the emerging stage has been characterized by intense inflows of labor, while growing clusters usually have established a local labor market and thus are less dependent on labor inflows. In Chapter 6, Gil Avnimelech and Morris Teubal investigate the coevolution between VC and cluster emergence by studying the example of the ICT cluster in Israel during the 1990s. Placing a special focus on the transition from the pre-emergence stage to the emergence stage they identify that venture capital exerts a twofold influence on the emergence of clusters: a direct effect by their portfolio and an indirect effect as they stimulate start-up creation and act as a focusing device or gatekeeper to global networks. In turn, the growth of firms and the cluster attracted more VC. They conclude with a discussion of the interactive relationship between cluster emergence and the development of a VC infrastructure. Finally, in Chapter 7 Paola Perez-Aleman uses the example of the Chilean salmon industry to highlight the importance of institutions as a crucial factor for the growth of firms in an emerging cluster. Being latecomers to the industry, the Chilean companies started setting up different standards like product classifications, quality-control mechanisms and regulations to reduce the mortality of fishes. During the process of standardization, the firms gradually improved their knowledge and products. As a consequence, the once small-scale industry became the world’s leading producer of salmon. All three chapters strongly indicate that there is a difference between the

8

Emerging clusters

emergence of clusters and their functioning, as the growth of the clusters is accompanied by the establishment of particular institutions like labor markets, venture capital, or common standards. The contributions do not only confirm previous studies (Bresnahan et al. 2001; Orsenigo 2001) but also amend them by showing that institutions neither precede nor follow cluster emergence, but co-evolve with it. Whether the emerging cluster becomes a growing and functioning cluster therefore highly depends on its capability to develop a distinct institutional setting. The chapters in Part III of the book, Patterns of Emergence and Growth, investigate the patterns of firm formation which are symptomatic for a successful emergence of clusters, namely spin-off processes. In Chapter 8, Ron Boschma and Floris Ledder link the window of locational opportunity concept to the approaches on industrial dynamics developed by Klepper (for example Klepper 2007) to describe why Amsterdam became the leading banking cluster of the Netherlands. In detail, their analysis is based on a unique database of all entries and exits in the banking sector in the Netherlands during the period 1850–1993. The authors examine the extent to which spin-off dynamics, the time of entry and the location of Amsterdam had a significant effect on the survival rate of banks during the last 150 years. They find that a window of locational opportunity was open until the first half of the 19th century and that Amsterdam’s leading role in the Dutch banking system is rather due to spin-off dynamics than to regional externalities. In Chapter 9, Donald Patton and Martin Kenney take a closer look at clusters in Illinois and Wisconsin that have evolved around large elite universities. One major finding of their comparison is that the fields of activity of those university spin-offs that are at the heart of their clusters reflect the relative strengths of their incubators. Another important result is that both clusters exhibit totally different structures: while in Illinois only a structural core of a cluster could develop so far, the cluster in Wisconsin has entered a second stage of cluster formation which has been characterized by spin-offs from existing firms. The authors attribute their findings mainly to differences in the entrepreneurial environments in the two locations. In Chapter 10, Max-Peter Menzel investigates the sources of ‘second generation growth’ using the biochip industries in Jena and Berlin as examples. The industries at both locations exhibit a growth pattern as predicted by Klepper’s (2007) heredity theory as only a few firms generated spin-offs. However, the knowledge embodied in the spin-offs had not been created in the parent firms but rather by interacting with other regional actors. From this observation Menzel concludes that it is not only birth and heredity but also the regional context which affects the propensity of second generation growth.

Emerging clusters: a conceptual overview

9

Elizabeth Garnsey, Erik Stam and Brychan Thomas describe the genealogical and ecological processes which have led to the emergence of the inkjet-printing cluster in Cambridge in Chapter 11. The Cambridge inkjet-printing cluster is marked by a substantial number of large firms, it has no direct university lineage, and it is more demand-driven than most other high-tech industries in the region. Spin-off processes were mainly responsible for the emergence of the cluster while the local ecology evolved gradually via the labor market. However, the recent renewal of the cluster was accompanied by a reconnection to the university. The four chapters of this part confirm existing studies which highlight the importance of spin-off processes for the emergence of clusters. Yet, each chapter gives different reasons for the spin-off phenomenon, which in a way reflects differences in the chosen industry, region and historical era. While the analysis of Boschma and Ledders confirms the Klepper literature by stating that the central process behind spin-offs are birth and heredity, Patton and Patton argue that a distinct entrepreneurial milieu is responsible for spin-off processes, and Menzel traces the origin of spin-offs to regional learning processes. Additionally, Garnsey, Stam, and Thomas show that spin-offs were responsible for the emergence of a cluster, but regional collaborations for its renewal. Part IV, Cluster Emergence and Emergence of Cluster Politics, changes the perspective from theoretical explanations and empirical investigations of cluster emergence to the question if and how policy support influences the emergence of clusters. In Chapter 12, Rolf Sternberg studies the impact of policy on cluster emergence by comparing ten high-technology regions in the USA, UK, Japan, France and Germany. His analytical framework contains factors derived from different approaches assumed to be relevant for cluster emergence. According to him, a variety of factors has to come together for making clusters emerge: in this context, the nation’s and/or a state’s technology policy activities together with the R&D infrastructure could be identified as the most important factors in the emergence of most clusters. However, most of the clusters analysed were not an intended outcome of explicit government policies, and non-governmental factors seem to have played a role as well. Matthias Kiese concentrates on the diffusion of cluster policies in Chapter 13. He conceptualizes the diffusion of cluster policies as repeated acts of policy transfer taking multiple channels such as manuals, consultants, knowledge communities and policy tourism. His interviews with practitioners in ministries and economic development agencies, cluster managers and consultants, as well as independent observers in Germany reveal that policy-makers and practitioners mainly learn from knowledge incrementally accumulated through learning-by-doing and less through

10

Emerging clusters

interregional policy transfer, which is mainly used as an inspiration. Hence, external ideas are taken up but adapted to fit to the specific regional context. The findings of the chapters in this part clearly indicate that dedicated cluster policies are only one, yet an important element in an array of factors responsible for cluster emergence. Obviously, there is not a best practice, and policies to support clusters have to be adapted to local conditions and the contemporary socio-economic context.

3

PATHWAYS FOR FUTURE RESEARCH

The contributions of this edited volume aimed at structuring the research questions and indicate directions for further research. One core question still to be answered is how new clusters are connected to established regional paths and assets. The bandwidth of the existent literature ranges from path dependency to windows of locational opportunities and accidental events. There are empirical studies for each explanation. While the contributions of this edited volume indicate that there are ex-post explanations which go beyond random events, they also point to the fact that there is no single explanation that accounts for all clusters. Why a particular path is chosen and which processes influence if, how and why a new cluster adheres to established development paths are questions still requiring further analysis. Another point deserving further consideration is how the endogenous dynamics of clusters come into being. The contributions of this book indicate that these dynamics are connected to the establishment of specific institutions. How and under which circumstances these institutions are created, which factors trigger this co-evolution and under which conditions this co-evolution takes place, are other questions for further research. A further promising road for research is the investigation of growth patterns. While the contributions of this book confirm existent studies emphasizing the importance of spin-offs for cluster emergence, they give different explanations for the drivers behind the spin-off processes: inheritance of firm routines, regional learning processes, or an entrepreneurial milieu. These different results point to the difficulties of empirically grasping the process of cluster emergence. Research on cluster emergence often is made ex-post, meaning that the unit of analysis lies in the past. Typically, approaches to gather the required information rely on historical descriptions of the sequences of events or factors that lead to the emergence of the cluster (for example in Pinch and Henry 1999), interviews with important

Emerging clusters: a conceptual overview

11

actors (for example in Bresnahan et al. 2001) or survival analyzes of firms (for example in Klepper 2007). Yet, the dependence on limited information aggravates the analysis of the processes which actually have taken place. Another question also important for developing policy measures is whether the processes identified as being responsible for cluster emergence in the past are also of importance under the contemporary socio-economic regime. Due to these methodological difficulties, we encourage the application of a variety of methods to find out different patterns and processes which resulted in or prevented cluster emergence in different times and socio-economic contexts. The many open and wide questions which could be only partly outlined above and also the many different methodological and theoretical approaches must be considered as evidence for the novelty of the field. Further research, especially on the linkage between new clusters and established paths as well as on the generation of institutions and endogenous dynamics, clearly would be helpful to develop appropriate policy measures to adequately support cluster emergence. The many failed attempts to artificially create clusters highlight this necessity quite plainly. Against this background we explicitly encourage any further discussion, criticism and comments and go in accord with Braunerhjelm and Feldman (2006, 3) when stating that ‘there is simply too much at stake to be uncritical when our focus is long-term economic growth and when significant public funds are used to promote cluster formation activities’.

REFERENCES Armington, C. and Z.J. Acs (2002), ‘The Determinants of Regional Variation in New Firm Formation’, Regional Studies, 36 (1), 33–45. Arthur, W.B. (1994), Increasing Returns and Path Dependence in the Economy, Ann Arbor, MI: University of Michigan Press. Audretsch, D.B. and M.P. Feldman (1996), ‘Innovative Clusters and the Industry Life Cycle’, Review of Industrial Organization, 11 (2), 253–73. Bathelt, H. and J.S. Boggs (2003), ‘Towards a Reconceptualization of Regional Development Paths: Is Leipzig’s Media Cluster a Continuation of or a Rupture with the Past?’, Economic Geography, 79 (3), 265–93. Bathelt, H., A. Malmberg and P. Maskell (2004), ‘Clusters and Knowledge: Local Buzz, Global Pipelines and the Process of Knowledge Creation’, Progress in Human Geography, 28 (1), 31–56. Braunerhjelm, P. and M. Feldman (2006), Cluster Genesis. Technology-Based Industrial Development, Oxford, UK: Oxford University Press. Brenner, T. (2005), ‘Innovation and Cooperation during the Emergence of Local Industrial Clusters: An Empirical Study in Germany’, European Planning Studies, 13 (6), 921–38. Bresnahan, T., A. Gambardella and A. Saxenian (2001), ‘“Old Economy” Inputs

12

Emerging clusters

for “New Economy” Outcomes: Cluster Formation in the New Silicon Valleys’, Industrial and Corporate Change, 10 (4), 835–60. Bünstorf, G. and S. Klepper (2009), ‘Heritage and Agglomeration: The Akron Tire Cluster Revisited’, Economic Journal, 119 (April), 705–33. Feldman, M.P. (2001), ‘The Entrepreneurial Event Revisited: Firm Formation in a Regional Context’, Industrial and Corporate Change, 10(4), 861–91. Feldman, M.P., J. Francis and J. Bercovitz (2005), ‘Creating a Cluster while Building a Firm: Entrepreneurs and the Formation of Industrial Clusters’, Regional Studies, 39 (1), 129–41. Frenken, K. and R.A. Boschma (2007), ‘A Theoretical Framework for Evolutionary Economic Geography: Industrial Dynamics and Urban Growth as a Branching Process’, Journal of Economic Geography, 7 (5), 635–49. Gordon, I.R. and P. McCann (2000), ‘Industrial Clusters: Complexes, Agglomeration and/or Social Networks’, Urban Studies, 37 (3), 513–32. Jacobs, J. (1969), The Economy of Cities, New York, US: Vintage Books. Harrison, B., M.R. Kelley and J. Gant (1996), ‘Innovative Firm Behavior and Local Milieu: Exploring the Intersection of Agglomeration, Firm Effects, and Technological Change’, Economic Geography, 72 (3), 233–58. Kenney, M. and U. von Burg (1999), ‘Technology, Entrepreneurship and Path Dependence: Industrial Clustering in Silicon Valley and Route 128’, Industrial and Corporate Change, 8 (1), 67–103. Klepper, S. (2007), ‘Disagreements, Spinoffs, and the Evolution of Detroit as the Capital of the US Automobile Industry’, Management Science, 54 (4), 616–31. Krugman, P. (1991), Geography and Trade, Cambridge, MA: MIT Press. Longhi, C. (1999), ‘Networks, Collective Learning and Technology Development in Innovative High Technology Regions: The Case of Sophia Antipolis’, Regional Studies, 33 (4), 333–42. Lorenzen, M. (2005), ‘Why do Clusters Change?’, European Urban and Regional Studies, 12 (3), 203–8. Malmberg, A. and Maskell, P. (2002), ‘The Elusive Concept of Localization Economies: Towards a Knowledge-based Theory of Spatial Clustering’, Environment and Planning A, 34 (3), 429–49. Malmberg, A. and Maskell, P. (2006), ‘Localized Learning Revisited’, Growth and Change, 37 (1), 1–18. Marshall, A. (1917), Industry and Trade, London: Macmillan. Martin, R. and P. Sunley (2003), ‘Deconstructing Clusters: Chaotic Concept or Policy Panacea?’, Journal of Economic Geography, 3 (1), 5–35. Martin, R. and P. Sunley (2006), ‘Path Dependence and Regional Economic Evolution’, Journal of Economic Geography, 6 (4), 395–437. Menzel, M.-P. and D. Fornahl (2005), ‘Unternehmensgründungen und regionale Cluster. Ein Stufenmodell mit quantitativen, qualitativen und systemischen Faktoren’, Zeitschrift für Wirtschaftsgeographie, 39 (3–4), 131–49. Orsenigo, L. (2001), ‘The (Failed) Development of a Biotechnology Cluster: The Case of Lombardy’, Small Business Economics, 17 (1–2), 77–92. Owen-Smith, J. and W.W. Powell (2004), ‘Knowledge Networks as Channels and Conduits: The Effects of Spillovers in the Boston Biotechnology Community’, Organization Science, 15 (1), 5–21. Owen-Smith, J. and W.W. Powell (2006), ‘Accounting for Emergence and Novelty in Boston and Bay Area Biotechnology’, in P. Braunerhjelm and M.P. Feldman

Emerging clusters: a conceptual overview

13

(eds), Cluster Genesis. Technology-Based Industrial Development, New York: Oxford University Press, pp. 61–83. Pinch, S. and N. Henry (1999), ‘Paul Krugman’s Geographical Economics, Industrial Clustering and the British Motor Sport Industry’, Regional Studies, 33 (9), 815–27. Porter, M.E. (1998), ‘Clusters and the New Economics of Competition’, Harvard Business Review, 76 (6), 77–90. Romanelli, E. and M. Feldman (2006), ‘Anatomy of Cluster Development: Emergence and Convergence in the US Human Biotherapeutics, 1976–2003’, in P. Braunerhjelm and M. Feldman (eds), Cluster Genesis: Technology-Based Industrial Development, Oxford, UK: Oxford University Press, pp. 87–112. Saxenian, A. (1994), Regional Advantage: Culture and Competition in Silicon Valley and Route 128, Cambridge, MA: Harvard University Press. Scott, A.J. (1988), ‘Flexible Production Systems and Regional Development: The Rise of New Industrial Spaces in North America and Western Europe’, International Journal of Urban and Regional Research, 12 (2), 171–85. Scott, A. (2006), ‘Origins and Growth of the Hollywood Motion-Picture Industry: The First Three Decades’, in P. Braunerhjelm and M. Feldman (eds), Cluster Genesis: Technology-Based Industrial Development, Oxford, UK: Oxford University Press, pp. 17–37. Storper, M. and R. Walker (1989), The Capitalist Imperative. Territory, Technology, and Industrial Growth, New York, US; Oxford, UK: Basil Blackwell. Torre, A. and A. Rallet (2005), ‘Proximity and Location’, Regional Studies, 39 (1), 47–59. Zucker, L., M. Darby and M. Brewer (1998), ‘Intellectual Capital and the Birth of U.S. Biotechnology Enterprises’, American Economic Review, 88 (1), 290–306.

PART I

Accidents, path dependency and strategic action

2.

Jacobian cluster emergence: wider insights from ‘green innovation’ convergence on a Schumpeterian ‘failure’ Philip Cooke

1

INTRODUCTION

As is well-known, for those interested in evolutionary economic geography, Schumpeter left almost no regional or spatial analysis of economic phenomena. From the evolutionary economic geography and policy viewpoints, this is clearly disappointing. His two brief allusions are highly time–space-specific. The first concerns Schumpeter’s fourth form of innovation, which is designated regional and exemplified by the process of ‘railroadization’ – the phenomenon by which US and other agricultural lands were opened up to markets by infrastructural investments, not only in railroads but farms, grain silos and even agricultural manuals that the railroad companies of the western USA had printed so that pioneers accessing cheap land on the plains would know how to farm that land. In Jutland, Denmark, 350 community schools were established to perform some of these and other functions. This ‘regional evolution’ of land and markets was, rightly, considered an externalized organizational innovation as compared with the internalized organizational innovation of a corporation adopting new management methods that gave it a, albeit temporary, competitive edge (Schumpeter, 1975; Andersen, 2007). The second allusion is even briefer where Schumpeter mentions innovation such as the department store only being feasible in the large city due to the level of demand required to sustain such an innovation. Hence the city is seen as having some economic specificity from its scale attributes, but Schumpeter says nothing about the dynamics of the entailed processes (Andersen, 1994, 2007). In this chapter it is argued that the rather ugly word ‘railroadization’ captures a crucial aspect of economic development from which generalizations

17

18

Emerging clusters

may be hypothesized. It is clearly a developmental ‘trigger’, by no means the only one, and it may become residual – even archaeological eventually. Nevertheless, railroads were emulated by other infrastructures with huge productivity contributions, notably the freeways in the US, regionalization of universities in many countries and, most recently, wireless broadband. Illustration of this pre-figurative theory of regional growth is offered by reference to recently identified processes of Jacobian cluster formation and emergence. Thus, two giants of economic evolution are put in harness for this task. The Jacobian dimension arises from a frequently observable cluster emergence consequent upon railroadization. What happens is that early clusters form in proximity to key network nodes or ‘hubs’. Every region that this author has explored, beginning with California, displays this emergent cluster formation, followed by cluster mutation, pathway. This is probably a chance effect but, as we shall see, it is nevertheless curious that Israel, Denmark and Wales (UK) all display this kind of cluster emergence process. Unfortunately, space does not permit the study of the fascinating case of Israel, whose ‘railroadization’ to all cities was only completed in 2007, when Be’er Sheva was connected. Almost immediately a green technology cluster, mainly focused on solar power and water conservation innovations, was identifiable, when it had not been before. The fact that it is associated with Israel’s Dimona nuclear facility Science Park may be a contributory factor to the former quietude amongst firms, but the new railway connection to nearby Be’er Sheva and its heightened academic interaction with, in particular, Tel Aviv, led to venture capital interest in forming a financial interest by Ben Gurion University of the Negev in Israel Cleantech Ventures, Israel’s first cleantech focused venture capital fund. Firms began to be put in better-networked interactive situations by the process of railroadization. Importantly, this also settles the dry and often unproductive debate about scale in cluster definition. Notions of ‘critical mass’ miss the key point of differentiation between a cluster and, say, an agglomeration, which is that firms in clusters interact, discuss, compete, co-operate and generally act as a group. Thus, as long as there are more than two interacting in some or other of these ways, including with any third parties like incubators or venture capitalists (two are a ‘couple’ and not a ‘group’), they constitute a (conceivably emergent) cluster. Otherwise, we get stuck in Zeno’s paradox of the bees – namely how many comprise a ‘heap’ or a ‘swarm’ in Schumpeter’s language. Our answer is three, where they are of one intention, namely seeking to create a new hive, usually because of overcrowding in the old one (Gould, 1992). Finally, the ‘discovery’ of cluster mutation was a by-product of new research into ‘green innovation’ (Cooke, 2008a). That is, in studying how clean technologies,

Jacobian cluster emergence

19

renewable energies and waste recycling innovations evolved it could easily be shown that entrepreneurs moved into ‘green technology’ from a nearby ICT, biotechnology, agro-food, oil-refining or nanotechnology type of cluster. It is, on reflection, an open research question whether such entrepreneurial migrants seek, like the bees, to escape overcrowding in their previous cluster. In what follows, the chapter first explores the meaning and implications of the theoretical perspective advanced. This involves attempting process definitions of railroadization, Jacobian cluster mutation, platform, and the relationship between the Schumpeterian and Jacobian elements of the theoretical framework. Thereafter, four empirical sub-sections ensue. These explore historical exemplars of the key processes under discussion. As noted, three of these involve cluster mutation from inter alia agro-food, agricultural and marine engineering, wireless telephony, information technology, biotechnology and energy clusters. To stress-test the underlying thesis, a non-cluster case is invoked from Norway, where large firm engagement in clean technology is pronounced. Interestingly, much of the intra-corporate production of silicon and other inputs has a high degree of geographical proximity of an intra-corporate kind. This, if in future accompanied by many more empirical illustrations, is suggestive of the importance of geographic proximity dominating even cognitive, intracorporate proximities, which are supposed to be placeless.

2

RELATING RAILROADS AND RELATED VARIETY

To continue, it is suggested that bemoaning Schumpeterian neglect of the spatial dimension may be misplaced. His category of innovation by railroadization helps in understanding regional innovation in which clusters ‘mutate’ through a Jacobian (after Jane Jacobs, 1969) process. This does not mimic the bees, dividing to escape, but rather indicates taking the opportunity to utilize the ‘related variety’ operating at the regional level in places like California, North Jutland (Denmark) and Wales (UK). Thus regional innovation through cluster mutation is illuminated by the interest of this chapter in ‘green’ innovation, which is pronounced in those regions. These are the only regions to have been examined from a ‘green innovation’ perspective thus far to this author’s knowledge.1 A possible reason for the illuminative aspect of taking a green perspective is that green innovation (such as the burgeoning cleantech industry) displays a high degree of innovation convergence across fields like information and communication technology (ICT), nanotechnology, biotechnology, agro-food, health,

20

Emerging clusters

environment, energy, production and materials management, and waste treatment. Thus innovation occurs laterally among distinctive parts of what may be described as an innovation platform. Other regions for which Jacobian clustering is probably true are those of the Third Italy, which has been studied from this perspective by Boschma and Lambooy (2002) also from an evolutionary economic geography viewpoint. They found that apparently different industrial districts displayed ‘related variety’ in their engineering competences and associated high lateral absorptive capacity towards innovations emanating from neighbouring industries and clusters. No claim is made here for the ubiquity of this process, on the contrary Jacobian cluster regions are probably not in the majority. But where they exist they can be propulsive in relation to national economies or aspects of them. To that extent they make a contribution to the understanding of regional and national unequal development between wealth and poverty, the issue that has animated economics since Adam Smith. They also constitute a particular type of explanation of cluster emergence, more satisfactory than idealizations such as ‘windows of opportunity’ that seem popular among many cluster emergence researchers. This is because the actual ‘mutation agents’, namely transmigrating entrepreneurs, can be identified, researched and their motivations investigated. It may reasonably be asked, since a new theorization of regional economic development is being proposed for the part of that complex field of study touched by the fascinating concept of ‘cluster emergence’, in what precise relationship do Jacobs and Schumpeter lie, intellectually speaking? It should be clear by now that the conceptual relationship tends to be ‘linear-interactive’. This means that the ‘opening-up’ of regional economic opportunity (Schumpeterian) is logically prior. Thereafter, for reasons discussed in detail in the empirical sections that follow, a cluster may emerge and once one has emerged it may divide or mutate into a new one or more over time. This species differentiation is, of course, predictable from an evolutionary as well as an evolutionary economic geographic perspective. Where numerous clusters co-exist in related variety and geographical proximity to each other we may speak of their form as that of a ‘platform’ of interrelated industries rather than the single clusters that may have preceded the platform historically. This theoretical explanation is ostensibly more satisfactory than the routine ‘window of opportunity’ account of cluster emergence. This is for three reasons. First, a possible hangover from evolutionary biology is the notion of ‘chance’ pronounced in the ‘window’ metaphor. For in human society, as Pasteur put it, ‘fortune favours the prepared mind’. Second, there seems no evident ‘agency’ as a result of that approach’s reliance upon a chance explanation, again disturbing from a human action perspective. Finally, there seems no possibility of ‘learning’

Jacobian cluster emergence

21

let alone ‘moulding’ the cluster outcome through social interaction or networking since, almost by definition, the particular chance opportunity is unrepeatable. Importantly, the Jacobian focus on ‘the city’ is relaxed completely in this analysis, which is interested more in regions. Jacobs’ analyses do not seem weakened by this scalar re-focusing. Having held out the promise of a neo-Schumpeterian theory of regional evolution, that aspiration has to be severely qualified. For a more truly evolutionary theory of spatial dynamics we have to turn to the midtwentieth century inheritors of Veblen’s concept of ‘cumulative causation’. A variant of the biblical ‘Matthew principle’ of ‘to those that have, more shall be given’ this profoundly disequilibrium perspective contains the missing dynamic element by virtue of Myrdal’s (1957) elaboration upon the various ‘backwash’ and ‘spread’ effects associated with regional evolution. Spread effects, on occasions, caused the dynamic element to seek to accommodate growth beyond its original boundaries. Backwash effects sucked back temporary gains made by competing locations to the larger, predominating accumulating entity, such as a strong city or regional economy. Observations of static relationships in the spatial evolution of the ‘knowledge economy’ have led to the preliminary postulation of a knowledge capabilities theory of regional evolution based on the distinctive distribution of two key components of the knowledge economy labour market (Cooke, 2007). Foremost here are, first, the knowledge-intensive business services (KIBS) such as finance, research, media, software and so on, while second, high technology manufacturing is a mainstay of the knowledge economy in computer and communications hardware, aerospace and biotechnology inter alia. Empirical observation of the static picture for the EU strongly suggested an urban–regional split between the locations of these. The former predominate particularly in primate cities (such as the major financial centre, sometimes, but not always, combining capital city administrative functions); the latter predominate in specialized satellite towns, often with appropriate knowledge centres like national research institutes or universities centred in them. This theory, in brief, is consistent with Myrdal–Hirschman2 theses about ‘cumulative causation’ and metropolitan regional concentrations of knowledge economy activities (Cooke, 2002). But as noted, the static picture merely hinted at the dynamism explicit in the idea of cumulative causation, which remained to be tested. The first such contemporary test was accomplished by access to and analysis of special runs of Israeli data in a dynamic perspective (Cooke and Schwartz, 2008). This chapter relies on those findings and then explores innovation outside primate cities where knowledge-intensive business services (KIBS) thrive to explore innovation at some distance from big cities altogether. Nevertheless Myrdal–Hirschman models

22

Emerging clusters

postulate innovation as capable of occurring there because such cities tend not to localize high technology manufacturing (HTM). To that we would add that they do not tend to have functional regional innovation systems and they are often the home of specialized rather than related variety clusters. The next sections take these insights and exemplify this by reference to some ‘cumulative causation’ peculiarities of innovative industry regions. Those selected display innovative convergence among high technology sectors to contribute to cleaner manufacturing, food and energy production – the so-called cleantech sectors, respectively (Cooke, 2008b). We might, writing from another perspective, wish to explore the phenomenon of cluster-emergence of a different kind from that investigated here. That would involve locational shift from possibly high rent urban spaces to satellite locations within reach of KIBS and other services but offering more affordable labour and land rents.

3

REGIONAL INNOVATION SYSTEMS: INTEGRATING REGIONAL NETWORKS AND REGIONAL INNOVATION POLICY

The enthusiasm for studying networks arose in a context of manifest decline in the co-ordinating capabilities of states and markets regarding leading edge research and innovation, which subsequent data (for example Chesbrough, 2003) shows set in from approximately 1991. But if the central state had become as debilitated as many large private corporations were to become regarding the lack of productivity from their large budgetary allocations to research and development (R&D), the ‘regional state’ seemed, from empirical reportage of the kind discussed above, to be on the rise. A parallel strand of research had evolved, which focused on regional innovation policy. Thus the connecting concept of regional innovation systems (RIS) evolved from this even earlier thinking about ‘regional innovation policy’, in relation to ‘regional innovation networks’ (the ‘systems view of planning’ intruding again). This happened in two publications, the more widely-cited one being less theoretically and empirically rich than the almost totally uncited one. The difference between Cooke (1992) and (1993) lies in the absence of any bibliographical influence from the ‘innovation systems’ literature in the 1992 paper, which thus has purer lineaments to economic geography. Contrariwise, the 1993 paper shows the author had by then read both Lundvall’s (1988) contribution on ‘innovation as an interactive process’ and Dosi et al. (1988) and was also influenced by Johansson (1991) and Grabher (1991) in probably the first proper book

Jacobian cluster emergence

23

on regional development from a ‘network regions’ perspective (Bergman, Maier and Tödtling, 1991). It seemed necessary to place these distinctive ‘network and policy’ concepts in relation to each other in a layered model. So, the innovation policy dimension evolved conceptually into the idea of a sub-system supporting with knowledge and resources the innovative firms in their networks. These formed a ‘superstructural’ sub-system dealing with actual innovation ‘near market’. As we have seen, they had been spoken of as carrying out ‘networking’ with each other, not only laterally in alliances or partnerships and vertically in sometimes partly localized supply chains but also with the innovation policy and knowledge generation sub-system (Meyer-Krahmer, 1990; Cooke, Alaez and Etxebarria, 1991; Malecki, 1991; Rothwell and Dodgson 1991). So these also had sub-system characteristics related to the governance of innovation support. Each sub-system was also seen to interact with global, national and other regional innovation actors, and even through technological or sectoral systems of innovation. Open systems ruled. Over the years the RIS framework has been analysed in terms of many different ‘varieties of innovation’ relating to localized, networked and hierarchical innovation ‘governance’ systems. Third Italy, BadenWürttemberg and French innovative regions exemplified each, respectively. Correspondingly, the ‘exploitation’ sub-system of firms, in the main, could be dominated by large firms or oligopolies – even foreign ones as with the Asian transplants to Wales in the 1980s and 1990s. Other regions, like Catalonia, had a mix of large (SEAT) and SME ‘district’-type innovation relations, while other places might have innovation regimes in which only small, entrepreneurial firms predominated, as in places with observable ‘industrial districts’, not only Third Italy but also some newer technology clusters. Later still, these more entrepreneurial SME systems, living by venture capital and exploitation of public research from universities, could be differentiated further as ‘entrepreneurial’ (ERIS), market-led systems, compared with those, especially in Europe, where they were more ‘institutional’ (IRIS) where state support was pronounced and entrepreneurship was less advanced (Cooke, 2004).

4

RECENT ADVANCES IN RIS RESEARCH

One of the most interesting research areas opened up in RIS research in the recent past concerns, once again, the insights of Jane Jacobs (1969) and can be referred to as addressing the challenging issue of ‘cluster emergence’. In particular by examining the emergence of a number of ‘green clusters’ on a

24

Emerging clusters

regional canvas, we see emphasis in ‘green innovation’ upon technological convergence among diverse industries. These include biotechnology, information technology and nanotechnology (but not limited to these hightech activities) and among them we also see a process of cluster ‘species mutation’. Of particular fascination here is that some regions have the capability relatively rapidly to mutate many ‘Jacobian’ clusters – so-called because although different they display evolutionary characteristics of ‘related variety’ (Frenken et al., 2007). A clear definition is called for here to denote this new concept. The key is the evolutionary concept of variety, whereby some new combinations of entrepreneurial and innovative opportunity might present themselves in geographically proximate space. This would arise from the mixture of knowledge spillovers and rather high absorptive capacity among neighbouring economic activities. Hence (Jacobian) variety is both a context and an ‘evolutionary fuel’ for cluster emergence as long as there is not too much cognitive dissonance or distance between neighbouring economic activities. So, as suggested earlier, Jacobian clusters emerge from new combinations of knowledge cross-fertilising among, for example, high-technology activities like biotechnology and information technology or medium ones like agro-food that may be foundations for a new clean technology cluster that adopts and adapts elements from each. Although we noted above that a different kind of cluster emergence might arise from lower rent-seeking practices by firms moving to geographically proximate locations outside cities, it shares with this ‘Jacobian’ model a ‘cumulative causation’ inheritance. However, while the Jacobian model is positively cumulative, the rentseeking model is, initially at least, the product of negative externalities of cumulative causation in the ‘mother city’. An important caveat regarding the meaning of related variety arises from this research. It is normally taken to mean firms in neighbouring NACE categories. However, for example, new combinations from agro-food and automotive industries that are historically not that close in technical let alone NACE terms may also arise if the new combination being sought concerns biodiesel or bioethanol. This is because adjustments in breeding of plants may have to be made if negative effects on engine performance cannot be made by the automotive side of the equation. Hence related variety is not fixed to sectoral relatedness but embodies also particular and contextuated technological convergences. In that respect it is much harder to predict cross-fertilization in the latter than the former case. The solution to this is yet another ‘after the fact’ type analysis where such unexpected interaction can be referred to as ‘revealed related variety’. But anyway, Jacobian variety rests not within but among the clusters according to this line of reasoning. This means the ‘Jacobian’ dimension is both a potentiality and

Jacobian cluster emergence

25

Clean Technology Biotechnology Wireless ICT Agro-Food Wine Film

Figure 2.1

California’s Jacobian clusters

(empirically) a product of its emergence. Moreover, as noted, it is likely to occur in the relative geographical proximity of regions. In what follows, empirical evidence is provided of regional evolution through innovation of differing intensities ‘mutating’ through processes of knowledge search and selection that happen to give rise to successive clustering phenomena in regional ‘platforms’ of related economic variety. Jacobian Clusters One such region is Northern California whose ICT, biotechnology and clean technology clusters overlap in proximity to San Francisco but also near various agro-food clusters like wine in the Napa, Sonoma and Russian River valleys and varieties of horticulture in the San Joaquin and Sacramento river valleys (see Figure 2.1). But notice, Figure 2.1

26

Emerging clusters

also shows Southern California having prominent Jacobian clusters in Los Angeles and San Diego (Cooke, 2008a). The content of Figure 2.1 is drawn entirely from secondary evidence supplied by the numerous studies of clusters in California as published in Saxenian (1994), Porter (1998), Simard and West (2003), Guthman (2005), Scott (2006) and Cooke (2007). North Jutland in Denmark is another such region, as apparently is Wales in the UK, as we shall see. North Jutland’s economy is the global centre of the wind turbine production industry whose profile and evolutionary trajectory was a key beneficiary from the outset of varieties of innovation. As will be shown, this recently ‘discovered’ cluster has all the required characteristics to warrant the cluster designation, conjoining university research at, for example, Ålborg and Århus Universities, the Danish Technological Institute (DTI) also at Århus, and both spin-out firms and larger, indigenously established firms that are involved in ‘green innovation’. Denmark’s ‘cluster’ has no geographical specificity of the kind Porter (1998) was rather more sensitive towards. He referred there to a cluster as: ‘a geographically proximate group of interconnected companies and associated institutions in a particular field, linked by commonalties and complementarities’. With regard to such clusters the most important analytical task is to establish the extent of interconnections, commonalities and complementarities since this is what distinguishes a localized cluster, its specialization or differentiation and its potential for exploiting knowledge spillovers for competitive advantage. In the research to be reported below, the Danish Wind Energy Association database was accessed and the details mapped by location and categorized according to point in the supply chain. Thus final assemblers were differentiated from major module suppliers (for example fibreglass blade manufacturers) and general components suppliers from them and services and logistics suppliers. Some 50 of the 70 members were found to be located in geographic proximity in Jutland, mostly in the more northerly part. Of considerable interest here is Denmark’s political commitment since 1970 to wind energy and what research reveals to be North Jutland region’s wind-turbine cluster has to be addressed. On this, Andersen et al. (2006) point to the wind energy industry having passed through an early phase characterized by numerous small and medium-sized enterprises (SMEs) producing domestically-scaled wind power for individual farms and householders. But latterly, especially since the government subsidy to domestic consumers was removed in 2000, exports have risen, the scale of equipment has increased tenfold and sea power from large-scale offshore wind farms has come to predominate. As wind turbines have only some ten years’ life expectancy, most early wind turbines in rural Denmark will

Jacobian cluster emergence

27

soon disappear if they have not already done so. So the current industry structure is large Danish (Vestas) or foreign (Siemens, Gamesa, Suzlon) producers and a supply platform of SMEs. There may be less local sourcing of key equipment like gearboxes than in the early days when North Jutland shipbuilding firms could adapt to meet the nascent wind energy demand. However, the scale and adaptability of German heavy engineering in cranes and related equipment means they now supply the Danish wind energy input market. Services and special logistics firms, the latter capable of transporting the now typically massive fibreglass turbine blades also exist in proximity as do a great many components suppliers (Figure 2.2). It is worth noting that the emergence of this cluster was closely associated with the triumph of Jutland’s three-blade solution over California’s two-blade effort. Jutland’s superior model was path-dependent on marine and agricultural engineering (ship propeller and plough) while California’s was based on propeller-driven aeroplane technology. The fact the latter faced upwind while the former faced downwind added to Californian windmill blade inferiority. Stoerring (2007) agrees with this evolutionary profile pointing out that scale was also partly induced in the early 1980s by the huge demand for wind turbines from the US and more particularly California. Then, in the late 1980s this market collapsed because California’s state administration removed its subsidy regime and the Reagan administration cut alternative energy research budgets. At this time many US turbines malfunctioned badly and even the superior Danish three-blade design was prone to breakdowns. Thereafter, the industry recovered as demand in European and Asian markets rose. Nowadays (Figure 2.2) around half the global production capacity is accounted for by Danish firms like world leader Vestas Wind Systems of Randers, near Århus (acquirer of Danish firms NEG-Micon; Nordtank; Wind World) and Siemens (Bonus) at Brande and Ålborg. Gamesa Wind Engineering, Spain’s largest producer of turbines is at Silkeborg, Jutland. Suzlon, India’s leader, is located at Århus. LM Glasfiber of Lunderskov near Århus in Jutland is the leading supplier of fibreglass wind turbine blades. The other members of the North Jutland cluster are in Figure 2.2. Of the Danish Wind Industry Association’s 70 members, 50 are in Jutland, mostly northcentral Jutland. More is said on the etymology of this ‘green cluster’ evolution in the final section of this chapter. Universities (AV) join DTI (see Table 2.1) as a knowledge generation sub-system of the RIS. As noted earlier, overlapping this substantial and globally leading wind turbine technology cluster is the main Danish solar thermal energy cluster (Figure 2.3). This is smaller in scope but consists of largely indigenous firms and their suppliers. These involve firms in two types of supply chain as follows:

28

Emerging clusters

Siemens Ålborg A U

S

Randers L Vestas

S

S Suzlon

Gamesa Siemens

Vest-Fiber L

S S LMGlasfiber L

Key:

Source:

S S

Århus

DOT

J&U

BSB M BSB9

D T I

L

A U

Reichhold S

L

Fiberline

Manufacturer

Materials

Services

Logistics

Components

Danish Wind Industry Association statistics.

Figure 2.2

The north-central Jutland wind turbine cluster

Jacobian cluster emergence

Nordvestjysk Folkecenter for Vedvarende Energi

NIRAS

Energi- og Miljødata ARCON Solvarme A/S Michael Madsen & Petersen PlanEnergi Rambøll A/S, Svendborg Djurs Solvarme I/S Ajdt Miljø A/S Cowi Consult Poul Lodberg

Grundfos Sensor

Viessmann A/S

ED Heating ApS Fyens Solvarme Danfoss A/S Esbensen Sol-Energi

Source:

Composed from ESTIF data.

Figure 2.3 1.

2.

North Jutland’s solar thermal energy cluster

Glazed (roofs) ● Solar collectors ● Flat plate collectors (a) Glass (b) Heat absorbent copper/aluminium (c) Coatings, paint (d) Pipes welded to absorber plate ● Vacuum collectors (a) Parallel glass tubes (b) Absorber (c) Transfer pipes (d) Vacuum is insulator Unglazed (swimming pools) long tubes ● Synthetic absorbent material ● Hydraulics in pool filtration system ● Heat storage and back-up heating ● Plumbing and installation

29

30

Emerging clusters

Biotechnology

Electronics/IT Informatics Telecommunication

Figure 2.4

Clinical/ Hospital

Biomedical Technology

Jacobian cluster emergence in North Jutland

Finally, exemplifying North Jutland’s Jacobian cluster profile it is worth considering Figures 2.4 and 2.5, the first of which reveals established cluster evolution in the shape of the NorCom wireless communications cluster at Ålborg and the possibly emergent and overlapping biomedical technology cluster in close proximity (Stoerring, 2007). Here, the longestablished wireless telecommunications cluster (Stoerring and Dalum, 2007) has given rise to possible cluster mutation by interaction with the healthcare activities associated with clinical trials and testing of biomedical equipment. Many of these activities are closely associated with science and technology commercialization through academic entrepreneurship at Ålborg University. In Figure 2.5 the most prominent (though many have yet to be fully researched) of North Jutland’s Jacobian clusters are shown which are characterized as emergent clusters or established ones by their ‘related variety’ characteristics in relation to each other. This is made clear in Table 2.1, which shows the stylized history of a significant part of the Jutland economy’s evolution since it was radically transformed by nineteenth century railroadization as proposed by Schumpeter (1975). This process created certain path dependences or developmental trajectories. This kind of analysis is rather important and useful in explaining the ontology of such regional economies and their clusterization. Recall that for Schumpeter ‘railroadization’ was the purest, most radical kind of innovation based upon the creative destruction of a preceding state of nature (or at least non-farming economy). The massive ‘entrepreneurial event’ of ‘railroadization’ creates evolutionary trajectories that act as constrained opportunity sets for regional evolution. Activities displaying ‘related variety’ to the originating entrepreneurial event comprise the selected trajectories as in Table 2.1. These may foster varying intensities of

Jacobian cluster emergence

31 Clean Tech Biotechnology Wireless Agro-Food (Organ.) Agro-Food (Conv.) Furniture Fashion Engineering (Fish) Engineering (Pipes)

Figure 2.5

Table 2.1

North Jutland’s Jacobian clusters and related variety

Jacobian cluster path dependences

Technology

Path dependence

Clean technology

Agricultural and marine engineering (for example wind turbine blades replicate plough and propeller design) Wireless ICT and medical technology Ship-to-shore marine technology ‘Railroadization’ of undeveloped land in Jutland

Biotechnology Wireless telephony Agro-food (conventional, intensive) Organic food Furniture Fashion clothing Fish equipment and pipework engineering

Reaction against conventional intensive food production in Jutland (mostly dairy) ‘Railroadization’, craft schools and the local forestry tradition Women’s craft schools skilling farmer’s wives Fishing and marine engineering

32

Emerging clusters

innovation, from disruptive, which cheapens (for example mobile telephony) an existing but specialized, uncommoditized technology (for example ship-to-shore radio), to incremental innovations around mobile telephony (first-, second-, third-generation and so on mobile telephony). There is insufficient space to offer a satisfying explanation for the Jacobian cluster mutation process in North Jutland but Kristensen (1992) underlines ‘railroadization’ as a key process where Jutland as a whole was opened up on a smaller scale but with similar inspiration to that of the Frontier West in nineteenth century USA. With this came two key movements. The first was the farmer’s co-operative movement where farmers supplied their own production and household needs, including banks. The second movement was the craft schools, established in over 350 centres, followed by the still flourishing Danish Technological Institutes from 1907. Together these made a form of social or collective entrepreneurship possible, that is, infrastructure, education, technical support, finance and markets. Hence ‘social capital’ remains an important dimension of the SME-based collective entrepreneurship of North Jutland. It makes technological branching by means of related variety evolution possible. Finally, this is assisted by the existence of an RIS infrastructure of technological institutes, technical and craft schools and universities, which sustains entrepreneurship and localized knowledge transfer.

5

BIOENERGY FROM CROPS IN WALES

One of the most surprising, perhaps, but unquestionably innovative developments in the bioenergy field has occurred in recent years in Wales. This case, to a greater extent than the others even, tends to show related variety and platform effects may occur in rural regions as well as more heavilypopulated ones. Descriptively speaking it involves patented knowledge derived by the Institute for Grassland & Environmental Research (IGER) based at Aberystwyth in rural, central Wales. This UK Biological Research Council-funded research institute has, for 70 years to 2007, been the UK’s main, specialist grassland research institute. It was tasked from the outset with improving the quality of fodder for cattle and sheep feedstock, which is mainly grass. By the early 1980s research, which involved not simply breeding richer grasses but understanding the rumen of these ruminative animals, had revealed that a limit to quality on these mountain-bred animals occurred because the enzymes that broke down fodder into protein were actually consuming a significant portion of the nutritional value of the fodder consumed by the animal. Following many years of lengthy field trials and laboratory research, cross-breeding the basic rye-

Jacobian cluster emergence

33

grass commonly utilized for cattle and sheep fodder with breeds possessing enhanced sugar content produced optimal results. The enzymes took some of the enhanced sugar content, transforming it directly into energy but left a substantial portion for the animal sufficient for the amount, nutritional value and flavour of the animal to be significantly enhanced. This came to the market at a time when consumer demand for leaner meat of the type raised in mountainous areas rose significantly and continuous improvement to the original AberDart strain of rye-grass, marketed by Germinal Holdings, over the intervening years led to it reaching 50 per cent of the UK market. It further secured the status of Welsh Black beef and Welsh lamb as premium products and enabled significant improvements to occur in comparable upland cattle breeds such as Aberdeen Angus. In 2003, it was realized that IGER had, in the form of these SugarGrasses, an indigenous product to add to its burgeoning portfolio of biofuels. Tests had shown that SugarGrass had twice the calorific value of sugar cane, the source of much of the world’s biofuel. IGER thus evolved a second string to its grassland expertise by developing a renewables research division. One of the biofuel feedstocks in which it became supreme early on was the growing and processing of Miscanthus, more popularly known as Elephant Grass, an Asian tall grass that grows on marginal land. Accordingly it does not compete for land with food crops, one of the criticisms of the US and Europe’s ‘bolt for biofuels’. This has seen the ears and cobs of wheat and corn being turned into ethanol because of easy availability and major subsidy, causing up to 40 per cent increases in the price of such cereals, and grief in developing country food markets. Tellingly, IGER is widely perceived as in a global class of its own in these specific bioenergy subfields, the official view being that maybe University of California, Berkeley, may become competitive now they have received a $500 million endowment for a Climate Change research institute from British Petroleum (BP). The University of Illinois is also mentioned as a possible future competitor, but these are the only two and IGER has a current lead on both of them. But in any case, SugarGrass is also twice as calorific as Miscanthus and SugarGrass is thus favoured as the technology with the best long-term prospect to replace oil. IGER has the patent for SugarGrass, currently earning royalties of £100 000 per year from sales of seed varieties for fodder. But as the world awakens to the relatively simple processes of biorefining the product, these are likely to grow substantially – so much so that agreement has been reached with Welsh Government officials about the promise of funds to help build an experimental biorefinery. Thinking had gone as far as to speculate that when oil ceases to be refined at the huge Milford Haven refineries in neighbouring Pembrokeshire, the pool of talent and infrastructural sunk costs would

34

Emerging clusters PHOTOVOLTAICS CLUSTER Royce Renewables PJB Systems Sharp Solar, Dulas; PV Systems; Corus Colours, ICP Solar, Jantec Solar, IQE VE Heating Bright Light Solar

LlaniSolar

Sunset Solar

Solar Housing

Sundance Renewables Thermonax SB Alternative Micaul Solarfit Energy Solar ICP Solar Solarfit G24i Eco Energy InterSolar Systems KDUK Clear Sky

Figure 2.6

Solar energy equipment manufacturers in Wales

make them ideal candidates for becoming SugarGrass (and Miscanthus) biorefineries. These would continue to meet a huge share of the UK’s future energy. But it is not simply a spin-out-venture capital model that is in mind, possibly because a spin-out model does not yet work as well as a commercialization outsourcing model in this nascent field. For example, Molecular Nature, the key spin-out of IGER, burnt-up its venture capital, but because of the value of its patent for biofuels potential as well as its fodder market, it was acquired by spin-in company Summit. Moreover, true to the traditions of co-operation among Welsh mountain farmers, IGER promotes a new vision of mixed farming whereby groups of farmers grow Miscanthus on their poorest soil, devote some fields for SugarGrass fuel cropping and raise quality Welsh lamb or Welsh Black beef on their best SugarGrass land. Photovoltaics produce solar thermal energy as in North Jutland. In Wales, this has been studied by authors (Hendry et al., 2001) comparing the broader opto-electronics cluster, which also specializes in fibreoptic cabling, with those such as that associated with Carl Zeiss in Jena, eastern Germany. However, in relation to this present discussion about ‘green clusters’, it is the photovoltaics capability that comes to the fore. Figure

Jacobian cluster emergence

35

2.6 reveals the presence of sub-divisions of multinationals such as Japanese electronics corporation Sharp whose Sharp Solar subsidiary is based at St Asaph alongside Corus Colours, a subsidiary of Corus, the UK–Dutch steel manufacturer, acquired in 2007 by Indian giant Tata Steel. Utilizing polymer science and surface treatments Corus Colours has innovated radically a Solar Paint product capable of generating solar energy, especially from prefabricated steel buildings. Other firms in the photovoltaics cluster at St Asaph are indigenous, such as Cardiff-headquartered microprocessor firm IQE and ‘green engineering’ firm Dulas, headquartered in mid-Wales. Hence, in conclusion, we see that numerous indications of clustering among small firms, but also some large firms, along with an applied and basic research infrastructure characterizes important locations of ‘green clusters’ mainly, in this analysis, focused upon the production of nonfossil fuel energy that contributes to the moderation of global warming. A key feature to be discussed in the concluding section of this chapter is that in some cases there is an element of cluster ‘species’ multiplication which, from an evolutionary economic geography perspective can readily be hypothesized. As shown in Figure 2.1, the California cleantech clusters are to be found in juxtaposition to the ICT and biotechnology, food and wine clusters of the San Francisco region of northern California and the wireless telecom and biotechnology clusters of San Diego in southern California. Indeed, so-called cleantech is widely seen as arising from the combination of biotechnology (including biopolymers and biofuels), ICT (sensors) and nanotechnology (catalysts and filtration membranes). However, while agro-food is also one of California’s key industries, agrofood path dependence seems even more pronounced in the cases of Jutland and Wales, as we have seen, while in yet another case forestry is important to Sweden’s biofuels cluster in Örnsköldsvik (Cooke, 2007).

6

RELATED VARIETY BY OTHER MEANS: CLEANTECH IN NORWAY

The Norwegian model of developing green innovation usually involves large organizations evolving towards green innovation from intracorporate related variety. Perhaps Norway’s greatest strength in green energy is Carbon Capture and Storage (CCS). In 2007 the Norwegian government and Statoil made an agreement to establish a full-scale CO2 capture and storage project at Mongstad (near Bergen, Hordaland). To limit technical and financial risks the project will progress in two stages. The first stage covers the Mongstad CO2 capture testing facility which will be operational at the same time as the cogeneration plant starts operation

36

Emerging clusters

in 2010. The testing facility/pilot plant will have the capacity to capture at least 100 000 tonnes of CO2 per year. The second stage involves full-scale capturing of approximately 1.5 million tonnes of CO2 per year and will be in place by the end of 2014. The technology development phase of the project is currently progressing according to the project execution plan. The main objective for the pilot is to develop more cost-effective technology for CO2 capture for a wider international application, that is to develop, test, verify and demonstrate technology that would allow construction of full-scale CO2 capture plants with reduced costs and reduced technical and financial risks. A technology company will be set up to construct and operate the capture pilot, CO2 Test Centre Mongstad. The government is currently in the process of inviting companies to participate in the technology company. The invited companies are potential users of CO2 technologies and the aim is to establish a group of participants in May 2007. Several technological solutions will be tested in parallel in the project. This approach should ensure that technological developments in Norway could have broad international relevance. With the Mongstad CCS project we move from the research/small-scale phase to actual construction of a full scale CO2 capture facility. Another larger firm with a leading position in Norway’s solar energy industry is REC. This firm is the largest silicon foundry for photovoltaics in the world. The company has three divisions: first, REC is the world’s largest dedicated producer of silicon materials for photovoltaic applications and holds all rights to its proprietary production technology. Solar grade silicon produced by REC can be used in the production of both mono and multicrystalline wafers, as well as wafers based on ribbon technologies. REC is also the world’s largest producer of monosilane gas which, in addition to being used internally by REC to make solar grade silicon, can be used by others in all types of thin-film silicon applications. REC is also the world’s largest producer of multicrystalline wafers, with a history of rapid business expansion and introduction of leading production management techniques to increase productivity. REC combines high quality manufacturing equipment with proprietary technologies to achieve high productivity. Third, REC Solar’s cell and module facilities are among the most automated plants in Europe, and REC is currently developing new technology to strengthen its competitiveness and ensure future growth. The facilities are focused on few products and customers, allowing a lean approach to production. REC’s main production centres are at Sandvika and Porsgrunn in southern Norway (near Oslo) and Narvik and Glomfjord in the north. In each case a significant number of specialist suppliers are located nearby.

Jacobian cluster emergence

7

37

CONCLUSIONS AND THEORETICAL IMPLICATIONS

By virtue of an examination of the emergence of green clusters, often involving the production of new forms of non-fossil fuel energy aimed at lessening of overall greenhouse gas (GHG) emissions derived from human economic activity, a curious feature of economic evolution has been revealed. The clue lies in the element of convergence that characterizes green innovation. As hinted at in the cases of northern and southern California, not studied in detail here but examined elsewhere (Cooke, 2007), the type of ‘cleantech’ industry emerging in the clustered form described by Burtis et al. (2004, 2006) evolves from agro-food, ICT and biotechnology. In North Jutland we see something comparable having occurred. Thus the wind turbine and solar thermal clusters are found in the more agricultural and marine engineering regions of Denmark. In writing the history of the former industry, Jørgensen and Karnøe (1995) and Karnøe and Jørgensen (1996) note how the Danish design of wind turbines defeated the main global competitor from where a significant renewable technology demand also arose simultaneously from the 1970s, namely California. As noted, Danish wind turbine blade design was influenced by the agricultural engineering industry, notably the design of modern ploughing equipment. In the experimental innovation phase when some 30 firms engaged in the design of prototype turbine blades, knowledge spillovers from the design of propellers by marine engineers in the Jutland shipbuilding industry were also absorbed. This resulted in a three-blade solution and the idea that the greater efficiency in the operation of such blades came from pointing them into the wind. California’s aeronautics tradition, by the 1970s predominantly relying on jet propulsion, led to the recovery of historic knowledge of propeller-driven aeroplanes. This suggested a two-blade solution pointing downwind. The Danish solution proved far superior to the Californian in this technological contest. Hence in these multi-cluster locations, it is clear that a good deal of technological convergence is possible and probably necessary. But, interestingly, comparable technological assets do not necessarily produce optimum solutions from such Schumpeterian ‘new combinations’. Nevertheless, it is clear that in some regions cluster forms can evolve quite readily from other cluster forms, the cluster ‘species’ multiplication giving the region more of a cluster ‘platform’ characteristic to its industrial organization. On further inspection, both California and Jutland prove to have spawned many clusters. In the former case, wine clusters overlap the horticultural zones, Hollywood’s film cluster is well-known and Porter (1998) also profiles other, sometimes highly specialized clusters

38

Emerging clusters

such as the alloy golf club cluster at Carlsbad in the southern Californian desert. Further inspection of the cluster history of Jutland reveals the detailed cases of Salling (furniture) and Ikast (clothing), the even more closely studied NorCom wireless telephony cluster at Ålborg (Stoerring and Dalum, 2007), the emergent BioMedico cluster also at Ålborg, and as yet unexamined cluster candidates in insulated pipework near Ålborg, and fish-processing equipment near Skagen, at Jutland’s northern tip. At Barritskov, East Jutland is the estate that sustains the Årstiderne Organic Food Network, a co-operative retail network that delivers 30 000 boxes per week of organic food throughout Denmark. It could also be argued that there is a high degree of knowledge transfer from varieties of agricultural production to bioenergy production in Wales leading to possibly nascent cluster-formation, but also from glass technology to fibre-optic cables and then photovoltaics by a different route into renewable energy in a multi-functional opto-electronics cluster. Species multiplication or mutation of this kind would be perfectly consistent with an underlying theory of evolutionary economic geography, especially that part referring to the opportunities for innovation and growth arising where there is related variety among industries. Absorptive capacity for adaptation to new combinations based on easily understood knowledge spillovers would be the mechanism by which such species multiplication is explained, as the case of Jutland’s wind turbine technology illustrates especially clearly. In other cases focusing upon ‘green innovation’ cluster specialization as ascribed to Marshall–Arrow–Romer (MAR) thinking seems on the face of it to be more convincing than the idea of Jacobian clustering (after Jane Jacobs’ notion of innovation through variety). Yet even where limited clustering occurs, as in Rhineland or Brazil, previously existing industries such as the coal, steel and chemicals super-clusters of the Ruhr Valley or the sugar producing industry in Brazil are suggestive of the presence of important spillovers from knowledge of filtration and ventilation in the former and fermentation in the latter cases that were of profound importance to the evolution of new, convergent combinations of innovative products and processes. This tends to confirm clearly the widespread and common sense policy experience that clusters cannot be easily built in vacuo but may find it a less rigorous evolutionary trajectory to emergence where the regional context gives opportunities for Schumpeterian ‘new combinations’ from regionalized ‘related variety’. Where such related variety is more attenuated, as perhaps with biofuels in Brazil or NE England, fewer ‘Jacobian clusters’ emerge. However, that is not the whole of the explanation for Jacobian cluster mutation, rather it is an important contextual factor as noted, for example, in the work of Cantwell and Iammarino (2003). Other key features that

Jacobian cluster emergence

39

may be hypothesized, but further research is needed, is that Jacobian clustering benefits from other more social, institutional and organizational assets, such as those listed below, in addition to more economic assets concerning related variety, knowledge spillovers and high lateral absorptive capacity: ● ● ● ● ● ●

Social capital; Collective entrepreneurship; Technological branching (‘new combinations’ opportunities); Peripherality (perceived distance from key governance core); Infant industry subsidy; Innovation system – regional research and technological institutes, universities, regional innovation platform policy and funding.

The key concluding point of this section is that, for the first time in regard to new industries, we see replication of processes that have historically underpinned successful regional economies that once spawned many traditional industrial districts or clusters. Evolutionists like Klepper (2002), for example, would also highlight the transfer of routines from one to another industry by means of ‘mobility of talent’, as in the cases of the US, German and Italian automotive and engineering industries (see also Boschma and Wenting, 2007). Probably the key findings of this contribution in relation to evolutionary theory are the following. First, while Schumpeter had little to say about regional innovation, his concept of innovation by ‘railroadization’ proves to be highly apposite as an explanation of at least the case of Denmark’s opening up of North Jutland and elsewhere in the west in the nineteenth century and its modern evolution into an arena of Jacobian clustering in related variety industries. Second, the green perspective has somehow thrown the evolution of this kind of industry organization into clearer perspective because it focuses on a horizontal and convergent technology ‘platform concept’ rather than a more traditional industrial economics perspective that emphasizes vertical structures like sectors or clusters. Third, regarding cluster emergence within a regional innovation systems context the research reported showed the importance of social capital – which even in California may be considered strong, as the work of Saxenian (1994) on Silicon Valley showed – as an evolutionary driver of certain kinds of regional innovation system. Indeed, whether as ‘bonding’ or more institiutional ‘bridging’ social capital it is the key element of the hidden power of networks, both social and institutional, that has always been at the heart of the RIS approach to evolutionary science. Finally, it could be seen that the evolutionary processes described were capable of hosting differing intensities of ‘innovative bursts’. Railroadization itself

40

Emerging clusters

was said by Schumpeter to be the most radical kind. Divergent, possibly disruptive, innovations like the semiconductor industry in California and mobile telephony infrastructure in North Jutland cheapened and ‘democratized’ key technologies based upon new knowledge combinations. That other types of cluster-emergence can evolve, as around larger corporate interests in Norway, is beyond dispute and a different cluster biography from the dominant ‘mutation’ model discussed in this chapter must be composed. In the main cases discussed here, incremental, narrowly path-dependent innovation can evolve among cluster firms in proximity. History also shows there may be punctuated evolution with the more radical innovations around biotechnology from cancer-defeating therapeutics to fodder-based biofuels as knowledge evolves and broader economy regimes, notably those associated with the chemicalization of fossil fuels, approach exhaustion and make way for a potentially cleaner bioeconomy regime.

NOTES 1. Subsequently, Israel’s cleantech clusters were examined and found to be similarly convergent with agro-food, ICT and biotechnology. As noted, Israel’s new cluster at Be’er Sheva in the Negev actually coincides with the recent completion of the railway connection from Tel Aviv to that desert location (Cooke 2008c). 2. The Myrdal–Hirschman theory of economic development has been influential in the emergence of ‘new economic geography’ (for example Krugman, 2001). Anticipating the latter’s solution to the neoclassical location theory impasse by positing ‘increasing returns to scale’ rather than the rubric of ‘constant returns’ thus demonstrating the growth of cities to be a function of spatial monopoly, Myrdal (1957) proposed spatial development to be characterized by ‘cumulative causation’ with associated ‘spread’ and ‘backwash’ effects. This implies increasing returns to scale (through ‘backwash’) and developmental ‘spread’ to other nearby areas. Hirschman’s (1958) elaboration on this was that ‘spread’ would be driven by the innovative capacity of competing technology users. Under ‘knowledge economy’ conditions we hypothesize that, over relatively short time periods, primate cities grow through increasing returns (to knowledge) and ‘satellites’ of leading technology innovators ‘spread’ nearby. Our preliminary static pictures of EU NUTS 2 regions are consistent with this, while our dynamic picture of spatial divergence in Israel 1995–2002 (Cooke and Schwartz, 2008) is consistent with Myrdal– Hirschman rather than Krugman (2001), who himself admits his ‘two-locations competing’ models are misleadingly simplistic. In this respect, it can be argued, evolutionary economic geography trumps ‘new economic geography’.

REFERENCES Andersen, E. (1994) Evolutionary Economics: Post-Schumpeterian Contributions, London, Pinter. Andersen, E. (2007) Schumpeter’s Evolution, Ålborg, Ålborg University.

Jacobian cluster emergence

41

Andersen, P., Borup, M. and Olesen, M. (2006) Innovation in energy technologies, Risø Energy Report, 5, 21–7 Bergman, E., Maier, G. and Tödtling, F. (eds) (1991) Regions Reconsidered: Economic Networks, Innovation and Local Development in Industrialized Countries, London, Mansell. Boschma, R. and Lambooy, J. (2002) Knowledge, market structure, and economic coordination: dynamics of industrial districts, Growth and Change, 33, 291–311. Boschma, R. and Wenting, R. (2007) The spatial evolution of the British automobile industry: does location matter? Industrial and Corporate Change, 16, 213–38. Burtis, P., Epstein, R. and Hwang, R. (2004) Creating the California Cleantech Cluster, San Francisco, Natural Resources Defence Association. Burtis, P., Epstein, R. and Parker, N. (2006) Creating Cleantech Clusters, San Francisco, Natural Resources Defence Association. Cantwell, J. and Iammarino, S. (2003) Multinational Corporations and European Regional Systems of Innovation, London, Routledge. Chesbrough, H. (2003) Open Innovation, Boston, Harvard Business School Press. Cooke, P. (1992) Regional innovation systems: competitive regulation in the new Europe, Geoforum, 23, 365–82. Cooke, P. (1993) Regional innovation systems: an evaluation of six European cases, in P. Getimis and G. Kafkalas (eds) Urban and Regional Development in the New Europe, Athens, Topos. Cooke, P. (2002) Knowledge Economies, London, Routledge. Cooke, P. (2004) Introduction: regional innovation systems – an evolutionary approach, in Cooke, P., Heidenreich, M. and Braczyk, H. (eds), Regional Innovation Systems, London, Routledge. Cooke, P. (2007) Growth Cultures: the Global Bioeconomy and its Bioregions, London, Routledge. Cooke, P. (2008a) Regional innovation systems, clean technology and Jacobian cluster-platform policies, Regional Science Policy and Practice, 1, 9–30. Cooke, P. (2008b) Cleantech and an analysis of the platform nature of life sciences: further reflections upon platform policies, European Planning Studies, 16, 375–94. Cooke, P. (2008c) Green innovation and regional development, Presentation to Regional Science Association Meeting, Be’er Sheva, Israel, April 16. Cooke, P. and Schwartz, D. (2008) Regional knowledge economy variations: an EU–Israel comparison, Tijdschrift Voor Ekonomische en Sociale Geographie, 99, 178–92. Cooke, P., Alaez, R. and Etxebarria, G. (1991) Regional technological centres in the Basque country: an evaluation of policies, providers and user perceptions, Regional Industrial Research Report No. 9, Cardiff University. Dosi, G., Freeman, C., Nelson, R., Silverberg, G. and Soete, L. (eds) (1988) Technical Change and Economic Theory, London, Pinter. Frenken, K., Van Oort, F. and Verburg, T. (2007) Related variety, unrelated variety and regional economic growth, Regional Studies, 41, 685–97. Grabher, G. (1991) Building cathedrals in the desert: new patterns of co-operation between large and small firms in the coal, iron and steel complex of the German Ruhr area, in E. Bergman, G. Maier and F. Tödtling (eds) op cit. Gould, J. (1982) Why do honey bees have dialects? Behavioural Ecology and Sociobiology, 10, 53–6.

42

Emerging clusters

Guthman, J. (2005) Agrarian Dreams: the Paradox of Organic Farming in California, Berkeley, University of California Press. Hendry, C., Brown, J., Ganter, H.-D. and Hilland, S. (2001) Industry Clusters as a location for technology transfer and innovation: the case of opto-electronics’, Industry and Higher Education, 15, 33–41. Hirschman, A. (1958) The Strategy of Economic Development, New Haven, Yale University Press. Jacobs, J. (1969) The Economy of Cities, New York, Vintage. Johansson, B. (1991) Economic networks and self-organization, in E. Bergman, G. Maier and F. Tödtling (eds) op cit. Jørgensen, U. and Karnøe, P. (1995) The Danish wind turbine story: technical solutions to political visions, in A. Rip, T. Misa and J. Schot (eds) Managing Technology in Society: the Approach of Constructive Technology Management, London, Pinter. Karnøe, P. and Jorgensen, U. (1996) The International Position and Development of the Danish Wind Turbine Industry, Copenhagen, AKF. Klepper, S. (2002) Capabilities of new firms and the evolution of the US automobile industry, Industrial and Corporate Change, 11, 645–66. Kristensen, P. (1992) Industrial districts in West Jutland, Denmark, in F. Pyke and W. Sengenberger (eds) Industrial Districts and Local Economic Development, Geneva, International Institute for Labour Studies. Krugman, P. (2001) Where in the world is the ‘new economic geography’?, in G. Clark, M. Feldman and M. Gertler (eds) The Oxford Handbook of Economic Geography, Oxford, Oxford University Press. Lundvall, B. (1988) Innovation as an interactive process, in G. Dosi et al. (eds) op cit. Malecki, E. (1991) Technology and Economic Development, London, Longman. Meyer-Krahmer, F. (1990) Science and Technology in the Federal Republic of Germany, London, Longman. Myrdal, G. (1957) Economic Theory and Underdeveloped Regions, London, Duckworth. Porter, M. (1998) On Competition, Boston, Harvard Business School Press. Rothwell, R. and Dodgson, M. (1991) Regional technology policies: the development of regional technology transfer infrastructures, in J. Brotchie (ed.) Cities of the 21st Century, London, Longman. Saxenian, A. (1994) Regional Advantage, Cambridge, Harvard University Press. Schumpeter, J. (1975) Capitalism, Socialism and Democracy, New York, Harper. Scott, A. (2006) Spatial and organizational patterns of labour markets in industrial clusters: the case of Hollywood, in B. Asheim, P. Cooke and R. Martin (eds) Clusters and Regional Development: Critical Reflections and Explorations, London, Routledge. Simard, C. and West, J. (2003) The role of founder ties in the formation of San Diego’s ‘Wireless Valley’, Paper to DRUID Summer Conference 2003; Creating, Sharing and Transferring Knowledge: the Role of Geography, Organizations and Institutions, Copenhagen, June 12–14. Stoerring, D. (2007) Emergence and Growth of High Technology Clusters, PhD Thesis, Dept. of Business Studies, Ålborg University. Stoerring, D. and Dalum, B. (2007) Cluster emergence: a comparative study of two cases in North Jutland, Denmark, in P. Cooke and D. Schwartz (eds) Creative Regions: Technology, Culture and Knowledge Entrepreneurship, London, Routledge.

3.

Economic policy and its impact on the evolution of clusters and spatial systems exemplified by German TV programme production Ansgar Dorenkamp and Ivo Mossig

1

INTRODUCTION

When news emerged throughout German media in November 2008 that the private broadcasting company Sat.1 intended to give up its location in Berlin and relocate its business to Munich in order to reduce costs (Spiegel Online 2008), this news provoked different political echoes. While Bavarian politicians interpreted the company’s plans as a proof of Bavaria’s economic attractiveness, Berlin’s governing mayor Klaus Wowereit said that such a decision was unacceptable. This comment was justified by the fact that the federal state of Berlin had supported the Sat.1 company before by providing investment grants. These financial aids were granted on condition that Sat.1 was ready to guarantee to stay in Berlin and maintain jobs until 2012. In this context Berlin’s Minister of Economic Affairs Harald Wolf announced that he would reclaim these grants if the company did not give the binding promise to preserve jobs in Berlin (Morgenpost Online 2008). Two months later it was noted that Sat.1, starting from January 2009, would continue to employ 114 of formerly 350 employees in Berlin in spite of relocating to Munich. Capital investment grants like those paid to Sat.1 in Berlin rank among governmental incentives. By granting these incentives political authorities try to influence the behaviour of companies in favour of specific locations (Schätzl 1994, 46 ff.). Schätzl (1994, 13) calls this deliberate promotion of certain locations ‘location policies’ (Standortpolitik). Location policy aims at stimulating economic growth and employment by politically supporting existing enterprises or business start-ups. Furthermore, it creates incentives for business units to relocate to a certain place (Meyer-Stamer 1999a, 1; 1999b, 11). Efforts of location policy can be seen in conjunction

43

44

Emerging clusters

with the political support of clusters, if the efforts of location policy aim at encouraging the development of spatial concentrations and especially of the ideal types of cluster characteristics (Hospers and Beugelsdijk 2002, 383; Fromhold-Eisebith and Eisebith 2005, 1252). The paramount objective of cluster promotion is, ultimately, an increase in the competitiveness of regional economic systems and therefore of economic growth in a regional perspective (Blien and Maier 2008, 2; Fromhold-Eisebith and Eisebith 2008, 81). If corresponding supportive measures are taken or organized by political authorities, these endeavours are called ‘cluster policies’ (Clusterpolitik) (Kiese 2008a, 130). Meanwhile, political efforts directed towards the economic development in a regional perspective by using the cluster concept are being made all over Europe and all over the world (Ahedo 2004, 1097; Burfitt et al. 2007, 1273; Bathelt and Dewald 2008, 163). Cluster policy is understood as a specific kind of ‘industrial policy’ (Industriepolitik) targeted at supporting regional economic characteristics (Kiese 2008a, 130). This industrial policy includes characteristics of science policy, technology policy and regional development policy (Hospers 2005, 383; Hospers et al. 2008, 5). Industrial policy is at the same time an important part of the government’s ‘economic policy’ (Wirtschaftspolitik) (von Einem 1991, 17). In general, there exists a strong link between state and politics. The state is called the organizational form of political issues (Schultze 2005, 944). Consequently, the state represents the arena that embodies the entire ensembles of institutions in which political decisions are made by political officials (Gellner and Glatzmeier 2004, 18). Thus the state can be regarded as the specific and institutionalized context in which politics are taking place (Gellner and Glatzmeier 2004, 19). All state-run activities are thus prepared and carried out by political activities and decisions. Cluster policy, representing a regionally-oriented special case of industrial policy, also turns out to be part of state-run measures of economic policy.1 The term ‘economic policy’ describes every kind of intentional influence exerted by public agencies and aiming at shaping the economy and its basic conditions (Luckenbach 2000, 1). Koch et al. (2008, 4 ff.) divide economic policy into three classes, namely ‘system policy’ (Ordnungspolitik), ‘adjustment policy’ (Prozesspolitik) and ‘structural policy’ (Strukturpolitik). Measures of cluster policy are assigned to structural policy, which aims at facilitating economic adaptation processes of sectors and regions. Governmental incentives like subsidies and investment grants are called direct measures of structural policy (Koch et al. 2008, 5). They are designed to shape entrepreneurial decisions in various ways. The same applies to measures of adjustment policy, which summarizes, for instance, decisions on monetary policy, fiscal policy and exchange rate policy (for example the

Clusters and spatial systems in German TV production

45

increase or cut in interest rates and direct taxes). Adjustment policy also often causes entrepreneurial modifications (Donges and Freytag 2004, 269; Welfens 2008, 489). However, the most extensive political influence on entrepreneurial activities is exercised by national system policy. National system policy aims at shaping the judicial and institutional arrangement of national economic systems, in Germany designed as social market economy (soziale Marktwirtschaft) (Wildmann 2007, 5). The constitution and arrangement of economic systems therefore means the setting of fairly long-term targeted binding rules and norms, which define the range of entrepreneurial activities and in this way normally restrict them (Welfens 2008, 489). However, the relationship between system policy and adjustment policy is a hierarchical one, characterized by the fact that measures of adjustment policy may only be carried out if the free-market system is sustained (Donges and Freytag 2004, 267). Even though economists disagree about the necessity, intensity and scale of economic policy interventions, these kinds of governmental interference are not unusual. In fact in Germany they have always been practiced. They concern industrial companies, but increasingly business services. On the one hand these remarks point at the important role of measures of cluster policy as a specific mode of efforts of economic policy. On the other hand they show that measures of cluster policy are less significant than the meaning of measures of adjustment policy and especially system policy in shaping entrepreneurial performance and consequent spatial behaviour. Although measures of system policy predominantly aim at the national level and are therefore not restricted to the regional level, they nevertheless can have an effect on the regional level. This is why Kiese (2008a, 130 f.) highlights the cluster-supporting role of even those staterun measures which do not explicitly aim at cluster promotion. For this reason it seems to be appropriate not to confine discussions about the conception, design, application and evaluation of political measures and their spatial effects to the regional level by using the label of ‘cluster policy’, as done in many recent publications (for example Hospers and Beugelsdijk 2002; Raines 2002; Hospers 2005; Burfitt et al. 2007; Alecke et al. 2008; Hegedüs 2008; Hospers et al. 2008; Aziz and Norhashim 2008; Bathelt and Dewald 2008; Burfitt and MacNeill 2008; Falck et al. 2008; Kiese 2008a; 2008b; Mans et al. 2008). We rather think that it is reasonable to basically analyse how national system policy, adjustment policy or structural policy can affect the course and the rules of the emergence and evolution of clusters and sectoral spatial concentrations. Due to the ‘primacy of system policy’ (Donges and Freytag 2004, 269), especially the decisions of system policy seem to be appropriate for analysing the spatial effects of national economic policy. Besides this setting of priorities it will be discussed in

46

Emerging clusters

this chapter how this process can be influenced by measures of structural policy aiming at the promotion of certain locations (location policy). The sectoral context for analysing these problems is represented by German broadcasting activities, especially by activities dealing with the production of TV programmes (called ‘TV industry’ in the following). German broadcasting, comprising sound broadcasting and television broadcasting, is characterized by strong interference of system policy, occurring in intense measures of regulation policy2 in all broadcasting markets. Furthermore, since the 1970s especially Germany’s federal states (Bundesländer) have focused their interests and activities with regard to structural policy on the TV industry. Closely related efforts of locational policy for attracting TV companies are based upon the positive effects of growth and employment that demonstrably occurred in Munich and Cologne (Hoffmann-Riem 1986, 14; Nünning 2003, 54; Mossig 2006, 113 ff.). For this reason politicians in Germany’s federal states have regarded the TV industry as an effective instrument for restructuring regional economic landscapes and transforming the industrial society into a ‘knowledge society’ (Sydow and Staber 2002, 221; Freundt 2003, 91 ff.; Mai 2005, 61). The analysis of the decisions of structural policy in the shape of measures of a federal state’s locational policy also seems to be interesting against the background that Germany’s federal system may lead to strong competition between federal states for attracting promising industries. This can exert an influence on the evolution of clusters and spatial systems. The chapter is structured as follows. According to theoretical principles, we argue in section 2 how clusters and spatial systems emerge in newly originating industries assuming that there is free market competition and no interference on the part of economic policy. On the basis of these findings we will show how the emergence and evolution of clusters and spatial systems can be shaped by measures of system policy and structural policy. In section 3 the current spatial pattern of Germany’s TV industry is presented. In section 4, the theoretical explanations will be taken up and tested as an example by analysing the evolution of Germany’s TV industry and its spatial pattern. In section 5 we come to a conclusion by summarizing the findings and pointing out the need for additional research in economic geography.

2

THEORETICAL FRAMEWORK: THE EVOLUTION OF CLUSTERS AND SPATIAL SYSTEMS AND THE SIGNIFICANCE OF ECONOMIC POLICY

In order to understand how decisions of economic policy influence processes of cluster formation and the emergence of spatial systems we present

Clusters and spatial systems in German TV production

47

a concept for explaining underlying processes in newly originating industries, while assuming the absence of measures on the part of economic policy. We use an evolutionary perspective concerning random events, presuming that the emergence and evolution of industrial spatial systems is mainly influenced by the action of companies and their organizational and technological abilities and routines. Thus they can create path dependency that may lead to a customary mode of regional economic development. Afterwards German broadcasting will serve as an example to show how these processes can be shaped by the evolution of system policy and structural policy and their respective measures. In their study about spatial clustering processes Storper and Walker (1989) argue on the assumption of free market forces and unfettered competition. They do not consider state-run market interventions. According to them, the first pioneer enterprises in new lines of business play a decisive role at the beginning of cluster formation because they decide about the place for locating their business units. The location is not predetermined, but the pioneer enterprise rather has some degree of free choice of localization (‘windows of locational opportunity’, ibid.). It is true that some urbanization and localization economies can have an influence on localization decisions (Boschma and Wenting 2005, 4 ff.), but the concrete localization decision of a pioneer business is mainly based on the founder’s individual circumstances and is for this reason also shaped by chance (Boschma and Lambooy 1999, 414; Mossig 2008). New lines of business are normally not opened up by multiple pioneer enterprises. In fact, emerging markets are at first entered by one pioneer who settles in a location. This first enterprise is called ‘market pioneer’ (Lieberman and Montgomery 1988; Robinson 1988; Min et al. 2006), or ‘first mover’ (for example Kerin et al. 1992; Koh 1993; Robinson et al. 1994; VanderWerf and Mahon 1997). In free market economies the first mover is able to gain some advantages. Literature calls them ‘market pioneer advantages’ (Robinson and Fornell 1985), ‘pioneer advantages’ (Bohlmann et al. 2002) or more familiarly ‘first-mover advantages’ (for example Lieberman and Montgomery 1988, 1998; Kerin et al. 1992; Robinson et al. 1994; Frynas et al. 2006; Min et al. 2006; also: Dorenkamp 2008, 63 ff.). The concept of ‘first-mover advantage’ shows how a first mover can benefit from pioneering in multiple ways and therefore capitalize on these advantages as soon as the following enterprises enter the market and break the first mover’s temporary monopoly. Altogether, the market pioneer’s numerous opportunities of realizing first-mover advantages cause barriers to market entry for followers since followers face higher costs and greater challenges than the market pioneer (Lieberman and Montgomery 1988, 43 ff.; Agarwal and Gort 2001, 162 ff.). But

48

Emerging clusters

followers do not face insurmountable barriers to entry, they are rather able to gain advantages, too (‘later entrant advantages’, Kerin et al. 1992, 46; ‘late-mover advantages’, Shankar et al. 1998; ‘second-mover advantages’, Hoppe and Lehmann-Grube 2001), whose realization might lead to ‘pioneer disadvantages’ (Bohlmann et al. 2002, 1176) or more familiarly ‘first-mover disadvantages’ (Lieberman and Montgomery 1988, 47, 1998; Robinson et al. 1994, 18; Min et al. 2006, 17). Therefore, first-mover advantages never occur automatically by entering a new market as the first enterprise. In fact, market pioneers only have the opportunity to realize first-mover advantages by the time their monopoly is broken. This does not mean that market pioneers possess better skills than their followers (Robinson et al. 1994, 12). The market pioneers will only be able to gain lasting advantages if they possess organizational routines that meet the challenges of the company’s environment in the best possible way, and enable them to realize the maximum of the potential first-mover advantages (Dorenkamp 2008, 67 f.). To what extent the market pioneers are able to benefit from these opportunities mainly depends on their own skills and less on the skills of the followers. However, if one takes a look at the development of the long-term business performance of market pioneers, it becomes obvious that market pioneers tend to achieve better results than the following enterprises. For example, numerous studies indicate that market pioneers possess a ‘market share advantage’ (Robinson 1988) and can generate long-term and higher market shares than the followers (for example Robinson and Fornell 1985; Robinson 1988; Urban et al. 1986; Kerin et al. 1992; Kalyanaram and Urban 1992; Huff and Robinson 1994), but this interrelationship does not exist inevitably (Frynas et al. 2006). However, it is assumed that first-mover advantages outweigh first-mover disadvantages (Robinson et al. 1994, 18). These facts are of considerable importance for the development of locations and clusters because the market pioneer represents the first enterprise on a particular market. According to Storper and Walker (1989, 70 ff.), the pioneer starts the industrial localization of incipient industrial activities by finding a location. In the course of such an industrial evolution competitive advantages by internal and external economies of scale will arise in the location that possesses enterprises with high market shares (and these are usually the market pioneers, see above). Thus, the spatial choice for following businesses will become dramatically smaller. Prospering businesses can realize economies of scale by vertical integration and exert a pull on employees and suppliers. This means that processes of agglomeration can be reinforced simply by the dominant size of this single enterprise (Storper and Walker 1989 78; Bathelt 1991, 363; Mossig 2000, 42). If externally

Clusters and spatial systems in German TV production

49

interwoven production complexes evolve (for this see Storper and Walker 1989, 79 ff.; Bathelt 1991, 363 f.; Bathelt and Glückler 2002, 209), the successful enterprise also acts as the unit that initiates the demand for products and services. Linkages of vertical and horizontal interdependence emerge while the increasing degree of interconnection heightens processes of concentration by enlarging local potentials for supply and marketing. Owing to their dynamic growth, these regions attract more and more mobile production factors like employees and capital from other regions (Kulke 2004, 118 ff.). Connected to these processes, the initial demand coming from the successful enterprise creates space for the forming of new enterprises at the location by setting incentives for the foundation or relocation of new firms. This will happen by firms moving in from other locations or by endogenous processes. If the number of firms exceeds a specific ‘critical mass’ (Witt 1997, 768; Brenner and Fornahl 2003, 137), positive agglomeration effects arise. These agglomeration effects determine the development of cluster-specific advantages, which in turn strengthen processes of clustering (Boschma and Wenting 2005, 5). In this way the market pioneer is able to achieve regional leadership which can be the basis for subsequent processes of cluster formation. Endogenous processes are of especially great importance for the formation of clusters. This is related to the multiplying effects of successful organizational routines by firm-specific reproduction intensities which appear in spin-off-processes (Klepper 2002, 662 ff.; Boschma and Frenken 2003, 186 ff.; 2006, 278 f.; Boschma and Wenting 2005, 2; Dorenkamp and Mossig 2006, 290 ff.; Buenstorf 2007, 4). Accordingly, various processes of clustering at different places are determined by the quality of organizational routines that are passed on by reproduction processes (Dorenkamp 2008, 69 ff.). This process is driven by entrepreneurial competition for the best organizational routines. Pioneer enterprises that possess very robust routines can sustain their position in the market and reproduce their successful routines via spin-off-processes to following businesses. These enterprises are called ‘incubators’ because by multiplying routines in a regional context they create clustering paths that can also be reinforced by spun-off enterprises. By this means the ‘incubator’ creates a local ‘seedbed’ if corresponding opportunities and space for the foundation of new firms exist locally (Hayter 1997). Due to the specific success of the pioneers that possess efficacious routines, the prospering seedbed is created especially by spin-off-processes that arise from these kinds of successfully organized firms. By this means sustainable processes of clustering can be launched, because spin-off enterprises locate their production units by reason of the founder’s personal connections to the particular region usually in close proximity to the incubator (Boschma and Lambooy 1999, 417; Mossig

50

Emerging clusters

2000, 52; May et al. 2001, 374; Boschma and Frenken 2003, 187; 2006, 292; Boschma and Wenting 2005, 4). If, however, an enterprise does not have appropriate organizational routines, it will not sustain its position on the market and will therefore fail to encourage sustainable regional reproduction processes by creating seedbeds. Consequently, such an enterprise will not be able to initiate successful clustering processes. These are the mechanisms leading to selective clustering as mentioned by Storper and Walker (1989). Altogether these remarks show that with regard to new industries the region in which the market pioneer is located has better conditions for successful clustering processes than the region where followers have settled. This is because market pioneers tend to have bigger chances of gaining a dominant market position because of the opportunities to realize firstmover advantages. Furthermore, they have more time and options available for developing successful and effective organizational routines which can be augmented in a regional context by spin-off-processes and at the same time, together with those new firms that are founded in or relocate to the same location, exceed the critical mass which itself leads to selfenergizing processes of clustering (Dorenkamp 2008, 67 ff.). It is true that the inconsistent interconnection between first entering a new market and the certainty of such a firm’s durable market shares shows that there are no predetermined results. This means that primarily the enterprises themselves exert influence on their successful business performance by using their routines. But in this regard the initial conditions for the market pioneer should be better than for the following companies. This is the reason why first-mover advantages will tend to develop into regional first-mover advantages. This means that the conditions for successful cluster formation are better at the market pioneer’s location, while it seems more probable that the processes of cluster formation will start later or progress less successfully at the location of the companies that follow later. In reality, these ideal courses of processes of spatial concentration in new industries are often influenced by measures of system policy (Shaffer 1995; Frynas et al. 2006). These measures can have a key influence on a company’s performance and have to be considered as a determining factor in business competition. In this regard, public authorities of system policy have a number of options for interfering in economic processes. Spatial effects especially arise from decisions in regulation policy. These regulating decisions of system policy are carried out if powerful natural monopolies exist. In this case regulation decisions prevail by the regulation of market patterns, which provides for the vertical splitting of upstream and downstream markets (Knieps 2007, 153 ff.; 2008, 79 ff.).

Clusters and spatial systems in German TV production

51

This can be observed in the German telecommunications and electricity supply sectors. Furthermore, regulating decisions of system policy manifest themselves by constraining market entry and market exit for the regulated businesses concerned (Agarwal and Gort 2001; Welfens 2008, 489). In this case framework regulations of system policy can protect market pioneers or even give them a kind of monopoly and therefore exclude prospective followers from market entry by establishing market entry barriers. At the same time the market pioneer may gain an additional first-mover advantage by being given the best geographical location or the best marketing channels. These market entry barriers created by decisions of regulation policy are extraordinarily effective because they artificially extend the market pioneers’ temporary monopoly beyond the normal dimensions prevailing under free market forces and allow them to expand their first-mover advantages over a rather long period and thus underpin their regional leadership in an extraordinary way. This can be the basis for subsequent processes concerning the establishment or relocation of firms, of firm founding, relocating and spin-off measures. These remarks especially apply to the German broadcasting business. Because of its specific characteristics the broadcasting business has been one of Germany’s highest regulated sectors of the economy up to now (Heinrich 2002, 83; Puppis 2007, 88). In German broadcasting measures of system policy mainly occur by decisions of media or broadcasting policy (Heinrich 2002, 74). While media policy comprises all kinds of activities ‘that aim at the creation and enforcement of generally binding rules and decisions concerning media organizations and public mass-media communications’ (Puppis 2007, 36), broadcasting policy deals with organizing and developing public communication inside broadcasting as a mass medium (Donges 2002, 39). In general, German media policy is mainly shaped by broadcasting policy (Jarren 2007, 135). Media policy forms public communications by regulatory means like law-making (Vowe 2007, 78). General principles of broadcasting regulation that constitute the basis for decisions in media policy are reflected in the Basic Constitutional Law, the inter-state treaty on broadcasting, the federal broadcasting laws, and in the legislation and jurisdiction of the European Union (Heinrich 2002, 85). The rulings of the Federal Constitutional Court are of great additional importance. Germany’s media policy is mainly based on historical experiences during World War II. The central aim is to avoid totalitarian control over the media again (Vowe 2007, 76 f.). Therefore the Federal Constitutional Court has specified the task of broadcasting. Broadcasting has to allow for the consumers’ freedom to form their own opinions. That is why diversity of opinion has to be presented as extensively as possible. At the same time broadcasting

52

Emerging clusters

has to be kept free from any direct influence and control on the part of the state and is to be protected from any control and exertion of influence on the part of a broadcasting monopolist (Heinrich 2002, 85). This is to ensure that the consumers’ freedom to form their own opinion remains protected (ibid., 86). Thus, the main objective of regulation based on system policy is the promotion of the public interest (Francis 1993, 1 f.). In particular, socio-political considerations necessitate the regulation of broadcasting. Mass-media are of great social and cultural importance because they bring information to the public and thus contribute to the process of the population’s formation of opinion (Beck 2005, 56). As a forum for political debates they assume an important role for the successful functioning of democratic systems (Puppis 2007, 82 f.). From an economic perspective the avoidance of market failure, the combating of the monopolistic abuse of power and of cutthroat competition are also of public interest (Heinrich 2001, 75; 2002, 83). Market failure occurs in media markets because media contents are public goods. Other reasons for market failure in media markets are the accumulation of external effects on media coverage and on consuming media contents, the structural problems of competition, the lack of transparency regarding media coverage (informational defects) and blurred or warped preferences of consumers (‘merit goods’) (Puppis 2007, 67 ff.). Altogether it becomes clear that media goods are not only economic but also cultural goods (ibid., 82 f.). The media have to deal with the conflicting aims of being true to their public mission and of maximizing profits. That is why economic competition cannot ensure the provision of media contents of great variety and of high quality. If there is a social desire for these media contents, media regulation can promote the production of this kind of goods. Freedom from a monopolistic abuse of power and the guarantee of the good quality of radio and TV programmes was first realized in Germany by the decision of broadcasting policy not to tolerate private broadcasters but rather install broadcasting corporations under public law (öffentlichrechtliche Rundfunkanstalt). These corporations under public law were given a specific mission concerning their programmes. Private broadcasters were debarred from market entry until 1984 and afterwards controlled by specific agencies of supervision, the so-called federal media institutions (Landesmedienanstalten), which can decide on the admittance of private broadcasters within the scope of licensing processes (Heinrich 2002, 84). Due to a federal state’s responsibility for broadcasting questions, each German federal state has such an agency at its disposal. In the course of licensing processes each federal state possessed a number of options for starting measures of structural policy. These options existed because

Clusters and spatial systems in German TV production

53

Germany’s federal states, or rather the federal media institutions, were in control of the powerful resource ‘licence’ and were thus able to make demands concerning the choice of location on private broadcasters that applied for such a licence. Between the foundation of the first broadcasting corporation under public law and the admittance of the first private broadcaster lay a period of more than 30 years, in which the broadcasting corporations under public law were monopolistic market pioneers that were protected by the government (Voelzkow 2002, 141). These explanations show that the formation and development of the television industry took place under the strong influence of system and structural policy. This is why the TV industry represents an appropriate line of business for analysing how processes of industrial location and selective clustering (Storper and Walker 1989) were shaped by measures of national system and structural policy. With reference to these findings, the following analysis will examine how pioneering broadcasting corporations under public law settled down in their locations. At the same time it will be analysed to what extent these market pioneers under public law were able to convert their monopolistic regional leadership into regional first-mover advantages by acting as incubators and thus creating seedbeds for subsequent processes of cluster formation. The role and influence of location policy in finding business locations for private broadcasters and in shaping processes of clustering will also be scrutinized. Altogether it will be investigated how decisions and influences of system policy and structural policy led to the localization of television companies and therefore initiated and strengthened processes of selective clustering, which led to the evolution of the spatial pattern of Germany’s TV industry today. First, the current spatial pattern of Germany’s TV industry is presented in the next section.

3

THE SPATIAL PATTERN OF GERMANY’S TV INDUSTRY, 2003

First, Figure 3.1 shows the spatial concentration of television production units in Germany in 2003. The figure reveals that most of the broadcasters’ revenues are earned in the cities of Cologne, Munich, Mainz, Berlin and Hamburg. The financially strongest location, Cologne, is home to Germany’s most important private broadcaster, Radio Television Luxemburg (RTL), as well as the broadcaster under public law, West German Broadcasting (Westdeutscher Rundfunk, WDR), and the private broadcasters VOX, SUPER RTL, VIVA and N-TV. The TV-related incomes in the second-best location, Munich, are mostly generated by

54

Emerging clusters

SchleswigHolstein Kiel MecklenburgVorpommern NDR

Schwerin

Hamburg Bremen RB

Niedersachsen

RBB

Berlin Hannover

MTV

Potsdam

Magdeburg Sat. 1

Sachsen-Anhalt

VIVA SUPER n-tv RTL

Brandenburg

Nordrhein-Westfalen Bochum

VOX

Düsseldorf

MDR

Dortmund

Leipzig

Kassel

WDR

Dresden

Köln

Erfurt

Hürth

Thüringen

Hessen

RTL

Sachsen

SWR

Wiesbaden

RheinlandPfalz

Frankfurt

Mainz

Revenues/Year

ZDF

Saarland

4 Bn SR

Bayern Saarbrücken

2 Bn RTL II SWR

1 Bn

SWR

Baden Baden

0.5 Bn

Stuttgart PRO 7

Ismaning

BadenWürttemberg

München 9Live N 24

0

Source:

50

0.1 Bn

BR

100 km

Unterföhring Grünwald Premiere

TV-production firm, TV-studio operator

DSF Kabel1

Outline: A. Dorenkamp Cartography: B. Goecke

Dorenkamp and Mossig (2006, 294).

Figure 3.1

Spatial concentration of the TV industry in Germany (2003)

Clusters and spatial systems in German TV production

55

three broadcasters belonging to the private ProSiebenSat.1 Media AG (ProSieben, Kabel1 and N24). The broadcasting corporation under public law Bavarian Broadcasting (Bayerischer Rundfunk, BR), is also very important, as are the private broadcasters Radio Television Luxemburg 2 (RTL II) and smaller channels like German Sports TV (Deutsches Sportfernsehen, DSF) and 9Live. Europe’s biggest broadcaster, Second German Television (Zweites Deutsches Fernsehen, ZDF), is situated at the financially third strongest location, Mainz, and earns three-quarters of the location’s TV income. This broadcaster under public law is followed by another one, Southwest Broadcasting (Südwestrundfunk, SWR), which operates a broadcasting channel in Mainz.3 Mainz is followed by Berlin, which houses the private broadcaster Sat.1, the broadcaster under public law, Broadcasting BerlinBrandenburg (Rundfunk Berlin-Brandenburg, RBB), and also MTV; and Hamburg, where the North German Broadcasting (Norddeutscher Rundfunk, NDR) has its seat. In Cologne, Munich, Berlin and Hamburg private telecasting production units, services and distributors have settled in direct neighbourhood to the broadcasters. With 103 production services Munich is the most important place (Munich: 70; Grünwald, Ismaning, Unterföhring: 33), followed by Berlin with 94 units (Berlin: 70; Potsdam: 20), Cologne with 73 units (63 in Cologne; 10 in Hürth) and Hamburg with 34 units. These figures show that successful clustering processes took place in Cologne, Munich, Hamburg and Berlin. Similar developments failed to appear in Mainz, even though the broadcasters ZDF, SWR and formerly Sat.1 (Sat.1 was located in Mainz between 1985 and 1999 and then relocated to Berlin) were or still are strong employers and therefore prospective customers (Dorenkamp and Mossig 2006; Dorenkamp 2008). In order to analyze the meaning of economic policy for the genesis of the German broadcasting system and its spatial pattern we firstly evaluated comprehensive sources of secondary literature. Owing to the fact that many decisions in broadcasting policy occur in informal systems (Isenberg 2007), we conducted 191 guided interviews with political and private representatives of German television production units. We conducted the interviews between 2001 and 2006 in Cologne, Munich, Hamburg, Berlin, Frankfurt/Mainz/Wiesbaden and in other locations (see Table 3.1) in order to measure the importance of the influence of economic policy on the processes of clustering in Germany’s television industry. In the following, the results of these empirical surveys are illustrated. They show that the evolution of the spatial pattern of Germany’s TV industry and of particular clusters can only be understood by considering the measures and decisions of economic policy on national and regional levels.

56

Source:

Own survey.

TV broadcaster TV production firm Studio operator/ technical service provider Other service providers (actors’ agency, casting, camera crews, consulting, cutter/ post-production, screenwriters, directors) Marketing (airtime marketing, public relations) Institutional environment (organizations, universities, funding, film subsidies, business development, policy) Overall number of Interviews

6 21 2 8

4 3

44

1 7

46

Cologne

9 22 4 3

Munich

28

–

–

– 5 – 23

Hamburg

43

2

–

13 26 2 –

Mainz/ Wiesbaden/Frankfurt

Interviewees

25

–

–

4 9 2 10

Berlin

5

1

–

1 2 1 –

Others

Interviews in Munich, Cologne, Hamburg, Berlin, Mainz/Wiesbaden/ Frankfurt and other locations

Position inside the production system

Table 3.1

191

3

5

33 85 11 44

All in all

Clusters and spatial systems in German TV production

4

57

EMPIRICAL LINKS: THE EVOLUTION OF THE SPATIAL PATTERN OF GERMANY’S TV INDUSTRY AS A RESULT OF DECISIONS IN ECONOMIC POLICY

Already during the Weimar Republic the state exerted an influence on German broadcasting (Wilke 2003, 157). During that time broadcasting was under the technical and legal control of the administration of the empire’s telegraphy (Reichstelegraphenverwaltung), which was subordinated to the empire’s post ministry (Reichspostministerium), and this possessed a monopoly of broadcasting and telegraphy (Fritze 1992, 23). This monopoly existed also because broadcasting was destined for military use (Beck 2005, 168). For this reason the empire’s post ministry had a dominant economic and technological position because all the transmitters belonged to that ministry (ibid.). In order to broadcast a programme, each broadcaster needed a licence that could only be granted by the administration of the empire’s telegraphy. Being an institution of the empire’s post ministry, this administrative department possessed the exclusive authority to allocate transmitting frequencies. Having control over licences meant having control over a powerful resource, and by using this resource the empire’s post ministry was able to gain influence on the contents of broadcasting. The first licensed company was ‘Deutsche Stunde, Gesellschaft für drahtlose Belehrung und Unterhaltung mbH’. This company was privately organized but in fact firmly integrated into the empire’s administration by complex holdings (Beck 2005, 167). However, during the 1920s technological constraints existed so that broadcasting a programme was not possible by using a central transmitter of the empire (Fritze 1992, 25). In conjunction with this fact considerations of regional peculiarities were responsible for the distinct foundation of regional stock corporations all over the republic. These corporations were distributed according to the competency areas of the Weimar Republic’s Foreign Office (Dorenkamp 2004, 2008; Beck 2005, 167). That is why regional stock corporations of the ‘Deutsche Stunde’ were founded and licensed for example in Berlin (29 October 1923), Munich (30 March 1924) and Hamburg (5 February 1924). The corporation that had been intended for Cologne had to be located in Münster at first (10 October 1924) because the Rhineland was being occupied. But when the French soldiers left the Rhineland, the head office of that corporation was relocated to Cologne in 1926 (Stuiber 1998, 148). Thus, the federal broadcasting structure during the Weimar Republic cannot be regarded as a result of federal virtue (Stuiber 1998, 145). In fact this structure necessarily arose from technical needs (Fritze 1992, 25). At the same time these circumstances satisfied the growing interest of the federal states in broadcasting (ibid., 26).

58

Emerging clusters

It is important to know that the previously-mentioned processes during the Weimar Republic happened by an internal interplay between the executive authorities of the empire and of the federal states (ibid., 29). Already here, there were indications of tendencies that were to become characteristic of the Federal Republic’s broadcasting policy, namely decision-making by informal bargaining between different political partners (Jarren 1997, 209). In the case of the Weimar Republic the obvious intention of the government was the basis for these processes, aiming at gaining a strong influence of the state on broadcasting. This development was completed between 1933 and 1945, when broadcasting was nationalized and forced into line. Until Germany’s surrender in May 1945 broadcasting was used by Hitler’s NSDAP as a means of propaganda. This propaganda misled millions of Germans into believing that it was right to wage World War II. After the Allies had gained control over Germany, all the state authority over Germany lay in their hands. Speedily they made an effort to rebuild German broadcasting and confiscated all the technical equipment of National Socialist television and nearly all the transmitters (Bausch 1980, 43 ff.; Altendorfer 2004, 319). Subsequent broadcasting began in all the zones of occupation by means of military transmitters. In the American occupation zone Radio Munich started on 12 May, 1945, Radio Hamburg in the British sector on 4 May 1945, Radio Berlin in the Soviet occupation zone on 13 May 1945 and Radio Koblenz in the French sector on 14 October 1945 (Bausch 1980, 43 ff.). However, besides rebuilding the technical infrastructure it was necessary to systematically redesign a new broadcasting system. As a consequence of the negative experiences people had during World War II by the misuse of broadcasting for purposes of propaganda, the Allies brought about a decision of regulation policy: the rebuilding of German broadcasting was not to be modelled on any form of state-run broadcasting (Bausch 1980, 20; Beck 2005, 170). However, the Allies could not agree on a collective concept for Germany’s new broadcasting system (Bausch 1980, 18). That is why the Allies developed different models for broadcasting inside their occupation zones (see in the following Altendorfer 2004, 305 ff.), but each model provided for the establishment of regional broadcasters in each sector. American broadcasting politicians decided not to impose the American commercial broadcasting system. In fact America aimed at creating a decentralized broadcasting system that was to be free from governmental and party political influence. Therefore they established a system of corporations under public law by founding the regional broadcasters Bavarian Broadcasting (Bayerischer Rundfunk, BR) and Hessian Broadcasting (Hessischer Rundfunk, HR) on 1 October 1948, Bremen Broadcasting

Clusters and spatial systems in German TV production

59

(Radio Bremen, RB) on 15 March 1949 and South German Broadcasting (Süddeutscher Rundfunk, SDR) on 12 May 1949. Inside the British sector media politicians decided to establish a central broadcasting corporation under public law, modelled on the British BBC, the broadcaster Northwest German Broadcasting (Nordwestdeutscher Rundfunk, NWDR), which was founded on 1 January 1948 and which had its headquarters in Hamburg and branches in Cologne, Hanover and Berlin. The French established Southwest Broadcasting (Südwestfunk, SWF) as a broadcasting corporation under public law on 30 October 1948 but put it under the control of the French commander-in-chief. Inside the Soviet sector a centrally directed broadcasting system was created between 1945 and 1949 with a main transmitter in Berlin and smaller transmitters in Leipzig, Dresden, Schwerin, Potsdam, Weimar and Halle. The decision of Allied media policy to organize German broadcasting according to the idea of corporations under public law was the first regulatory intervention in German broadcasting. The historical examination of the genesis of Germany’s broadcasting system reveals that broadcasters under public law were not able to decide freely on their locations. The decisions were made in accordance with the political guidelines of the Allies and were mainly governed by practical considerations (Bausch 1980, 46 ff.). A completely new location for the occupier’s military transmitters, Baden-Baden, was only chosen by the French occupying army. The British and Americans used the technical infrastructure and the transmitters in those cities that had been the headquarters of broadcasting institutions during the Third Reich. Predominantly in those places the control centres of broadcasting organizations were established. Hence, the decisions of regulation policy on the organization of the German broadcasting system under public law led to the settlement of broadcasters under public law in their respective locations. This means that their choice of location was not made due to existing ‘windows of locational opportunity’. The same facts applied to the additionally founded ZDF, which was located in Mainz after lengthy and informal bargaining processes between the prime ministers of Germany’s federal states and a concluding crucial vote (Dorenkamp 2008, 150 f.). The decision on the foundation and localization of the ZDF was again governed by politics and can be traced back to the decisions of regulation policy by the Allies to establish the German broadcasting system under public law. It was this decision that enabled German political authorities to influence the choice of the broadcasters’ location. Furthermore, the decisions of regulation policy resulted in the fact that the broadcasters under public law became publically protected monopolists that took advantage of the market in the absence of competition

60

Emerging clusters

and that were in a position to establish strong systems of production and distribution in their respective locations and therefore gain a regional leadership which could be the starting point for subsequent processes of cluster formation. In order to fulfil their legally required and defined programme tasks, which include the preparation of high-quality balanced and extensive information, the presentation of a variety of opinions and the production of cultural and entertaining programmes, the broadcasters under public law decided to produce telecasts by only using their own personnel and capacities of production and therefore not placing orders with private companies (Hoffmann-Riem 1999, 30). Thus, the broadcasters under public law established strong vertical-integrated systems of TV programme production and distribution and failed to create a regional demand for suppliers. By this means, the politically established regional leadership of broadcasters under public law could not be converted into a regional first-mover advantage at the respective locations. The broadcasters under public law did not become incubators and failed to create seedbeds for significant foundations of firms, for relocations and spin-offprocesses and consequent processes of cluster formation. When the Federal Constitutional Court decided in its third TV ruling in 1981 (BVerfGE 57, 295) that the introduction of private broadcasting was possible in principle on certain conditions, it paved the way for today’s dual broadcasting system with its coexistence of broadcasters under public law and private broadcasters. When private broadcasting companies were founded, the states found many opportunities to take an active role in location policy. Also the private broadcasters in Germany had to be subjected to regulation in order to ensure that they would meet the aims of broadcasting policy. Thus, rules on market entry, programme control and funding were also developed for private broadcasters. The adherence to these rules was secured by each federal state’s media agencies. If a private broadcasting company wants to broadcast its programme, it has to acquire a broadcasting licence. This licence can only be granted by the responsible media agency which is acting as provider, administrator and distributor of usage rights to broadcasting frequencies (Heinrich 2002, 199; Puppis 2007, 65). In 1984 three means of transmitting broadcasting signals existed: cable network, terrestrial transmission via antennas and satellite transmission. Despite the poor development of the German cable network until 1982, the transmission of programmes via cable networks proved to be profitable, especially in terms of a combination with satellite technology, which indeed was not yet fully developed. Anyhow, the televisor satellite ECS 1 served as a transponder for the transmission of broadcasting signals to central cable network stations, which were able to feed the programmes into nationwide cable networks. A licence that was

Clusters and spatial systems in German TV production

61

granted for the transmission of broadcasts on ECS 1 would enable the broadcasters to reach an increasing number of households with their programmes. But at first the terrestrial frequencies were of the main interest. Nearly every household possessed antennas, and this mode of transmitting broadcasts had the widest range (Beck 2005, 178). A wide range was very important to the private broadcasters because by this means they were able to charge higher prices for commercials, which was the only way to achieve the aim of maximizing profits (Heinrich 2002, 97). That is why private broadcasters tried to obtain licences for terrestrial, cable network or satellite frequencies. The cradle of private German broadcasting stood in Rhineland-Palatinate (Reese 1989, 375). On 1 January 1984, a pilot project for a cable network by also feeding in private broadcasts was started in Ludwigshafen. This pilot project was supervised by the RhinelandPalatinate media agency ‘Anstalt für Kabelkommunikation’ (AKK), which was located in Ludwigshafen. Pilot projects for cable networks were to serve for testing models of a future media landscape with simultaneous scientific observation (Eifert and Hoffmann-Riem 1999, 56). The private broadcaster PKS was also licensed for transmitting its programmes inside the local cable network of Ludwigshafen’s pilot project. However, in order to be granted this licence PKS had to relocate to Rhineland-Palatinate and because of technical considerations to Ludwigshafen: and then, of course it was like this: They said ‘Broadcasters that want to participate in the pilot project should be resident in Rhineland-Palatinate’. We did not want to grant borrowed licences. That is why it developed here the way it did, and the AKK as a media agency in Ludwigshafen did the technical handling and then it appeared natural to settle in a place next to the point of programme feed-in to the cable network. It would have been quite expensive if we had had to carry the signal from somewhere else in Germany. That is why they firstly settled in Ludwigshafen. (Interview 134; own translation)

AKK was also responsible for the feed-in of satellite broadcasting into cable networks (Reese 1989, 378) and was therefore authorized to grant a Europe-wide frequency on satellite ECS 1. Under the direction of PKS a consortium of companies applied for a licence for broadcasting on ECS 1’s TV channel. This licence was granted on 1 April 1984 (Vennebusch 1998, 80 f.). As from 1 January 1985, PKS was officially on the air under its new name ‘Sat.1’. In advance a relocation was necessary. This relocation was carried out from Ludwigshafen to Mainz and was mainly politically justified: Considering the political facts, only a relocation inside Rhineland-Palatinate was possible for us due to our Rhineland-Palatinate licence and the strong

62

Emerging clusters position of the Rhineland-Palatinate prime minister.4 We were quite happy to leave Ludwigshafen and relocate to Mainz. We felt quite important in Mainz because in Ludwigshafen we only had a small studio, a small administrative building between the cemetery and the slaughterhouse on the city’s periphery, which was quite difficult. (Interview 86; own translation)

However, a crucial factor for Sat.1’s decision to relocate to Mainz was also the availability of existing facilities like infrastructure and thus favourable production conditions due to the presence of other TV stations in Mainz. The following quotation underlines that the presence of broadcasters under public law actually attracted private broadcasters: Well, Ludwigshafen was difficult in terms of logistics, there were logistic problems, the place did not have any connections to the media and there were not any production plants and additional suppliers or contractors. It was difficult to find employees and so on. So we were proud to be able to argue: ‘Well, let us leave Ludwigshafen and move to the capital of Rhineland Palatinate, to Mainz’. And of course we hoped to find helpful facilities of infrastructure in Mainz due to the presence of ZDF, HR and SWR (formerly SWF). (Interview 86; own translation)

In order to enable more people to receive its programmes also via terrestrial antennas and not only via satellites (which provided the chance of feeding the Sat.1 programme into other cable networks), Sat.1 step by step applied for terrestrial broadcasting licences in each federal state. The media agencies on their part granted licences by demanding in return the broadcasting of regional information programmes within the private broadcaster’s programme. Moreover, they demanded a regional commitment on the part of private broadcasters. The following quotation underlines these processes and shows how Sat.1 got a terrestrial broadcasting licence in Berlin in August 1987: the federal state’s media agencies were responsible for the allocation of antenna frequencies. Sometimes only one broadcasting chain existed and there were always two main competitors, namely RTL and Sat.1. Then sheer political haggling followed. Berlin’s licence was about millions of viewers. (. . .) For example, in Berlin there was a night meeting. RTL and I . . . well, it was like a sale at an auction . . . in the morning at 3.30 I agreed to found also a Sat.1-production unit in Berlin. RTL . . . those folks had gone to bed and were not able to catch up and next morning at eight o’clock I got the licence. Actually it is unthinkable today how things happened at that time. Like I said: It was media policy, media policy, media policy, and that was tantamount to locational policy and was no economic policy. (Interview 86; own translation)

In this manner the ‘frequency rallye’ (Hanfeld 2003) of private broadcasters went on. The practice of granting licences as experienced for

Clusters and spatial systems in German TV production

63

example by Sat.1 shows that the allocation of licences was partly adventurous. It happened in informal bargaining systems and behind closed doors between politicians and the managers of broadcasting companies. Likewise it is obvious that the choices of location made by many private German broadcasters were considerably influenced by the guidelines of economic policy, which depended on the framework of system policy concerning the regulation of broadcasting. Concretely, the measures of structural policy of each federal state were carried out in the shape of location policy. Ambitious efforts of location policy can also be found during the phase of the foundation and licensing of other private broadcasters. The Luxembourgian broadcaster RTL Plus, which initially broadcast its German programme from Luxembourg, decided immediately after the first broadcasting law of North-Rhine Westphalia came into effect to relocate its headquarters to Germany. RTL did so because it did not want to be at a disadvantage regarding the allocation of North-Rhine Westphalian licences. These licences were preferentially granted to aspirants whose studios and production units were located in North-Rhine Westphalia. At the same time aspirants had to guarantee that they would buy the programmes produced by two North-Rhine Westphalian production firms (Nünning 2003, 106). The concrete locational decision of RTL in favour of Cologne is additionally due to the initiatives of Cologne’s municipal savings bank, which offered apartments, buildings and a comprehensive infrastructure (Geschwandtner-Andreß 1999, 47). At the same time North-Rhine Westphalian politicians and especially Prime Minister Johannes Rau and his chief of the state chancellery, Wolfgang Clement, promised to support the companies in the media scene in order to cope with structural change. They decided to provide generous funds to these new lines of business. Moreover, in Cologne the locational decisions of many private broadcasters were facilitated on account of a TV production infrastructure which already existed to a certain extent and had been created by the WDR, the broadcaster under public law. In Cologne, the WDR had been required by the city council already in the mid-1970s to abandon its vertically integrated production system in favour of increasingly placing orders with private service providers. This had a very positive influence on Cologne’s TV production landscape and very early attracted competent external service contractors to Cologne. This existing production infrastructure had a very positive effect on the locational decisions of many private broadcasters in favour of the North-Rhine Westphalian city of Cologne (Voelzkow 2002, 142). In other federal states the media agencies also linked the granting of licences to considerations of regional economic policy, for instance in Bavaria by reorganizing the private broadcaster

64

Emerging clusters

Tele 5, which became DSF, and by relocating it to Munich (1993), or by licensing Home Order Television 1995 (Nünning 2003, 82 ff.). Private broadcasters were players in the broadcasting market who were in urgent need of telecasts but did not possess sufficient capacity to produce these telecasts. That is why they established the principle of placing orders with independent production firms and therefore strong vertically desintegrated production systems (Mossig 2004, 256 f.), which caused an increase in regional demand for independently produced TV programme contents and therefore created opportunities for the foundation of new firms, for relocation and spin-off processes, which evidently took place in Cologne. There, the private broadcaster RTL particularly generated a strong regional demand for independently produced telecasts. This demand was conducive to converting the WDR-specific regional leadership into successful processes of cluster formation by causing the foundation or relocation of new firms and especially by spin-off processes (Voelzkow 2002; Beier 2003; Mossig 2006, 113 f.). Comparable developments could not be observed in the city of Mainz. Here, neither broadcasters under public law (ZDF, SWF/SWR) nor the private broadcaster Sat.1 became incubators and therefore failed to create seedbeds for successful processes of cluster formation. These facts are again due to certain measures of broadcasting policy (Dorenkamp 2008).

5

CONCLUSION

The descriptions and explanations in this chapter have indicated that the analysis of the evolution of processes of cluster formation and spatial patterns of industrial activities has to consider more carefully the role played by decisions relating to economic policy. The example of the German TV industry shows how decisions in system policy about regulating industrial activities can lead to politically influenced choices of industrial location and subsequent clustering processes. It became apparent that this spatial influence can take effect on the one hand if decisions of system policy on regulation create market entry barriers which lead to the politically motivated foundation of monopolistic companies under public law whose choices on location are subsequently made by informal bargaining inside political bodies and whose systems of production are established as a result of political decisions. On the other hand the analysis of the private broadcasters’ choice of locations showed that decisions of system policy on regulation create a considerable scope for measures of structural policy in the shape of location policy, if political authorities are entitled as a result of decisions on regulation to have powerful resources like licences,

Clusters and spatial systems in German TV production

65

which are necessary for private companies that intend to enter a special market. Concretely, in German broadcasting regulatory mechanisms of system policy manifested themselves in the shape of the control of market entry. The federal states’ broadcasters under public law were founded and located by the political directives of the Allies and of German politicians. At the same time they were established as vertically integrated and governmentally protected monopolists (Voelzkow 2002, 141). For these broadcasters market exit was never possible. However, their funding was always secure. That is why they were able to concentrate funds, qualified employees and technical know-how on their locations. Thus, the broadcasters under public law acted as market pioneers and built up strong regional production and distribution systems at their locations and therefore gained regional leadership. However, these production systems mainly remained in-house and were operating by the use of companyowned facilities. Nevertheless external service providers or producers could benefit from the presence of these broadcasters if broadcasters under public law had to abandon their vertically integrated production system in favour of placing orders with private firms. Broadcasters under public law acting as monopolistic market pioneers were therefore able to create successful seedbeds for subsequent clustering processes at their locations. Their fertility increased by the broadcaster’s growing financial power and the volume of orders placed with private companies. Thus, the market pioneers were able to provide regional first-mover advantages for their locations. This regional first-mover advantage later influenced the locational decisions of private broadcasters. After 1984, private broadcasters always strove for good contacts (‘political resources’, Frynas et al. 2006, 321) with the media agencies of the federal states. By doing this they tried to recommend themselves as candidates for the allocation of scarce broadcasting frequencies in order to enter the new market first. The media agencies of the federal states, equipped with powerful broadcasting licences, practised location policy in favour of their own federal state by working together with economic politicians who created financial incentives in terms of promises of economic support. Concrete locational decisions of private broadcasters were on the one hand based on the political demands of regional economic politicians who linked the granting of licences directly to a company’s promise to settle in the ‘right’ location. On the other hand the companies’ decisions were shaped by the entrepreneurial necessity of finding qualified personnel and an existing infrastructure for producing telecasts in proximity to their own headquarters. In fact, qualified personnel could only be found at locations with broadcasters under public law. The infrastructure for producing

66

Emerging clusters

telecasts could also only develop at these locations. If broadcasters under public law made contracts with external service providers, they created, acting as market pioneers, a local demand for externally produced programmes. In addition, their qualified personnel served as a resource of talents that could be attractive for private broadcasters looking for such personnel. That is why Germany’s most important private broadcasters RTL and Sat.1 located in Cologne and Mainz, respectively. RTL was able to build most of its production units in Cologne, and on account of its increasing demand for external programme productions in Cologne caused a boom in the production of TV programmes. This boom is reflected in local spin-off-processes and culminated in a successful cluster formation based on the seedbed created by RTL (Dorenkamp and Mossig 2006). By contrast, Sat.1 failed to create such seedbeds in Mainz and therefore did not initiate a successful cluster formation. Processes of selective cluster formation in the German TV industry therefore mainly depend on the presence and regional activities of private broadcasters. Altogether these explanations have shown how measures on the part of economic policy created and durably shaped the entire spatial pattern of a regulated industry in Germany. However, these political processes have been neglected in most studies in economic geography on the evolution of spatial concentrations of industrial activities. Surely this is due to the frequently informal bargaining processes between politicians and other parties which often lead to entrepreneurial choices of location but which are normally not documented or written down. Therefore it is difficult to gain access to these processes. However, the analysis of other lines of business as well as recent developments show that spatial effects of economic policy on cluster formation are surely not restricted to the German TV industry. Decisions in system policy on the regulation of industrial activities also occur in other sectors of the economy, for example in the postal service, in the telecommunication and energy industries, railway service and the banking sector as well as in multimedia and internet services (Weber 1986; Heinrich 2001, 91; Puppis 2007, 169 ff.). But especially the global financial meltdown since autumn 2008 and its impact on the whole economic order point to the outstanding role of national, European and global economic policy for companies, their performance and their decisions on the choice of location. At present, many nations which are affected by an economic downturn make an effort to subject the financial and capital markets to more effective regulation and to preserve individual companies or whole lines of business from bankruptcy by governmental declarations of surety or even by partial nationalization. Moreover, these nations take measures of wage policy, monetary policy and fiscal policy and introduce macropolitical economic stimulus packages in order to give

Clusters and spatial systems in German TV production

67

their companies better financial room for manoeuvre. We can assume that these measures of economic policy will lead to strong spatial effects, the consequences of which cannot be foreseen. But also in the absence of an economic crisis the economic policy framework exists. That is why economic geography will have to attach more importance to the supportive and/or restrictive influence of the economic policy framework when analysing the evolution of clusters and spatial patterns of industrial activity.

NOTES 1. Fromhold-Eisebith and Eisebith (2005, 1252) call this understanding of cluster policy as a state-run policy carried out by political authorities ‘explicit top-down initiatives’ and contrast this understanding of cluster promotion with ‘implicit bottom-up initiatives’, which are mainly employed by private groups. 2. Like competition policy (Wettbewerbspolitik), regulation policy (Regulierungspolitik) is a subdomain of system policy (Welfens 2008, 489). 3. The channels of SWR’s business are located in Stuttgart, Baden-Baden and Mainz. The entire income of this broadcaster is divided equally between the three locations. 4. The Rhineland-Palatinate prime minister traditionally holds the chairmanship of the federal states’ broadcasting commission. This commission acts as a discussion forum for a collective media policy and as a decision-making body. Decisions made in this commission are presented to the federal states’ parliaments and governments for assent.

BIBLIOGRAPHY Agarwal, Rajshree and M. Gort (2001), ‘First-mover Advantage and the Speed of Competitive Entry, 1887–1986’, Journal of Law and Economics 44 (1), 161–77. Ahedo, Manu (2004), ‘Cluster Policy in the Basque Country (1991–2002). Constructing Regional “Industry-Government” Collaboration Through Cluster-Associations’, European Planning Studies 12 (8), 1097–113. Alecke, Björn, C. Alsleben, F. Scharr and G. Untiedt (2008), ‘Geographic Concentration of Sectors in the German Economy: Some Unpleasant Macroeconomic Evidence for Regional Cluster Policy’, in U. Blien and Gunther Maier (eds), The Economics of Regional Clusters: Networks, Technology and Policy (New Horizons in Regional Science), Cheltenham, UK and Northampton, MA, USA: Edward Elgar, 209–33. Altendorfer, Otto (2004), Das Mediensystem der Bundesrepublik Deutschland, Band 2, Wiesbaden: VS Verlag für Sozialwissenschaften. Aziz, Kamarulzaman and M. Norhashim (2008), ‘Cluster-Based Policy Making: Assessing Performance and Sustaining Competitiveness’, Review of Policy Research 25 (4), 349–75. Bathelt, Harald (1991), Schlüsseltechnologie-Industrien. Standortverhalten und Einfluss auf den regionalen Strukturwandel in den USA und in Kanada, Berlin, Heidelberg and New York: Springer. Bathelt, Harald and J. Glückler (2002), Wirtschaftsgeographie. Ökonomische Beziehungen in räumlicher Perspektive, Stuttgart: Verlag Eugen Ulmer.

68

Emerging clusters

Bathelt, Harald and U. Dewald (2008), ‘Ansatzpunkte einer relationalen Regionalpolitik und Clusterförderung’, Zeitschrift für Wirtschaftsgeographie 52 (2–3), 163–79. Bausch, Hans (1980), Rundfunkpolitik nach 1945. Erster Teil: 1945–1962, Munich: Deutscher Taschenbuch-Verlag. Beck, Hanno (2005), Medienökonomie. Print, Fernsehen und Multimedia, Heidelberg: Springer. Beier, Oliver (2003), Spin-off-Gründungen im Kölner Mediencluster, unpublished diploma thesis at University of Cologne, Cologne. Blien, Uwe and G. Maier (2008), ‘The Starting Point’, in U. Blien and Gunther Maier (eds), The Economics of Regional Clusters: Networks, Technology and Policy (New Horizons in Regional Science), Cheltenham, UK and Northampton, MA, USA: Edward Elgar, 1–11. Bohlmann, Jonathan D., P.N. Golder and D. Mitra (2002), ‘Deconstructing the Pioneer’s Advantage: Examining Vintage Effects and Consumer Valuations of Quality and Variety’, Management Science 48 (9), 1175–95. Boschma, Ron A. and K. Frenken (2003), ‘Evolutionary Economics and Industry Location’, Jahrbuch für Regionalwissenschaft 23, 183–200. Boschma, Ron A. and K. Frenken (2006), ‘Why Is Economic Geography Not an Evolutionary Science? Towards an Evolutionary Economic Geography’, Journal of Economic Geography 6 (3), 273–302. Boschma, Ron A. and J.G. Lambooy (1999), ‘Evolutionary Economics and Economic Geography’, Journal of Evolutionary Economics 9, 411–29. Boschma, Ron A. and R. Wenting (2005), ‘The Spatial Evolution of the British Automobile Industry’, Papers in Evolutionary Economic Geography 1 (4). Brenner, Thomas and D. Fornahl (2003), ‘Theoretische Erkenntnisse zur Entstehung und Erzeugung branchenspezifischer Cluster’, in K. Dopfer (ed.), Studien zur Evolutorischen Ökonomik VII. Evolutorische Ökonomik–Methodologische, ökonometrische und mathematische Grundlagen, Berlin: Duncker and Humblot, 133–62. Buenstorf, Guido (2007), ‘Opportunity Spin-offs and Necessity-Spin-offs’, Papers on Economics and Evolution, No. 0718. Burfitt, A. and S. Macneill (2008), ‘The Challenges of Pursuing Cluster Policy in the Congested State’, International Journal of Urban and Regional Research 32 (2), 492–505. Burfitt, A., S. Macneill and J. Gibney (2007), ‘The Dilemmas of Operationalizing Cluster Policy: The Medical Technology Cluster in the West Midlands’, European Planning Studies 15 (9), 1273–90. BVerfGE (Bundesverfassungsgerichtsentscheidung) (Federal Constitutional Court decision) (1981), 57, 295. Donges, Patrick (2002), Rundfunkpolitik zwischen Sollen, Wollen und Können. Eine theoretische und komparative Analyse der politischen Steuerung des Rundfunks, Wiesbaden: Westdeutscher Verlag. Donges, Juergen B. and A. Freytag (2004), Allgemeine Wirtschaftspolitik, 2nd Edition, Stuttgart: Lucius & Lucius Verlagsgellschaft. Dorenkamp, Ansgar (2004), Bildung von Medienclustern in Deutschland am Beispiel des Standorts Mainz, unpublished diploma thesis, Gießen. Dorenkamp, Ansgar (2008), Blockierte Clusterbildung–das Beispiel der TV-Branche am Standort Mainz/Wiesbaden, unpublished doctoral thesis, Gießen. Dorenkamp, Ansgar and I. Mossig (2006), ‘Die Gründungsregionen Köln und

Clusters and spatial systems in German TV production

69

Mainz: Zur Rolle von Gründungen im Zuge der Clusterevolution am Beispiel der TV-Branche’, in Rolf Sternberg (ed.), Deutsche Gründungsregionen, Reihe Wirtschaftsgeographie 38, Berlin: LIT-Verlag, 281–308. Eifert, Martin and W. Hoffmann-Riem (1999), ‘Die Entstehung und Ausgestaltung des dualen Rundfunksystems’, in Dietrich Schwarzkopf (ed.), Rundfunkpolitik in Deutschland. Wettbewerb und Öffentlichkeit 1, Munich: Deutscher TaschenbuchVerlag, 50–116. Falck, Oliver, S. Heblich and S. Kipar (2008), ‘The Extensions of Clusters: Difference-in-Difference Evidence from the Bavarian State-Wide Cluster Policy’, CESifo Working Paper, No. 2463, Category 9: Industrial Organisation. Francis, John G. (1993), The Politics of Regulation. A Comparative Perspective, Cambridge, MA: Blackwell. Freundt, Andreas (2003), Entwicklungspotenziale der Kulturwirtschaft in altindustrialisierten Regionen, doctoral thesis at University of Dortmund. Fritze, Ralf (1992), Der Südwestfunk in der Ära Adenauer: die Entwicklung der Rundfunkanstalt von 1949 bis 1965 unter politischem Aspekt, Baden-Baden: Nomos-Verlagsgesellschaft. Fromhold-Eisebith, Martina and G. Eisebith (2005), ‘How to Institutionalize Innovative Clusters? Comparing Explicit Top-Down and Implicit BottomUp Approaches’, Research Policy 34 (8; Special Issue: Regionalization of Innovation Policy), 1250–68. Fromhold-Eisebith Martina and G. Eisebith (2008), ‘Clusterförderung auf dem Prüfstand. Eine kritische Analyse’, Zeitschrift für Wirtschaftsgeographie 52 (2–3), 79–94. Frynas, Jedrzej G., K. Mellahi and G.A. Pigman (2006), ‘First Mover Advantages in International Business and Firm-specific Political Resources’, Strategic Management Journal 27, 321–45. Gellner, Winand and A. Glatzmeier (2004), Macht und Gegenmacht. Einführung in die Regierungslehre, Baden-Baden: Nomos-Verlagsgesellschaft. Geschwandtner-Andreß, Petra (1999), ‘Medienwirtschaft in Köln. Theoretische Erklärungsansätze und politische Bestimmungsfaktoren eines regionalen Produktionsclusters Medien’, Working Papers, Institute for Broadcasting Economics at University of Cologne 116, Cologne. Hanfeld, Michael (2003), ‘20 Jahre Privatfernsehen – Der erste Mann’, Frankfurter Allgemeine Zeitung, 18 December 2003, No. 294, 38. Hayter, Roger (1997), The Dynamics of Industrial Location. The Factory, the Firm and the Production System, Chichester, New York, Weinheim, Brisbane, Singapore and Toronto: John Wiley & Sons. Hegedüs, Veronika (2008), ‘Theoretical Approach to Clusters and Applied Cluster Policy’, Development & Finance (2), 77–87. Heinrich, Jürgen (2001), Medienökonomie. Band 1: Mediensystem, Zeitung, Zeitschrift, Anzeigenblatt, Wiesbaden: Westdeutscher Verlag. Heinrich, Jürgen (2002), Medienökonomie. Band 2: Hörfunk und Fernsehen. Wiesbaden: Westdeutscher Verlag. Hoffmann-Riem, Wolfgang (1986), ‘Internationale Medienmärkte–Nationale Rundfunkordnungen. Anmerkungen zu Entwicklungstendenzen im Medienbereich’, Rundfunk und Fernsehen 34, 5–22. Hoffmann-Riem, Wolfgang (1999), ‘Der strategische Stellenwert der Programmproduktion für die Rundfunkveranstaltung’, in Hermann-Dieter Schröder (ed.), Entwicklung und Perspektiven der Programmindustrie, Symposium

70

Emerging clusters

des Hans-Bredow-Instituts, Band 17, Baden-Baden and Hamburg: NomosVerlagsgesellschaft, 13–41. Hoppe, Heidrun C. and U. Lehmann-Grube (2001), ‘Second-Mover Advantages in Dynamic Quality Competition’, Journal of Economics Management Strategy 10 (3), 419–33. Hospers, Gert-Jan (2005), ‘Best Practices and the Dilemma of Regional Cluster Policy in Europe’, Tijdschrift voor Economische en Sociale Geografie 96 (4), 452–57. Hospers, Gert-Jan and Beugelsdijk, S. (2002), ‘Regional Cluster Policies: Learning by Comparing?’, KYKLOS, International Review for Social Sciences 55 (3), 381–402. Hospers, Gert-Jan, F. Sautet and P. Desrochers (2008), ‘The Next Silicon Valley? On the Relationship between Geographical Clustering and Public Policy’, International Entrepreneurship and Management Journal 4 (Online First™), 1–15. Huff, Lenard C. and William T. Robinson (1994), ‘Note: The Impact of Leadtime and Years of Competitive Rivalry on Pioneer Market Share Advantages’, Management Science 40 (10), 1370–77. Isenberg, Meike (2007), Verhandelte Politik. Informale Elemente in der Medienpolitik, Berlin: VISTAS Verlag. Jarren, Otfried (1997), ‘Rundfunk und Rundfunkregulierung in Deutschland– Probleme, Defizite und Zukunftsaufgaben’, in Heribert Schatz, Otfried Jarren and Bettina Knaup (eds), Machtkonzentration in der Multimediagesellschaft? Beiträge zu einer Neubestimmung des Verhältnisses von politischer und medialer Macht. Wiesbaden: Westdeutscher Verlag, 203–15. Jarren, Otfried (2007), ‘Die Regulierung der öffentlichen Kommunikation. Medienpolitik zwischen Government und Governance’, Zeitschrift für Literaturwissenschaft und Linguistik 146, 131–53. Kalyanaram, Gurumurthy and Glen L. Urban (1992), ‘Dynamic Effects of the Order of Entry on Market Share, Trial Penetration, and Repeat Purchases for Frequently Purchased Consumer Goods’, Marketing Science 11 (3), 235–50. Kerin, Roger A., R. Varadarajan and R.A. Peterson (1992), ‘First-mover Advantage: A Synthesis, Conceptual Framework, and Research Propositions’, Journal of Marketing 58, 33–52. Kiese, Matthias (2008a), ‘Mind the Gap: Regionale Clusterpolitik im Spannungsfeld von Wissenschaft, Politik und Praxis aus der Perspektive der Neuen Politischen Ökonomie’, Zeitschrift für Wirtschaftsgeographie 52 (2–3), 129–45. Kiese, Matthias (2008b), ‘Stand und Perspektiven der regionalen Clusterforschung’, in Matthias Kiese and Ludwig Schätzl (eds), Cluster und Regionalentwicklung. Theorie, Beratung und praktische Umsetzung, Dortmund: Verlag Dorothea Rohn, 9–50. Klepper, Steven (2002), ‘The Capabilities of New Firms and the Evolution of the US Automobile Industry’, Industrial and Corporate Change 11, 645–66. Knieps, Günter (2007), Netzökonomie. Grundlagen–Strategien–Wettbewerbspolitik, Wiesbaden: Gabler Verlag. Knieps, Günter (2008), Wettbewerbsökonomie. Regulierungstheorie, Industrieökonomie, Wettbewerbspolitik, 3rd Edition, Heidelberg: Springer. Koch, Walter A.S., C. Czogalla and M. Ehret (2008), Grundlagen der Wirtschaftspolitik, 3rd Edition, Stuttgart: Lucius & Lucius Verlagsgesellschaft.

Clusters and spatial systems in German TV production

71

Koh, Winston T.H. (1993), ‘First-mover Advantage and Organizational Structure’, Economics Letters 43, 47–52. Kulke, Elmar (2004), Wirtschaftsgeographie, Paderborn, Munich, Vienna and Zurich: Ferdinand Schöningh. Lieberman, Marvin B. and D.B. Montgomery (1988), ‘First-mover Advantages’, Strategic Management Journal 9, 41–58. Lieberman, Marvin B. and D.B. Montgomery (1998), ‘First-mover (Dis)advantages: Retrospective and Link with the Resource-based View’, Strategic Management Journal 19, 1111–25. Luckenbach, Helga (2000), Theoretische Grundlagen der Wirtschaftspolitik, 2nd Edition, Munich: Vahlen. Mai, Michael (2005), Medienpolitik in der Informationsgesellschaft, Wiesbaden: Verlag für Sozialwissenschaften. Mans, Peter, F. Alkemade, T. van der Valk and M.P. Hekkert (2008), ‘Is Cluster Policy Useful for the Energy Sector? Assessing Self-Declared Hydrogen Clusters in the Netherlands’, Energy Policy 36, 1375–85. May, William, C. Mason and S. Pinch (2001), ‘Explaining Industrial Agglomeration: The Case of the British High-Fidelity Industry’, Geoforum 32, 363–76. Meyer-Stamer, Jörg (1999a), ‘Strategien lokaler/regionaler Entwicklung. Standortpolitik und systemische Wettbewerbsfähigkeit’, Nord-Süd aktuell, 3/1999. Meyer-Stamer, Jörg (1999b), ‘Lokale und regionale Standortpolitik– Konzepte und Instrumente jenseits von Industriepolitik und traditioneller Wirtschaftsförderung’, INEF-Report 39/1999. Min, Sungwook, M.U. Kalwani and W.T. Robinson (2006), ‘Market Pioneer and Early Follower Survival Risks: A Contingency Analysis of Really New Versus Incrementally New Product-Markets’, Journal of Marketing 70, 15–33. Morgenpost Online (2008), ‘Senat verzichtet auf Geld, wenn Sat.1 in Berlin bleibt’, available online: http://www.morgenpost.de/berlin/article987737/Senat_verzich tet_auf_Geld_wenn_Sat_1_in_ Berlin_bleibt.html, 31 March 2009. Mossig, Ivo (2000), Räumliche Konzentration der VerpackungsmaschinenbauIndustrie in Westdeutschland. Eine Analyse des Gründungsgeschehens, Wirtschaftsgeographie Band 17, Münster, Hamburg and London: LITVerlag. Mossig, Ivo (2004), ‘Steuerung lokalisierter Projektnetzwerke am Beispiel der Produktion von TV-Sendungen in den Medienclustern München und Köln, Erdkunde 58, 252–68. Mossig, Ivo (2006), Netzwerke der Kulturökonomie. Lokale Knoten und globale Verflechtungen der Film- und Fernsehindustrie in Deutschland und den USA, Bielefeld: transcript Verlag. Mossig, Ivo (2008), ‘Entstehungs- und Wachstumspfade von Clustern. Konzeptionelle Ansätze und empirische Beispiele’, in Matthias Kiese and Ludwig Schätzl (eds), Cluster und Regionalentwicklung: Theorie, Beratung und praktische Umsetzung, Dortmund: Verlag Dorothea Rohn, 51–68. Nünning, Volker (2003), Medienpolitik als Standortpolitik. Vergleichende PolicyAnalyse der Medienpolitik Bayerns und Nordrhein-Westfalens am Beispiel des bundesweiten Privatfernsehens, unpublished diploma thesis at Leibniz University Hanover. Puppis, Manuel (2007), Einführung in die Medienpolitik, Konstanz: UVK Verlagsgesellschaft.

72

Emerging clusters

Raines, Philip (2002), ‘Cluster Policy–Does it Exist?’, in Philip Raines (ed.), Cluster Development and Policy, Aldershot: Ashgate Publishing. 21–33. Reese, Jürgen (1989), ‘Rheinland-Pfalz: Landeszentrale für private Rundfunkveranstalter (LPR)’, in Gerd-Michael Hellstern, Wolfgang Hoffmann-Riem and Jürgen Reese (eds), Rundfunkaufsicht in der Bundesrepublik Deutschland, Band 3: Rundfunkaufsicht in vergleichender Analyse, Paderborn: Bonifatius Verlag, 375–98. Robinson, William T. (1988), ‘Sources of Market Pioneer Advantages: The Case of Industrial Goods Industries’, Journal of Marketing Research 25, 87–94. Robinson, William T. and C. Fornell (1985), ‘Sources of Market Pioneer Advantages in Consumer Goods Industries’, Journal of Marketing Research 22, 305–17. Robinson, William T., G. Kalyanaram and G.L. Urban (1994), ‘First-mover Advantages from Pioneering New Markets: A Survey of Empirical Evidence’, Review of Industrial Organization 9, 1–23. Shaffer, Brian (1995), ‘Firm-level Responses to Government Regulation: Theoretical and Research Approaches’, Journal of Management 21 (3), 495–514. Shankar, Vishwakarma, G.S. Carpenter and L. Krishnamurthi (1998), ‘Later Mover Advantage: How Innovative Late Entrants Outsell Pioneers’, Journal of Marketing Research 35, 54–70. Schätzl, Ludwig (1994), Wirtschaftsgeographie 3–Politik, Paderborn: Ferdinand Schöningh. Schultze, Rainer-Olaf (2005), ‘Staat’, in Dieter Nohlen and Rainer-Olaf Schultze (eds), Lexikon der Politikwissenschaft. Theorien, Methoden, Begriffe, Band 2 (N–Z), Munich: C. H. Beck, 944–6. Spiegel Online (2008), ‘ProSiebenSat.1 legt radikales Sparprogramm auf’, available online: http://www.spiegel.de/wirtschaft/0,1518,590305,00.html, 31 March 2009. Storper, Michael and R. Walker (1989), The Capitalist Imperative. Territory, Technology, and Industrial Growth, Oxford, Cambridge, MA: WileyBlackwell. Stuiber, Heinz-Werner (1998), Medien in Deutschland. Band 2: Rundfunk. Teil 1: Zum Rundfunkbegriff, Rundfunktechnik, Geschichte des Rundfunks, Rundfunkrecht, Konstanz: UVK Verlagsellschaft. Sydow, Jörg and U. Staber (2002), ‘The Institutional Embeddedness of Project Networks: the Case of Content Production in German Television’, Regional Studies 36 (3), 215–27. Urban, Glen L., T. Carter, S. Gaskin and Z. Mucha (1986), ‘Market Share Rewards to Pioneering Brands: An Empirical Analysis and Strategic Implications’, Management Science 32 (6), 645–59. VanderWerf, Pieter A. and John F. Mahon (1997), ‘Meta-Analysis of the Impact of Research Methods on Findings of First-Mover Advantage’, Management Science 43 (11), 1510–19. Vennebusch, Angela (1998), Die Neugliederung der deutschen Fernsehlandschaft, Frankfurt am Main, Berlin, New York, Paris, Wien: Verlag Peter Lang. Voelzkow, Helmut (2002), ‘Die “neue Kultur der Selbständigkeit” und ihr institutionelles Umfeld: Erfahrungen aus der Medienwirtschaft Köln’, in Rolf G. Heinze and Frank Schulte (eds), Unternehmensgründungen. Zwischen Inszenierung, Anspruch und Realität, Wiesbaden: Westdeutscher Verlag, 130–48. Von Einem, Eberhard (1991), ‘Industriepolitik: Anmerkungen zu einem

Clusters and spatial systems in German TV production

73

kontroversen Begriff’, in Ulrich Jürgens and Wolfgang Krumbein (eds), Industriepolitische Strategien, Berlin: Rainer Bohn Verlag, 11–33. Vowe, Gerhard (2007), ‘Ordnung durch Medienpolitik und der Beitrag der Wissenschaft–das Beispiel Deutschland’, in Otfried Jarren and Patrick Donges (eds), Ordnung durch Medienpolitik?, Konstanz: UVK Verlagsgesellschaft, 71–82. Weber, Rolf H. (1986), ‘Wirtschaftsregulierung in wettbewerbspolitischen Ausnahmebereichen’, in Ernst-Joachim Mestmäcker (ed.), Wirtschaftsrecht und Wirtschaftspolitik, Band 86, Baden-Baden: Nomos-Verlagsgesellschaft. Welfens, Paul J.J. (2008), Grundlagen der Wirtschaftspolitik. Institutionen– Makroökonomik–Politikkonzepte, 3rd Edition, Berlin and Heidelberg: Springer. Wildmann, Lothar (2007), Wirtschaftspolitik. Module der Volkswirtschaftslehre, Band 3, Munich and Vienna: Oldenbourg Wissenschaftsverlag. Wilke, Jürgen (2003), ‘Kommunikations- und Mediengeschichte’, in Günter Bentele, Hans-Bernd Brosius and Otfried Jarren (eds), Öffentliche Kommunikation. Handbuch Kommunikations- und Medienwissenschaft, Wiesbaden: Westdeuscher Verlag, 151–68. Witt, Ulrich (1997), ‘Lock-In’ vs ‘Critical Masses’ – Industrial Change Under Network Externalities’, International Journal of Industrial Organization 15, 753–73.

4.

Bridging ruptures: the re-emergence of the Antwerp diamond district after World War II and the role of strategic action Sebastian Henn and Eric Laureys*

1

INTRODUCTION

‘Diamonds love Antwerp’ – these three words constitute the present slogan of the Antwerp World Diamond Centre, the organization concerned with the promotion of the local diamond sector and with keeping the Scheldt city an important hub for trading and manufacturing the precious stones in the age of globalization with competing centres evolving especially in low-cost countries like China and Thailand (The New York Times, 31 May 2005; Even-Zohar 2006). In fact, diamonds seem to have loved the Antwerp region even in the past as this part of Belgium has been a major centre for trading and polishing them since the 15th century (Walgrave 1993). Despite some ups and downs, there had not been any interruption of the commercial activities in this sector until World War II reached the country and trading as well as processing of the stones gradually were discontinued (Laureys 2005, chapter 5f.). After 1945, however, the Belgian diamond sector experienced a long-lasting boom which contributed significantly to the country’s economic power. This is surprising as the former infrastructure had partly been taken away or destroyed, many workers had fled, been deported or killed and promising diamond centres had evolved during the years of the German occupation (van Dyck 1989). Against this background, this chapter aims at analysing the factors which led to the re-emergence of the cluster at its former location. For this purpose it is structured as follows: first, the theoretical framework will be outlined showing that the development of clusters can be regarded as a path-dependent process. In this context the so-called window of locational opportunity approach will be introduced as a central concept of cluster development incorporating the idea of path dependency. Though it

74

The re-emergence of the Antwerp diamond district

75

provides a sound theoretical basis for the evolution of clusters (section 2), the insufficient conceptualization of strategic measures will be considered a major shortcoming the more so as purposeful action is important both for the creation and the continuation of a trajectory (section 3). Having given some details on the research design and methodology (section 4), the trajectory of the diamond district in Antwerp will be reconstructed (section 5) with special attention being paid to those factors which brought about the break of the path (section 6). Moreover, it will be shown that the war years led to the development of still existing growth centres at different locations – an aspect which to some extent resembles the stage of ‘shifting centres’ outlined by Storper and Walker (1989) (section 7). Finally, the reemergence of the former cluster structures will be discussed as an outcome of strategic actions and chance aspects, both of them limiting the degree of openness to the window of locational opportunity which existed at this time (section 8). The chapter ends by summing up the most important results and shedding light on further research questions (section 9).

2

CLUSTERS AS A PATH-DEPENDENT PHENOMENON

Whilst for a long time clusters were treated as quasi-stable entities exhibiting certain positive impacts on their regional economic environment in terms of increases in local productivity, entrepreneurial activities and corporate innovativeness (Cooke 2001), the research focus gradually has shifted towards a more dynamic view analysing the alterations of certain cluster characteristics (for example employment structure, technological diversity, network structure, adaptability) in the course of time (Dalum et al. 2005; Press 2006; Henn 2008). Amongst the different concepts addressing this aspect, evolutionary approaches recently have attracted great interest (Braunherhjelm and Feldman 2006). Most of them refer to the concept of path dependency which has its earliest precursors in the 19th century (Martin and Sunley 2006) and was re-introduced into modern economic thinking especially by David (1985) and Arthur (1994). The key characteristic of a path-dependent process is that its outcomes arise from its own history. More generally speaking, the long-term development of a system is considered to depend on its point of departure and on interferences occurring during its history1 with even minor accidents assumed to exert great influence on the course of development (so-called trajectory). Following this line of reasoning, different approaches aim at explaining the emergence of regional clusters as specific spatial and industrial configurations from an evolutionary point of view: stochastic concepts

76

Emerging clusters

(see, for example, Arthur 1994) making up one important conceptual subgroup seek to analyse the evolution of economic agglomerations on the basis of mathematical models (for example polya urn experiments). As this approach, however, is characterized by several major shortcomings2 (for a criticism see Bathelt and Boggs 2005; Boschma 2007), the so-called window-of-locational-opportunity concept (Storper and Walker 1989; Boschma 1997) has attracted much more interest so far. Its proponents argue that the requirements of a new industry cannot be met at any location since they are far too innovative and therefore yet have to be provided by the companies themselves (for example by cooperating with other firms). As there are no specific regions offering such conditions there is great freedom when choosing the location, usually referred to as ‘window of locational opportunity’ (Storper and Walker 1989, p. 74).3 Once established, the development of the sites will vary since only a few of them will develop into growth centres. Others, on the contrary, will stagnate, grow only slowly or even decline. In other words: by and by companies in certain locations will gain competitive advantages over their competitors due to different positive feedbacks. As soon as the latter exceed a certain threshold, the window of locational opportunity closes (Storper and Walker 1989) and selective growth patterns dominate. Moreover, during their life-cycle most industries tend to establish remote locations (growth peripheries). Depending on whether this happens for reasons of expansion or cost reduction, the former locations typically will be kept or closed. As an industrial development path finally comes to an end, a reorganization of the respective industry is likely to take place leading to new windows of locational opportunity which favour the dislocation of the incumbents (stage of shifting centres). In summary, it can be said that for explaining the early formation of a cluster the model implicitly refers to two main factors: First, ‘human agency’ (Boschma 1997, p. 15) as the ability of actors to shape local structures which hardly provide a stimulating environment in order to generate growth conditions for the new companies. Second, agglomeration economies which set in once a ‘critical mass’ of companies has been reached at a location. While in the past many studies have focused on the latter aspect (see, for example, Press 2006; Buenstorf and Klepper 2009), the first one was neither conceptualized nor illustrated in more detail. In contrast, the results of intentional action of single actors often were regarded as accidential incidents, so-called ‘small events’ (Schamp 2000, p. 43). This view, however, has several shortcomings. First, attributing the emergence of a cluster to chance must be considered undersocialized as the social processes underlying its emergence are not regarded sufficiently (Martin and Sunley 2006). Second, referring to chance as an important explanatory

The re-emergence of the Antwerp diamond district

77

factor in general bears the risk of failing ‘to explain why some apparently trivial events in some places are selected and become significant, whilst other similar chance events in other locations do not trigger the birth of new industrial trajectories’ (Martin and Sunley 2006, p. 425). Third, chance does not constitute a basis for policy advices which is not unproblematic in a discipline increasingly aiming at shaping the policy debate (Martin 2001; Menzel 2008). The role of chance during the emergence of clusters, however, should not completely be ruled out here. Rather chance can be said to be of indirect importance as it is able to modify the scope of action of the local players (Hoffmann 2005).4

3

STRATEGIC ACTION AND PATH CREATION

With reference to the evolution of technological paths, Garud and Karnøe (2001) were the first to discuss the role of strategic action from a conceptual perspective. According to them, ‘path dependence assigns too much weight to history; it inadequately characterizes the fragility of any path as it is produced and reproduced through micro level practices where social rules and artifacts are enacted’ (p. 8). Based on this view, the authors developed the concept of path creation which departs from the path dependency perspective in two main aspects. First, Garud and Karnøe (2001) argue that ‘entrepreneurs meaningfully navigate a flow of events even as they constitute them . . . entrepreneurs attempt to shape paths in real time, by setting processes in motion that actively shape emerging social practices and artifacts, only some of which may result in the creation of a new technological field’ (p. 3). Second, they highlight the process of ‘mindful deviation’ implying that entrepreneurs often have to change existing social practices, regulations or institutions. According to this line of reasoning, path creation highlights ‘the active role of the entrepreneur and the firm, for it is these actors that help shape the evolution of markets and the rules by which markets operate’ (Stack and Gartland 2005, p. 421). Regarding the relation between both concepts, Puffert (2000) stresses that purposeful action should not be regarded as an alternative to path dependency, as the latter would make actors even more interested in starting a new path based on their technologies and techniques. Furthermore, according to Meyer and Schubert (2007), the driving force behind emerging paths should be situated on a continuum between unplanned processes on the one hand and intentionally controlled action on the other hand. As the concept of path creation is introduced into geography, local actors (entrepreneurs, policy makers) can be regarded as being able to exert influence on the emergence of development paths (for example by

78

Emerging clusters

mobilizing resources) (Martin and Sunley 2006; Hassink 2007). Following Martin and Sunley (2006) two kinds of mindful deviation can be distinguished in this context. First, there is an incremental form of path creation, implying that actors actively search for new solutions, that is processes, products, technologies and so on. Second, there may be ‘critical junctures and large-scale events or shocks’ (ibid.) making agents shape new strategies. Examples of the latter aspect are ruptures of cluster trajectories caused for instance by wars, natural catastrophes or other historical events (Porter 1990) with the cluster infrastructure being destroyed or the relevant actors intentionally or unintentionally hindered in carrying out their businesses. It can be hypothesized that strategic action in these cases is of particular relevance for several reasons. First, based on past experiences actors formerly associated to the sector might be willing to use their power to reinvent the cluster structures for different reasons (for example personal motives). Second, if the sector had performed well once, regional policy might be interested in redeveloping it by providing adequate support. Third, depending on the kind and duration of the rupture, some parts of a local knowledge-base or infrastructure might still exist and in principle facilitate a comparatively rapid growth of the sector.

4

RESEARCH DESIGN AND METHODOLOGY

To empirically analyse the role of strategic action in a re-emerging cluster, the diamond district in Antwerp was chosen as an example. Unlike the majority of empirical work on clusters the study thereby primarily makes use of historical methods: In detail, one of the authors conducted searches in 34 different archives (amongst them the Archives Nationales in France, the Bundesarchiv in Germany and the National Archives and Records Administration in the US) throughout the years 1997–2003 in order to analyse the development of the diamond sector during the Nazi rule. The results of these inquiries also provide the basis for several publications (Laureys 2003a, 2003b, 2005). In addition, an in-depth study in the Antwerp City Archive (Stadsarchief Antwerpen) carried out in 2008 aimed at explaining the developments in the post-war era. More than 70 000 datasets of the registers van aankomst en vertrek (Registers of arrival and departure) covering the period 1948–65 were analysed to shed light on the immigration of diamond people who contributed to the stabilization of the resuscitated district. In doing so, a total of 1135 diamond people could be identified as having migrated to Antwerp in the abovementioned period. Furthermore, the archive was searched for files documenting the re-emergence of the local diamond industry. Relevant

The re-emergence of the Antwerp diamond district

79

dossiers could not only be found in the modern archive (Modern Archief) but also in the Camille Huysmans mayor office archives (both belonging to the Antwerp City Archive) where the correspondence of the former mayor of Antwerp is to be found. Since Jews play an important role in the history of the district (see section 5), it was important to consider their role during the rupture and the re-emergence of the structures as well. In this context, the authors were kindly provided with a list of all registered Belgian Jews dated as of 1940 as well as with the lists of deportation to the concentration camps by the Jewish Museum of Deportation and Resistance in Mechelen. Given the kind permission of its owner, it was furthermore possible to make use of the Brachfeld Archive thus gaining valuable information on Jews in the diamond trade. Finally, important insights into the rebuilding of the Jewish community and the cluster itself also could be gained by several interviews with (former) diamantaires from Antwerp or other people related to the Belgian diamond sector.

5

THE TRAJECTORY OF THE DIAMOND DISTRICT UNTIL 1940

The history of the diamond sector in Antwerp dates back to the 15th century (Walgrave 1993). The reasons for its location in this harbour are to be found in an early chapter of globalization. In the 14th century, Bruges traded Indian diamonds with Venice. With the Portuguese taking over the trade routes, however, Venice was replaced by Lisbon as the main European trade centre. Later, as Bruges’ sea access silted up during the 16th century, the trade between the emerging commercial centre Antwerp and Lisbon started to rule out the connection between Venice and Bruges. While for a long time the economic activities in the sector remained on a comparatively low level, the discovery of South-African diamond deposits in the 19th century provoked a massive influx of rough diamonds allowing for the employment of thousands of workers. In the aftermath of this event, a significant strengthening of the institutional basis set in: between 1887 and 1907 three unions were founded, and as a countermeasure the ‘Syndicate of the Belgian diamond industry’ (Syndicaat der Belgische Diamantnijverheid), an employers’ federation, was established in 1927. Moreover, the Diamantclub (established in 1893), the Beurs voor Diamanthandel (1904), the Diamantkring (1929), the Fortunia (1910) and the Vereniging voor Vrije Diamanthandel (1911) were founded as so-called diamond bourses allowing the trade of diamonds to be carried out in an adequate manner (Kinsbergen 1984; Laureys 2005).5

80

Emerging clusters

Furthermore, in the 1930s different vocational schools with special foci on polishing were established for apprentices and unemployed people in Antwerp and its surroundings (Laureys 2005). Last but not least, the sector was granted credit through specialized diamond banks such as the Comptoir Diamantaire Anversois SA6 and the Amsterdamsche Bank voor Belgie in 1934, both of them having an exclusive focus on the diamond sector (Laureys 2005; ADB 2008). The continuous progress of the industry since the waning years of the 19th century is reflected in the growing number of employees (see Figure 4.1, section 8.2). In 1908, there were around 66 working benches with 1184 polishing wheels and around 1500 employees in the district. During World War I the supply of rough stones was interrupted due to the occupation of the city. In 1920, however, the number of diamond workers in the Antwerp region amounted to approximately 20 000 and in 1929 even to about 27 000. The Great Depression of the 1930s resulted in a tangible, albeit temporary impact on the diamond industry. It was not unusual for polishing shops to close for several weeks in these days. In 1937 the number of workers fell to 13 312 while as a result of immigration waves just before the German occupation in 1940 there were about 23–25 000 diamond workers, 2000 manufacturers, 4000 traders and 400 brokers in the district. Due to some overlaps, however, about 15 000 people probably were dependent on the diamond sector as a whole. Antwerp at this time accounted for cutting about 80 per cent of the world’s diamonds (measured by value) (Laureys 2005). From a geographical point of view it should be stressed here that in these days the sector was centred around Antwerp with the strongest concentration in a small area close to the Antwerp main station. While larger stones were cut in the city, too, for different reasons its surroundings (the so-called Kempen region) had become home to the fabricants of smaller stones since the turn of the century (Kinsbergen 1984). Given the spatial concentration of specialized economic actors and activities, the cluster term can be applied when referring to the spatial configuration of the industry. However, when considering the fact that both trade and processing were in the hands of tightly interconnected family-owned businesses the notion of a ‘diamond district’ alluding to the works on industrial districts (Marshall 1920; Markusen 1996) seems even more apt.7 For understanding the further course of the cluster trajectory it has to be mentioned here that the development of the diamond sector has been closely connected with the local Jewish community since its beginnings (Laureys 2005; Vanden Daelen 2006, 2008).8,9 Depending on whether the trade or the processing of diamonds is considered, the presence of the

The re-emergence of the Antwerp diamond district

81

Jews differs significantly: while 80–90 per cent of the bourse members were immigrant Jews, the largest share of the workers (estimated at 80 per cent) was of traditional Flemish origin (Kinsbergen 1984; Laureys 2005). The strong Jewish influence did not remain without consequences for the institutional framework of the sector: For instance, even though second generation immigrants mostly adopted the French language, Yiddish was to have some influence in the diamond district as indicated by the expressions mazl un brokhe (good luck and blessing) which are still used when a deal is closed (Vanden Daelen 2006). Moreover, working rhythms still nowadays are adapted to the Jewish calendar (the luach) and they are acknowledged by non-Jewish traders as well (for example the diamond bourses are still closed on Jewish holidays like Yom Kippur).10

6

THE NAZI RULE AND THE END OF THE TRAJECTORY

The looming war and the invasion of German troops into Belgium in May 1940 brought about some confusion into the local diamond world. While at first there were some uncertainties whether to stay in Belgium or to move abroad, it was finally decided to relocate the industry in Cognac, France.11 In fact, around 5000 people, most of them Jews, moved there (Laureys 2005). For different reasons, however, the establishment of the refugees in France proved difficult. As a consequence, a new escape wave set in resulting in two types of refugees. A first group moved overseas mainly to the USA, Palestine, the UK, South Africa, Brazil and Cuba and started new businesses there (ibid.; Laureys 2003a) (see also section 7). These refugees often benefited from support from relatives and friends of the Jewish diaspora. A second group decided to return to Antwerp, the more so as many of them were persuaded to do so by the German occupants.12 It has to be mentioned, however, that not all diamond people had fled. Indeed, most of the fabricants and workers of Belgian origin had stayed in Belgium hoping that the Germans would allow their activities to continue (Laureys 2005). All diamond people, whether they had stayed in Antwerp or returned from France were soon put under German military control. Shortly after that, several measures taken by the Germans not only led to a reorganization of the sector but also resulted in a complete discontinuation of the commercial activities in 1942 (Laureys 2003b, 2005): First, intermediaries like brokers were excluded because their business was regarded to be clearly Jewish and to contribute to price increases. Second, only people who had been active in the diamond trade for at least ten years were allowed to

82

Table 4.1

Emerging clusters

Members of the Diamantclub (1941–44)

Number of members Number of members abroad

1941

1942

1943

1944

1533 0

699 822

683 1161

339 1159

Source: Own calculations based on data from the Brachfeld Archive (BA 441, balances for the period between 01 January 1941 and 31 December 1946).

continue their activities.13 Third, in 1941 organized raids were carried out in two of the diamond bourses by the Devisenschutzkommando and the Sipo-SD14 (see Vanden Daelen 2008). Fourth, the traders’ stocks were confiscated. Fifth, the Germans removed about 2500 workbenches from the ten largest companies while smaller machinery for about 10 000 workers was left untouched (The New York Times, 7 September 1944; 11 September 1944). Moreover, large numbers of qualified workers were deported to Germany for compulsory labour. The measures by the Nazis culminated in the deportation of 984 diamond people to the concentration camps as the analysis of the deportation reveals. Given the fact that in addition many people died due to enemy actions, it can be assumed that approximately 40 per cent of the city’s diamond dealers and about 15 per cent of the workers were killed during the war (The New York Times, 27 March 1960). The gradual reduction of the trading activities alluded to above is well illustrated by the decrease in the number of bourse members: in 1941, 1533 members were registered at the Diamantclub, the most important diamond bourse in Antwerp in those days; in 1944 only 339 were left. At the same time the number of absent members rose from 0 to 1159 (see Table 4.1). The number of members of the Beurs voor Diamanthandel decreased from 1654 in 1939 to 355 in 1945 (Beurs voor Diamanthandel 1979, 27). Despite missing data it can be assumed that the other bourses (except for Fortunia which did not have any Jewish members) met the same fate. The diamond processing industry also suffered under the German occupation, however, in a rather indirect way: the allies did not want the Nazi regime to take possession of diamonds which were of strategic importance to the war industry and thus had ceased the supply of raw diamonds to foreign countries immediately after the German attack on Poland in 1939 (cf. Jack 1941). Due to the lack of rough stones to be polished, the industry gradually wasted away: from 15 000 workers being active in the sector in 1940, only 2000 workers remained in March 1941 and 1000 in July 1941. A mandatory diamond deposit eventually hailed the end of the industrial activities, too (Laureys 2005).

The re-emergence of the Antwerp diamond district

Table 4.2

83

Number of diamond workers by country before the war and in 1945

USA (without Puerto Rico) South Africa Brazil Cuba Canada UK Puerto Rico Palestine / Israel (since 1948)

Pre-war

1945

350 150 – – – 10 – 200

4000 600 4 2 300 750 1,5 3800

Source: Antwerp City Archive (MA-KAB 1723, letter from Romi Goldmuntz to Camille Huysmans, 25 August 1945 concerning the activities of COFDI).

7

EVOLUTION OF COMPETING CENTRES

Many of the refugees had left Belgium with their stocks trying to resume their activities overseas both in diamond trade and processing. In addition, the rise of the new diamond centres was encouraged by the gradual fading-away of the former main competitor Antwerp due to the reasons mentioned above. As a consequence, the industry had to pass through a development which resembles the stage of ‘shifting centres’ although the underlying reasons for the emergence of the new centres differ from those outlined by Storper and Walker (1989).15 In these years new windows of locational opportunity mainly opened up at two locations (see Table 4.2): in the United States to where many diamond dealers emigrated, the number of diamond workers rose from 350 (pre-war) to 4000 in 1945. Up to 70–80 per cent of the world diamond production was to be found there after the war.16 The share of sales even rose from approximately 75 per cent (pre-war) to 85 per cent in 1948. In spatial terms, New York stood out as the most important centre for both manufacturing and commercial activities. The polishing industry in Palestine (after 1948 Israel) – destination of many Zionist Jews hoping to contribute to the emerging Jewish state – experienced a similar development. Here, the number of enterprises rose from 4 in 1939 to 45 in 1947 while the number of polishers increased from 200 just before World War II to about 3800 after, and to approximately 8000 in 1965 (Szenberg 1973).17 Besides these two countries, the UK gained in importance as a centre for diamond-related activities, too: on the one hand, it became home to some industrial activities, on the other hand,

84

Emerging clusters

the so-called Correspondence Office for the Diamond Industry (COFDI) was established in London by the Antwerp mayor Camille Huysmans and the prominent diamantaires Romi Goldmuntz and Herman Schamisso in October 1940 to support the former Antwerp diamond community during the war (for example by returning diamonds to refugees that had been confiscated by different states) (Laureys 2005).

8 8.1

RE-EMERGING STRUCTURES AS AN OUTCOME OF STRATEGIC ACTION Generating Advantages on the Supply Side

The evolution of the new growth centres was a major threat to the reemergence of the Antwerp diamond district which was desired by some Belgian policy makers and also by people from the diamond sector. Considering the fact that New York and Ramat Gan (Israel) still belong to the most important trading centres in the diamond industry in terms of turnover and employees shows that the new windows of locational opportunity were not of temporary nature and that the individual concerns of the Antwerp diamantaires were completely justified. As a reaction to the growing competition COFDI had started looking for a solution thereby taking some particularities of the sector into account: Since the late 19th century, the diamond industry had been characterized by cartel structures. By merging different smaller South African companies the ‘De Beers Consolidated Mines Ltd.’ was founded in March 1888 controlling a vast share of the produced rough diamonds (the evolution of the diamond cartel is subject to many publications – see for example Carstens 2001; Campbell 2003 – and therefore will only be outlined here). By and by, the company managed to acquire different mines which started selling their stones to a London-based de Beers subsidiary known as the Diamond Corporation Ltd (Dicorp). De Beers also managed to persuade outside diamond mining companies to sell their production via Dicorp, thus sparing them the establishment of an expensive distribution system. As a consequence, the company gained control over about 80 to 85 per cent of the world’s production of rough diamonds.18 Besides that, De Beers also centralized the distribution of raw diamonds through a unique system which was effective until 1979 (Kinsbergen 1984; Legrand 1991).19 Given De Beers’ control over the distribution of rough diamonds, it was important to convince this company to favour Antwerp’s rough diamond supply if the city was to regain its competitive edge again. Several factors, indeed, prompted De Beers to do so.

The re-emergence of the Antwerp diamond district

85

During the war, approximately 80 per cent of the world’s raw diamonds were extracted in the Congo which at that time was a Belgian colony. The majority of the stones were industrial grade (Kinsbergen 1984). The mining was carried out by the Forminiere company (Societe Internationale Forestiere et Miniere du Congo) with the Belgian government being its main shareholder. Forminiere distributed its production through De Beers’ London operation, based on a contract dated 1926 (Laureys 2005). In 1942 this contract had to be renewed. The Belgian government was very keen on the restoration of the diamond district once the war was over – it had accounted for 6 per cent of the Belgian exports before 1940 – and successfully negotiated the inclusion of a clause into the new contract specifying that large amount of cuttable raw diamonds should be reserved for the post-war diamond industry in Belgium. This matter had been brought to the Belgian government’s attention thanks to COFDI’s relentless lobbying. In return, Forminiere committed to supply the allies at advantageous prices hoping that this might stimulate a continued demand for industrial diamonds after the war. Last but not least, the mentioned measures only were possible as De Beers was interested in a long-term binding of Forminiere in order to keep its influence on the diamond market. As a consequence, the cartel agreed to compromise with its contractual partners (Laureys 2005). In addition to these measures being highly important for the reemergence of the district in terms of an initial spark,20 COFDI also succeeded in convincing De Beers to reduce or even stop the rough diamond supply to Brazil, Mexico and Cuba – where only well- established companies would be supplied. In addition, the supply to Palestine and to India was temporarily ceased for strategic reasons: it was feared that Germany could get a hold of stones from Palestine through Syria and Turkey. Japan could have imported stones from India (ibid.; Jack 1941) (Laureys 2005). All in all, it becomes apparent that through its strategic action COFDI was able to contribute to closing the window of locational opportunity by securing a brilliant post-war position for Antwerp in terms of supply and thus outnumbering the other centres. As the other locations did not hold any mines in pledge, this explains why Antwerp had a better negotiating position with De Beers and why similar decisions could not be made in the newly emerging centres. 8.2

Recovery of Labour

The worldwide dispersal of the diamantaires from Antwerp was regarded as another major problem in view of the potential attempt at resuscitation

86

Emerging clusters

of the cluster structures. Against this background, yet during the war, COFDI started to encourage the return of the fugitives through material and psychological incentives. Already in 1941, Goldmuntz and Huysmans had tried to convince the London-based Belgian government in exile to contact different organizations related to the diamond industry, especially the ones located in New York in order to trigger a friendly attitude towards Antwerp. In addition, the organization implemented several measures aiming at a quick recovery of labour after the war. In principal, they can be classified into two categories. The first one consists of general reparation measures such as the granting of transport benefits, Belgian nationality, visas and compensation for war damages. Personal assets were repatriated, and apartments and furniture were resituated to their pre-war owners. Not unimportantly the fight against anti-Semitism was of particular concern to Belgian and Antwerp authorities, too. The second category of measures was meant to rekindle business activities in the diamond sector: supply of rough stones, fiscal and financial advantages and support from the authorities when establishing business contacts. While at first the refugees tended to reject any thought of going back, many of them changed their minds with the first allied military successes in 1942 and with COFDI having successfully achieved to restrict the rough

25000 20000 15000 10000 5000

1954

1953

1952

1951

1950

1949

1948

1947

1946

1945

1944

1943

1942

1941

1940

1939

1938

1937

0

Note: The data were based on information provided by the compensation fund. As they refer to the entire Belgian diamond industry and are dated as of 31 December for each year they deviate partly from the details given in the text. Source:

Michielsen (1955, 87).

Figure 4.1

Number of diamond workers in Belgium

The re-emergence of the Antwerp diamond district

87

diamond supply to the new centres (see above). Although more and more diamond workers came to the district after the retreat of the Nazis (see Figure 4.1), convincing them to do so must be considered as one of the hardest challenges with regard to the recovery of the former structures (Vanden Daelen 2008) and a quick rebirth of the structure was all but certain. The most important reasons for the initial retention to return even after the Nazi retreat were twofold (see Laureys 2005): first, it was for fear for discrimination and latent anti-Semitism. Second, the business opportunities were difficult because (a) the diamond exchanges were occupied by allied troops (until 1947) and needed to be made available to resume trade; (b) the access to raw diamonds in the early post-war days was limited and (c) polishing plants had been damaged and were only partially operational. Generally speaking, the mentioned aspects made many Jewish diamantaires wait for the Antwerp trade to gain strength again

Share of Migrants by Provenance (1948–1965) >0 – <2 ⱖ2 – <4 ⱖ10 – <20 ⱖ20

Regions out of map N 3,6

0

750

1.500

3.000 Kilometers

Status: May 2008 Mapping: K. Leimer Layout: S. Henn

Source: Own depiction based on the analysis of immigration files of the city archive of Antwerp (N = 1135).

Figure 4.2

Migrants to the Antwerp diamond district by provenance (1948–65)

88

Emerging clusters

before returning whereas their very absence prevented this from happening earlier (ibid.). Though most of the measures outlined above clearly referred to the emigrants in the US (Vanden Daelen 2008), consulting the foreigner files for the period between 1948 and 1965 suggests that they might also have attracted people from other countries. Figure 4.2 shows that the vast majority of the diamond workers coming to Antwerp in fact returned from those states having experienced the strongest immigration before, with the US and Israel but also France and the UK being most relevant.21 Another important aspect is that according to the files many of the returning diamantaires were born in the Southern Carpathian region and – regarding their names – can (with caution) be considered Jews. It can be assumed that additional pull-factors were effective as a kind of subsidiary factor for them. First, neither diplomas nor any other certificates were required for a job in the diamond industry. Second, in many cases Jews took on jobs in sectors with limited government control for they often had temporary or no work permits. Actually the Belgian government adopted a very liberal attitude towards the diamond business and there were not as many formal rules as in other trades. This also allowed new immigrants to take up just a temporary profession in the diamond trade to get around until they left by ship for relatives in the US (Vanden Daelen 2006). 8.3

Chance and Other Factors

Besides the COFDI initiatives, emotional aspects (‘homesickness’) surely may have made some people go back to where they had lived before, too.22 In fact, apart from the Palestine refugees, many of the fleeing diamond dealers never intended to settle abroad. Furthermore, the migration to the district was favoured by factors that could not be influenced by the local actors. In a sense, these aspects could be termed as ‘chance’: according to Vanden Daelen (2008) the sector experienced a considerable growth after the first deliveries of stones, mainly triggered by the resumption of exports to the US. This, in fact, convinced some refugees to return. Moreover, during the Korean War (1950–53) the United States were interested in the compilation of a strategic stockpile. Some Antwerp diamantaires succeeded in towing this market to them which resulted in a massive influx of capital to the city and thus attracted other diamond dealers to the location as well (Laureys 2005). A different development was of importance in these years, too: since the 1950s the Dutch diamond district, a traditional competitor to Antwerp, had been struggling. Efforts were made to restart the industry and an international consulting company was commissioned to analyse and suggest ways to improve the local quality and production. This in turn

The re-emergence of the Antwerp diamond district

89

resulted in a strike in 1954 marking the final curtain of the Dutch diamond industry and encouraging many of its employees to move to Antwerp (Interview with Hans Wins, former De Beers PR manager in Antwerp, 16 April 2008). In fact, the analysis of the immigration files suggests that 9 per cent of the 1135 identified diamond people moving to Antwerp in the period between 1948 and 1965 came from the Netherlands. Finally, the reemergence of the structure was facilitated by the availability of parts of the former structures: many of the diamond workers were of autochthonous Flemish origin and thus had not been persecuted during the war. In other words, the knowledge of how to cut and polish a diamond was still to be found in the region. In addition, although some of the machines had been taken away by the occupants, most of the infrastructure was still available thus facilitating a recovery of the pre-war activities. In the end, the abovementioned immigration of Jews to Antwerp was facilitated by another aspect: due to some new anti-Semitic actions many Jews, especially from Poland but also from Russia and Czechoslovakia, fled from their countries to Belgium which was known for the liberal attitude of its government. Many of them started as unskilled workers cleaving the stones (Interview with Hans Wins, 16 April 2008; Interview with Yvan Verbraeck, long-time editorial journalist for the Syndikaat der Belgische Diamantnijverheid, 2008). Finally, yet before the war was over it was assumed that Antwerp would soon recover as a major trading hub, not only because tools and machinery were expected to be reinstalled quickly but also because smaller stones (so-called melee) could not be worked upon in New York since the labour costs were too high (Jack 1941) implying that there was a ‘big potential demand for Antwerp types of cut gems’ (The New York Times, 11 September 1944).23 All in all, together with the COFDI initiatives the mentioned aspects resulted in a ‘selective clustering’ as reflected in the number of diamond workers: On 1 April, 1945, there were 3480 cleavers, cutters and polishers, in May 1946 their number had grown to a considerable 13 570 (Kinsbergen 1984).

9

CONCLUSION

This chapter has discussed the re-emergence of the diamond district in Antwerp after World War II. It could be shown that the city had been a prospering diamond centre until different measures by the German military administration led to a complete discontinuation of all activities related to the sector in 1942. Many of the diamantaires who were able to

90

Emerging clusters

flee from the Nazis started new businesses abroad and contributed to the rise of competing centres. As a consequence of this, Antwerp’s recovery at first was all but sure. The development of the new locations, however, could be slowed down by purposeful action by a handful of actors related to the Belgian diamond industry. When detaching these results from their specific regional context and trying to identify some general patterns with regard to the re-emergence of cluster structures, several aspects deserve further consideration. To begin with, strategic action not only favours the creation of new paths but also the continuation of ruptured paths. In this context, key persons like the diamantaires Romi Goldmuntz and Herman Schamisso, representatives of the colonial diamond extraction company Forminiere, and politicians like Camille Huysmans can be regarded as the drivers of changing cluster structures. This aspect is consistent with other studies implying the high relevance of individuals in the process of cluster evolution (see Zucker et al. 1998 for the impact of key scientists on the emergence of clusters). These key persons are characterized by both having the power to change existing structures (for example due to economic strength and/or relationships like the key persons in the case study) and making use of it by taking strategic actions for several reasons. In principal, they have the choice to support the growth of economic activities at the location in question (for example by setting incentives to make people return to their former location), to avoid the development of competing locations (for example by restricting the supply to a competitor) or to adapt a mix of these options. While the measures relevant to the re-emergence of the structures need not necessarily be taken at the location in question (the Belgian exile government and COFDI were located in London), the key persons, of course, have to be convinced of the fact that their measures are likely to succeed and that the reasons for the rupture will cease to exist. With regard to the case study it is hard to imagine that similar steps would have been taken if the respective actors had expected the Nazi rule for a longer time. This line of reasoning is partly reflected in the evacuation of the diamond people to Cognac which was believed to be only a temporary hideaway. Even though strategic action could be elaborated as a central aspect to the recovery of former cluster structures, it must not be considered the only factor leading to the re-emergence of clusters. On the one hand, parts of the former infrastructure or knowledge base still may be in place supporting the rebirth of the structures at their former location. On the other hand, the re-emergence of a cluster in general should be attributed to a mix of both strategic action and ‘accidental’ events which cannot be influenced by the local actors but nevertheless affect the future development of

The re-emergence of the Antwerp diamond district

91

the trajectory (see Meyer and Schubert 2007). For instance, as explained above, the growth of the local cluster structures was facilitated by the Korean War, and the demise of the Amsterdam diamond industry as well as anti-Semitic sentiments in Eastern Europe. In general, this suggests that chance should be regarded as a factor opening up certain ‘spaces’ for strategic actions (like the new contract between Forminiere and De Beers given the lucky coincidence that it had to be renewed). All in all, the findings of the case study ask for a deeper consideration of redeveloping former cluster structures, an aspect which apart from some exceptions (see for example, Bathelt and Boggs 2003) has widely been neglected so far due to the concentration on path dependency. In this context research should put its focus on the interplay between key persons, strategic action and chance aspects. In addition, as the diamond sector has been characterized by monopolistic structures which have also influenced the regional key persons in their choice of measures, there is a need to address the question whether different market structures favour different types of actions. Given the fact that purposeful action need not be taken as soon as the rupture has set in (as in the case of the airship industry cluster in Friedrichshafen, Germany, which re-emerged about 50 years after the interruption of the trajectory), the determinants which provoke strategic actions at a certain point of time (for example the political framework or technological developments) should by analysed in more detail. Finally, future studies should address the question whether strategic actions differ depending on whether a cluster emerges for the first time or whether it re-emerges at its former location. In the literature it is argued that embryonic clusters can hardly be distinguished from their environment (Menzel and Fornahl 2005) so that potentially competing regions cannot be identified. The case of re-emerging clusters, however, seems to be different as existing clusters might already exist or can be observed during their emergence thus influencing the strategic measures taken by the relevant key persons.

NOTES * 1.

Acknowledgement: The authors thank the German Research Foundation (DFG) for supporting the underlying project ‘Transnational Communities and regional cluster dynamics. The cases of the diamond districts in Antwerp and New York’. A common and often-cited example of the phenomenon of (technological) path dependency illustrating it very well is the development of the typewriting keyboard (David 1985). Although different studies showed that other keyboard arrangements proved more efficient, the QWERTY pattern established itself as the most common keyboard in the second half of the 19th century (Grote 2004).

92 2.

3.

4. 5. 6. 7. 8. 9.

10.

11.

12.

Emerging clusters On the one hand, these models neither consider firm dynamics nor the heterogeneity of firms. On the other hand, they disregard geographical variety or countervailing powers to positive externalities and do not allow for any change or rupture of paths. Furthermore, these models exclude the co-existence of paths and view actors from an atomistic perspective thereby neglecting their embeddedness in social relations (Bathelt and Boggs 2005; Boschma 2007). In a seminal refinement Boschma (1997, 2007) suggests that industries do not really start from scratch but rather typically rely on certain generic factors a priori not being evenly spaced and therefore limiting the range of the window of locational opportunity. Accordingly, regions possessing these factors will, to a certain probability, contribute to the development of the said industry while in regions without them, development seems unlikely. This is not to imply, however, that the latter should be excluded from the analysis from the start: it can rather be assumed that these factors will develop by and by. In general, it therefore cannot be predicted where an industry forms; it can only be stated that the more regions are endued with these generic factors the more the window of locational opportunity is open. The notion here simply refers to any action in pursuit of an objective based on the conscious, rational calculation of likely actions of others. Originally the trading took place in cafes close to the Antwerp central railway station (Laureys 2005). Since 1937 Banque Diamantaire Anversoise SA, also known as Antwerpse Diamantbank NV and Antwerp Diamond Bank NV. However, industrial districts can be regarded as a special type of regional clusters (Koschatzky 2001; Sternberg and Litzenberger 2004). See the document prepared by Ephraim Schmidt in the Brachfeld Archive (BE SA 7537, Joden in de diamantnijverheid). Vanden Daelen (2006) lists several reasons for explaining the Jewish engagement in the diamond industry (for a more theoretical consideration see Richman 2006). First, as immigrants, Jews often were looking for professions that did not require intensive training. Furthermore, they preferred jobs allowing for an adaptation of their working rhythms to their religious practices. In fact, a closer analysis of the Jodenregister (Registry of Jews) dated 1940 shows that most of the Jews had worked as diamond cutters (40.7 per cent). In other words, jobs that easily could be carried out in a very flexible way at home or in small workshops allowed for the practice of religious rituals. Traditionally, and mostly after World War I, established diamond dealers expressed themselves in French which was also the dominant language in the diamond bourses whilst Jewish diamond people of Dutch origin kept to Dutch. More recent immigrants or less educated diamond workers, however, still spoke Yiddish which they had spoken in Eastern Europe (Teitelbaum-Hirsch 2001; Interview with David Urlik, former president of the Vrije Diamanthandel, Antwerpen, 26 March 2002; Interview with Nathan Ramet, Member of the board of the Diamantclub van Antwerpen, Chairman of the Museum for Deportation and Resistance in Malines, 1 October 2008). Cognac was (among other things) chosen as a location because it was assumed that the German armies would not advance in much the same way as they did in World War I – a surrender of France had not been expected. Furthermore, it can be assumed that the peripheral town was only selected as a refuge because a cosmopolitan and wealthy group of Jewish diamantaires would not settle there indefinitely, thus increasing the probability of their return to Antwerp (Jack 1941; Laureys 2005). The Germans at first were interested in keeping the Antwerp diamond industry alive for several reasons: on the one hand, the export of diamonds provided a hard currency which was needed to import raw materials into Germany; on the other hand, industrial diamonds had a high strategic value. Laureys (2003b, pp. 59ff.) points out that the Germans, being well aware that the diamond industry was in Jewish hands to a large extent, did not push for anti-Jewish measures at the beginning of the occupation but

The re-emergence of the Antwerp diamond district

13. 14. 15.

16.

17. 18. 19.

20.

21.

22.

23.

93

tried to recover the stocks that the refugees had taken with them. This ‘friendly’ behaviour, however, changed around 1941 when stocks were consigned by the Germans. As the US entered the war and hopes for more diamond dealers to return to Antwerp dwindled, the Germans started considering the implementation of a ‘final solution to the Jewish problem’ in the diamond industry as well. As many Jews had arrived in the 1930s, this step clearly can be regarded as an antiJewish measure, too (Laureys 2003b). Sicherheitspolizei/Sicherheitsdienst (Gestapo in occupied territories). Here, the establishment of new centres is based on external intervention while according to the original approach the emergence of new growth peripheries is regarded as a consequence of corporate endeavours either to expand or to cut costs (Storper and Walker 1989). However, data concerning the exact number of diamond workers are inconsistent. One source, for example, points out, that in 1948 there were about 3500 diamond cutters (The New York Times, 9 May 1948), while another one assumes about 800 diamond workers for the same year (Szenberg 1973). Volksgazet, 25 February 1965. The rest of the production was sold by the mining countries which benefited from the stability of the market (Legrand 1991). Approximately ten times a year, the Diamond Trading Company Ltd., a sister company of Dicorp, organized so-called sights in London (UK), Lucerne (Switzerland) and Kimberley (South Africa) where a certain amount of diamonds was offered at a fixed price (approximately 75 per cent of the market wholesale price) to only a small number of exclusive buyers (so-called sightholders) who had proven to be financially stable and loyal. Once a sightholder did not agree with the offer he was not invited again (Kinsbergen 1984; Legrand 1991). According to Laureys (2005), shortly before the war about 300 sightholders were specialized in jewellery diamonds and 30 in industrial diamonds. Though different files (see Antwerp City Archive, MA-KAB 1723, summary of the recovery of the Belgian diamond industry) point out, that the first deliveries from London were not sufficient – neither in quantity nor in quality (see Laureys 2005) – they actually were of great importance because they generated thousands of jobs (Vanden Daelen 2008). Comparing the Jodenregister with the immigration files, however, it becomes obvious that only a minor part of the pre-war Jews actually returned to Antwerp. To be exact, only 32 persons could be identified as having lived in Antwerp before World War II and having immigrated after 1948. These data, however, may be lower than the actual number for several reasons. First, the immigration files did not start until 1948 so that people who were repatriated between 1944 and 1948 could not be accounted for. Second, people may have gained Belgian nationality in a municipality outside Antwerp and thus were not registered as immigrants to the city. Third, people might not have registered in Antwerp but in a different municipality. One diamond-finisher quoted in the Sunday Express from 1 October 1944, said: ‘One thing we don’t like – the Welsh Sunday. Nothing to do but sit and think of the Continental Sunday, with all the pubs, cinemas, theatres and other places open to give people full enjoyment on their one full day free from work’. At the same time, however, it was feared that Antwerp would relieve the wartime shortage of smaller gems thus contributing to a fall of the prices for larger stones as well (The New York Times, 7 September 1944, 11 September 1944). Given the shortage of diamonds and diamond workers outside Europe, the supply of diamonds was greatly restricted in the period between 1940 and 1945. As a consequence, prices had risen strongly. By taking up activities in Belgium again, it was assumed that the costs for cutting and polishing would decrease. In any case it could be expected that the supply of diamonds would increase and prices would fall (The Times, 10 October 1944).

94

Emerging clusters

REFERENCES ADB (Antwerp Diamond Bank) (ed.) (2008), ‘History’, www.antwerpdiamond bank.com/index.php/ADB_en/profile/5, accessed 08 May 2008. Arthur, W.B. (1994), Increasing Returns and Path Dependence in the Economy, Ann Arbor, MI: University of Michigan Press. Bathelt, H. and J. Boggs (2003), ‘Towards a Reconceptualization of Regional Development Paths. Is Leipzig’s Media Cluster a Continuation of or a Rupture with the Past?’, Economic Geography, 79 (3), 265–93. Bathelt, H. and J. Boggs (2005), ‘Continuities, Ruptures and Re-bundling of Regional Development Paths: Leipzig’s Metamorphosis’, in G. Fuchs and P. Shapira (eds), Rethinking Regional Innovation and Change: Path Dependency or Regional Breakthrough? New York: Springer, pp. 147–70. Beurs voor Diamanthandel (ed.) (1979), Diamantbeurs 75 Jaar. 1904–1979, Antwerp, Belgium. Boschma, R.A. (1997), ‘New Industries and Windows of Locational Opportunity. A Long-Term Analysis of Belgium’, Erdkunde, 51 (1), 12–22. Boschma, R.A. (2007), ‘Path Creation, Path Dependence and Regional Development’, in J. Simmie and J. Carpenter (eds), Path Dependence and the Evolution of City Regional Development. Papers Presented at a Workshop at St. Catharine’s College, Cambridge University, Cambridge, 11 September 2007, Oxford: Oxford Brooks University, pp. 40–55. Braunerhjelm, P. and M. Feldman (2006), Cluster Genesis. Technology-Based Industrial Development, Oxford: Oxford University Press. Bünstorf, G. and S. Klepper (2009), ‘Heritage and Agglomeration: The Akron Tire Cluster revisited’, Economic Journal, 119, 705–33. Campbell, G. (2003), Blood Diamonds. Tracing the Deadly Path of the World’s Most Precious Stones, Cambridge, MA: Basic Books. Carstens, P. (2001), In the Company of Diamonds. De Beers, Kleinzee, and the Control of a Town, Athens, OH: Ohio University Press. Cooke, P. (2001), ‘Clusters as Key Determinants of Economic Growth: The Example of Biotechnology’, in A. Mariussen (ed.), Cluster Policies – Cluster Development?, Stockholm: Nordregio, pp. 23–38. Dalum, B., C.R. Pedersen and G. Villumsen (2005), ‘Technological Life-Cycles: Lessons from a Cluster Facing Disruption’, European Urban and Regional Studies, 12 (3), 229–46. David, P.A. (1985), ‘Clio and the Economics of QWERTY’, American Economic Review, 75 (2), 332–7. Even-Zohar, C. (2006), From Mine to Mistress. Corporate Strategies and Government Policies in the International Diamond Industry, Edenbridge, Kent: Mining Journal Books. Garud, R. and P. Karnøe (2001), ‘Path Creation as a Process of Mindful Deviation’, in R. Garud and P. Karnøe (eds), Path Dependence and Creation, Mahwah, NJ: Erlbaum, pp. 1–38. Grote, M. (2004), Die Entwicklung des Finanzplatzes Frankfurt, Berlin: Ducker and Humblot. Hassink, R. (2007), ‘Path Creation in City Regional Economies: The Case of the Computer and Video Game Industry in Seoul’, in J. Simmie and J. Carpenter (eds), Path Dependence and the Evolution of City Regional Development. Papers

The re-emergence of the Antwerp diamond district

95

presented at a workshop at St. Catharine’s College, Cambridge University, Cambridge 11 September 2007, Oxford: Oxford Brookes University, pp. 75–85. Henn, S. (2008), ‘Formierung und Wirkungsgefüge regionaler Technologiecluster. Das Beispiel Nanotechnologie im Saarland und in Berlin-Brandenburg’, Zeitschrift für Wirtschaftsgeographie, 52 (2–3), 95–113. Hoffmann, A. (2005), Zufall und Kontingenz in der Geschichtstheorie. Mit zwei Studien zu Theorie und Praxis der Sozialgeschichte, Frankfurt: Klostermann. Jack, A. (1941), ‘Nervous Ice – The World’s Diamond Trade Flees the War’, Saturday Evening Post, 19 April 1941, 105–17. Kinsbergen, A. (1984), Antwerpen, Briljant aan de Top in de Diamantwereld, Antwerp: Provincieraad van Antwerpen. Koschatzky, K. (2001), Räumliche Aspekte im Innovationsprozess. Ein Beitrag zur neuen Wirtschaftsgeographie aus Sicht der regionalen Innovationsforschung, Münster: LIT. Laureys, E. (2003a), ‘De Joodse Diamantdiaspora en de Antwerpse Diamantindustrie 1940–1945’, in H.A.M. Klemann and D. Luyten (eds), Thuisfront. Oorlog en Economie in de Twintigste Eeuw, Zutphen, Netherlands: Nederlands Instituut voor Oorlogsdocumentatie, pp. 179–91. Laureys, E. (2003b), ‘The Plundering of Antwerp’s Jewish Diamond Dealers, 1940–1944’, in Center for Advanced Holocaust Studies (ed.), Symposium Proceedings of Confiscation of Jewish Property in Europe 1933-1945. New Sources and Perspectives, Washington, DC: Center for Advanced Holocaust Studies at the United States Holocaust Memorial Museum, pp. 57–74. Laureys, E. (2005), Meesters van het Diamant. De Belgische Diamantsector Tijdens het Nazibewind, Tielt, Belgium: Lannoo. Legrand, J. (1991), ‘Die Central Selling Organization’, in R. Maillard (ed.), Der Diamant. Mythos, Magie und Wirklichkeit, Erlangen, Germany: K. Müller Verlag, pp. 188–9. Markusen, A. (1996), ‘Sticky Places in a Slippery Space: A Typology of Industrial Districts’, Economic Geography, 72 (3), 293–313. Marshall, A. (1920), Principles of Economics, London: Macmillan. Martin, R. (2001), ‘Geography and Public Policy: The Case of the Missing Agenda’, Progress in Human Geography, 25 (2), 189–210. Martin, R. and P. Sunley (2006), ‘Path Dependence and Regional Economic Evolution’, Journal of Economic Geography, 6 (4), 395–437. Menzel, M.-P. (2008), ‘Zufälle und Agglomerationseffekte bei der Clusterentstehung. Eine vergleichende Diskussion des Core-periphery-Modells, des Window of locational opportunity-Konzepts sowie stochastischer Ansätze’, Zeitschrift für Wirtschaftsgeographie, 52 (2–3), 114–28. Menzel, M.-P. and D. Fornahl (2005), ‘Unternehmensgründungen und regionale Cluster. Ein Stufenmodell mit quantitativen, qualitativen und systemischen Faktoren’, Zeitschrift für Wirtschaftsgeographie, 49 (3–4), 131–49. Meyer, U. and C. Schubert (2007), ‘Integrating Path Dependency and Path Creation in a General Understanding of Path Constitution. The Role of Agency and Institutions in the Stabilisation of Technological Innovations’, Science, Technology & Innovation Studies, 3 (1), 23–44. Michielsen, A. (1955), De Diamanteconomie. Waarde, Prijs en Conjunctuur, Antwerp, Belgium: Christelijke Belgische Diamantbewerkerscentrale. Porter, M.E. (1990), The Competitive Advantage of Nations, London: Macmillan.

96

Emerging clusters

Press, K. (2006), A Life Cycle for Clusters? The Dynamics of Agglomeration, Change, and Adaptation, Heidelberg: Physica-Verlag. Puffert, D. (2000), ‘Path Dependence, Network Form and Technological Change’, Paper Presented at the Conference to Honour Paul David – History Matters: Economic Growth, Technology, and Population, Stanford University, Stanford, CA. Richman, B. D. (2006), ‘How Community Institutions Create Economic Advantage: Jewish Diamond Merchants in New York’, Durham, NJ (Harvard Law and Economics Discussion Paper 384). Schamp, E.W. (2000), Vernetzte Produktion. Industriegeographie aus institutioneller Perspektive, Darmstadt, Germany: Wissenschaftliche Buchgesellschaft. Stack, M. and M. Gartland (2005), ‘The Repeal of Prohibition and the Resurgence of the National Breweries. Productive Efficiency or Path Creation?’, Journal of Management History, 43 (3), 420–32. Sternberg, R. and T. Litzenberger (2004), ‘Regional Clusters in Germany – their Geography and their Relevance for Entrepreneurial Activities’, European Planning Studies, 12 (6), 767–91. Storper, M. and R. Walker (1989), The Capitalist Imperative. Territory, Technology, and Industrial Growth, New York, NY; Oxford, UK: Basil Blackwell. Sunday Express (1 October 1944), ‘“Red-Tape” may Rob Britain of a Trade’. Szenberg, M. (1973), The Economics of the Israeli Diamond Industry, New York, NY: Basic Books. Teitelbaum-Hirsch, V. (2001), Diamantaire. L’Univers & les Coulisses d’une Passion, Brussels, Belgium: LABOR. The New York Times (7 September 1944), ‘More Diamonds Due Freeing of Antwerp is Expected to Reopen Huge Market’. The New York Times (11 September 1944), ‘Antwerp’s Diamond Industry Seen Restarting Despite Handicaps’. The New York Times (9 May 1948), ‘Diamond Cutting Continues in US’. The New York Times (27 March 1960), ‘Diamond Market Busy in Antwerp’. The New York Times (31 May 2005), ‘Diamond Polishing Is One More Dynamic Facet of China’, http://www.nytimes.com/2005/05/31/business/ worldbusiness/31diamonds.html, accessed 16 June 2008. The Times (10 October 1944), City Notes. van Dyck, P. (1989), De Diamantsector en de Antwerpse Economie, Antwerp (Dissertation). Vanden Daelen, V. (2006), ‘Antwerp Jews and the Diamond Trade: Jews shaping Diamonds or Diamonds shaping Jews?’ Working paper prepared for the XIV International Economic History Congress in Helsinki, Finland, 21–25 August 2006. Session 22: Ethnic, Religious or Cultural Plurality and Economic Institution Building, Antwerp. Vanden Daelen, V. (2008), Laten we hun lied verder zingen: de Heropbouw van de Joodse Gemeenschap in Antwerpen na de Tweede Wereldoorlog (1944–1960), Amsterdam: Aksant. Volksgazet (25 February 1965), ‘Internationale Toestand van de Diamantnijverheid’. Walgrave, J. (1993), ‘Diamond in Antwerp. A Brillant Story’, in Federation of Belgian Diamond Bourses (eds), Antwerp. The Diamond Capital of the World, Antwerp, pp. 33–49. Zucker, L., M. Darby and M. Brewer (1998), ‘Intellectual Capital and the Birth of U.S. Biotechnology Enterprises’, American Economic Review, 88 (1), 290–306.

PART II

Institutions and endogenous dynamics

5.

Origins of human capital in clusters: regional, industrial and academic transitions in media clusters in Germany Anne Otto and Dirk Fornahl

1

INTRODUCTION

A successful cluster relies on a strong local human capital base and on a high level of inter-firm job mobility as a mean for knowledge transfer. However, the large body of cluster studies gives little evidence on the development of such local labour market pools. Thus, the question is how this central element of clusters emerges. While in the cluster emergence stage, the demand for human capital is relatively modest, in the growth stage firms in the cluster rely strongly on an extensive human capital base. Besides this quantitative aspect, the competencies demanded in the two stages probably also differ with a demand for quite specialised ones in the emergence stage and a much wider range in the growth stage. There exist different mechanisms of creating or accumulating human capital inside a cluster. Based on the peculiar demand for human capital just described the relevance of these mechanisms probably varies across the cluster life cycle. This is to say, human capital can be created within the cluster region either by educating and training pupils and students or by job changes from non-cluster firms to cluster firms (intra-regional mobility). In addition, the inflows of extra-regional labour are a major source of human capital, particularly in the stage of cluster emergence whereas local human capital production seems to be more important in the growth stage. Accordingly, a change of the pattern of labour mobility between these stages is assumed. In addition, in the growth stage labour mobility probably generates the most externalities because a large labour pool is present and high-rates of intra-regional mobility leads to increased knowledge transfer. Empirical studies dealing with cluster life cycles have

99

100

Emerging clusters

not taken into account these stage-specific accumulation and transfer processes of (local) human capital so far (Menzel and Fornahl 2007). Compared to recent cluster studies investigating the local labour force and its mobility which mostly focus on one single cluster, this chapter provides a comparative perspective. Thus, the objective of this chapter is to provide an interregional comparative analysis of processes of human capital creation – with a focus on labour mobility – between emerging and growing clusters in Germany’s audio-visual media industry (AVM) in order to reveal systematic evidence for the above-stated assumptions. This chapter explores whether a specialised pool of labour emerges in the selected AVM clusters and analyses a range of assumptions dealing with the level and patterns of job mobility in these emerging and growing clusters. Labour mobility is measured by labour inflows. The AVM industry is in many respects an interesting unit of analysis. First, this industry is highly concentrated within Germany and there is a range of local media clusters. Second, our database covers nearly all important periods of industrial development. Since our time series starts in the beginning of the 1980s, we are able to examine the first growth stage of the audiovisual media industry in Germany which occurred during the 1980s and the second one which took place at the end of the 1990s. Before the 1980s all clusters consisted of public broadcasting companies or their suppliers, but starting from the 1980s new clusters emerged and the older ones changed due the establishment of private broadcasting firms. Hence, today the clusters have a mixed composition: some are still dominated by public firms, while others are dominated by private firms and again others are composed of both types of firms. The remainder of this chapter is structured as follows. The next section gives a brief review of the theoretical and empirical literature on the interaction between industrial clusters and human capital (mobility). Section 3 depicts the applied criteria for the identification of emerging and growing media clusters and describes the database of our analysis. Then, the assumptions on the patterns of job mobility within clusters are introduced and we test them by our empirical analysis (section 4). In the last section the outcomes of the comparative analysis are discussed.

2 A SHORT LITERATURE REVIEW: THE RELEVANCE OF HUMAN CAPITAL AND LABOUR MOBILITY FOR INDUSTRIAL CLUSTERS In cluster research both the availability of a local labour pool and the mobility of local employees are considered to be crucial for the emergence,

Origins of human capital in clusters

101

existence and success of clusters. This section gives a short review of theoretical literature which either explains the development of a cluster-specific human capital base or the importance of labour mobility as a mechanism for the creation of human capital and as a mean for knowledge transfer. A short summary of empirical evidence on these issues is also provided. The theory of Storper and Walker (1989) takes into account the development of a local labour pool and the mechanisms of human capital creation within a cluster whereas location and economic theory argue the other way around and highlight the relevance of the local labour pool and of job mobility for the existence of clusters. The evolutionary approach of Storper and Walker (1989) explains the development of industrial clusters based on the resources these clusters generate themselves. Storper and Walker (1989, p. 96) postulate that industries create regional resources and not the other way around. The composition of inputs in an industry and the scope of the market are the result of innovations in product, process, and organization that generate competitive advantage, dynamic economies, and high rates of accumulation. [. . .] Firms and sectors generate their own input histories, and those of their chosen regions, at the same time. It follows that the central motor of regional development is not industry location as a response to prior resource endowments, but geographical industrialization as a process of growth and resource creation.

The emergence and growth of industrial clusters relies, therefore, mainly on the resource creation by the firms and not on the availability of resource endowments within the region. This approach provides particular insights into the development of a cluster-specific human capital base during the emergence and growth stages. In the stage of cluster emergence, above all in new industries, technologies and competence fields, there are no specific requirements with respect to the location, resources and qualifications of human capital and so on. First, firms try to compensate for the lack of available employees inside the region by hiring qualified employees from other regions. But Storper and Walker (1989) state that this is not always possible, since the needed human capital is not available in any region in this stage. This may be true especially for academic workers which embody key knowledge which is not available in each single regional labour market. This type of highlyskilled workers is specially needed in knowledge- and technology-based industries and human-capital-intensive creative service industries. Second, the adaptability of the (local) workforce to the needs of the new industries and firms is of utmost importance in this stage. In contrast to the abovementioned assumption of Storper and Walker, general qualifications such as jobs for secretaries or marketing and sales activities are available

102

Emerging clusters

on nearly every regional labour market. In-house training and advanced education of local workers may enable them to adapt quickly to the new necessities of the cluster firms. To sum up, there is a large need for an extra-regional workforce in the emergent stage, particularly for highlyqualified employees. Interregional mobility might be a major mechanism for the creation of a cluster-specific human capital base. However, the demand for general jobs which are required in nearly every firm should be satisfied also by the local workforce. It is expected that employees in this cluster stage have worked beforehand to a large extent in other industries than the cluster industry itself. Institutions such as schools, universities or schools providing vocational education may adapt to the qualification necessities of cluster firms in the growth stage, for instance by establishing new course programmes. The cluster region is, therefore, able to endogenously satisfy these needs in the growth stage. But, the demand for human capital is extraordinarily large in this stage as well. Hence, the additional inflow of extra-regional labour is assumed to be strong in this stage. The prosperity of the cluster region and the related high rate of extra-regional labour inflow might result in a tradeoff between the availability of a high number of potential employees and the likelihood that these employees match the demands of the cluster firms. The intra-regional mobility is probably also an important source of human capital for the cluster firms: employees who stem from a noncluster-firm might move to cluster firms. Even job changes among cluster firms become more likely. Storper and Walker (1989) differentiate between two different forms of regional clusters. A cluster dominated by a vertically integrated large enterprise will possibly exhibit weak local inter-firm mobility because in-house job careers are more attractive for the staff. In a localised production complex with vertically disintegrated firms job changes are more likely to occur among firms and, thus, inter-firm mobility is expected to be strong. All in all, the development stage of the cluster and the cluster structure can both influence the patterns of job mobility. In contrast, a cluster may also emerge when a specialised local labour pool is already available. This might be the case because of a shut-down of a large local organisation (for example research centre) or of a big firm (Feldman 2001). Klöpper (2005) demonstrates, for instance, the emergence of an ITcluster in Paderborn (Germany) as the consequence of an economic crisis of the large high-tech firm Siemens. The following theoretical approaches highlight both the relevance of the local labour market and of job mobility for the ‘functioning’ and dynamism of industrial agglomerations. These approaches are Weber’s neoclassical location theory of agglomeration (Weber 1909) and the competition-oriented cluster concepts (Porter 1998) which regard the

Origins of human capital in clusters

103

local labour pool as a substantial part of clusters. Yet another prominent approach is Marshall’s labour pooling thesis which is seen by him as one out of three major reasons for the formation and existence of clusters (Marshall 1890; Krugman 1991). Marshall’s thesis assumes that the search for a new job is relatively effortless for workers and employees located in an industrial cluster. This holds especially true for those employees whose professions, qualifications and working experience are oriented towards the core activities of the cluster firms. The job search process within a cluster is alleviated because there is a large pool of similar jobs and the firms offer quite favourable job conditions, for instance high salaries, in order to attract and to retain adequate and highly skilled employees (Amend and Herbst 2008). The risk of becoming unemployed and staying so for a longer period of time is, therefore, relatively low within a cluster. Such a labour market offers two advantages for the firms: a numeric flexibility and a functional flexibility taking into account that firms have access to high numbers of workers with clusterspecific competencies (Angel 1991). Otherwise the firms would have to look for employees on the national and international labour market which is much more expensive or the quality of the matching might be lower. All in all, these pooling effects contribute substantially to facilitate the search and screening processes of employers and employees on the local labour market. Thus, the matching process becomes more efficient. This improved matching process is presumed to alleviate the intra-regional mobility of workers. The cluster firms can make use of positive external economies by virtue of these pooling effects. These specific external economies based on labour market pooling are one important element of the so-called Marshall–Arrow– Romer (MAR) spillovers (Gordon and McCann 2000). At least, during the growth stage labour supply and demand should profit more and more from such pooling effects on the local labour market. The other side of the coin is, however, that firms have to face competition with other firms when looking for new employees. Furthermore, labour poaching is part of this competition in which employees are hired by other firms. To avoid labour poaching the cluster firms might be more inclined to pay higher wages. Thereby, they can retain their working staff and also the firm-specific incorporated knowledge of their employees. According to the theory of efficient wages (Franz 2006) firms are thereby able to keep their staff and also to spare the costs for the vocational adjustment of the new personnel. Cluster firms are therefore confronted with a sort of stand-off situation with respect to the conflictive effects of labour pooling and poaching. Localised knowledge spillovers provide another theoretical link between job mobility and cluster dynamics. Implicit knowledge gained through working experience in cluster firms is incorporated in local human capital.

104

Emerging clusters

The higher the mobility rates of workers within the cluster, the faster such uncodified knowledge diffuses among firms. This knowledge diffusion may shape, facilitate and improve other cluster processes significantly such as collective learning processes, invention and innovation activities of firms and in the end sustain the competitiveness of the cluster. The mobility of high-qualified employees with embodied key knowledge is in this respect of major importance. The access to such key knowledge might be crucial for the invention of new products and for the improvement of innovation processes within firms. The latter is especially true, since the employees do not only transfer this embodied knowledge to the new firm, but also their ability to access certain local (informal) networks (Zellner and Fornahl 2002). Hence, increased job mobility also strengthens the density of local networks leading to higher diffusion rates of information and knowledge inside the cluster. The local structure of inter-firm job mobility and thus indirectly also the intra-cluster diffusion patterns of knowledge depend on a multitude of factors. As mentioned beforehand, this is foremost the stage of cluster development and the cluster structure. In addition, geographical mobility is determined by socio-demographic and professional characteristics of individuals and by the regional environment (Mertens and Haas 2006). To sum up, this process is a complex one which is determined at the individual, the firm and the industrial level and also at different spatial scales (regional, national and international scale). Although there is a large body of cluster studies in regional economics and economic geography, only a few studies highlight cluster-specific labour markets. Most of them explore the thesis of strong intra-cluster job mobility. These studies refer either to the mobility of the total workforce or to the mobility of certain groups of employees (for example high- and low-skilled workers) in different clusters, situated mostly in technologyand knowledge-intensive industries. The following studies corroborate the mobility thesis: Angel (1991); Dahl (2002); Power and Lundmark (2004); Henry and Pinch (2000); Carnoy, Castells and Brenner (1997); Fallick, Fleischmann and Rebitzer (2005). The works of De Blasio and Di Addario (2005) and Frank (2008) reveal, however, opposite outcomes. In all these aforementioned cluster studies the mobility rates within the clusters are compared to different reference rates, for instance to mobility rates at the regional, industrial or national level. Hence, the extent to which the results of these studies are comparable is limited. Interregional studies comparing labour mobility between several clusters may reveal some more systematic results in this respect. For instance, Lawton Smith and Waters (2005) explain the differing levels of job mobility in the two high-tech clusters Oxford and Cambridge with the different structures of these two clusters.

Origins of human capital in clusters

105

Labour mobility in ‘functioning’ clusters is a mean for positive externalities such as knowledge spillovers. The evidence of localised knowledge spillovers is emphasised by several studies tracing the mobility paths and career tracks of star scientists in industrial agglomerations. The first studies, highlighting this aspect, refer to biotechnology clusters in the United States (see for example Zucker, Darby and Armstrong 1998 or Zucker, Darby and Brewer 1998). They find that biotechnology clusters emerge in close vicinity to star scientists due to the important role they play in this industry and at the same time their relative immobility. This body of literature, which is critically reviewed in the seminal work of Breschi and Lissoni (2001), takes into account only the key knowledgeagents. Thus, one cannot generalise these results to the overall patterns of job mobility and therefore of knowledge diffusion within clusters. In most cases the phenomenon of labour poaching is analysed by theoretical modelling in labour market economics, for instance by Combes and Duranton (2001) or Amend and Herbst (2008). Empirical studies refer mostly to the case of Silicon Valley (for example Fallick, Fleischmann and Rebitzer 2005). Frank (2008) shows that audio-visual media firms hesitate to locate within the media cluster of Potsdam/Berlin because of the risk of losing key knowledge via labour poaching. The aspect of labour poaching, is however, not taken into account in this analysis.

3

IDENTIFICATION OF EMERGING AND GROWING AVM CLUSTERS IN GERMANY

The data on job mobility of German employees, which is used in our analysis, derives from the Historic Employment Database and the Historic Firm Panel of the Institute of Employment Research. These longitudinal databases are originally based upon the employment notifications in Germany. These two datasets cover a period from 1975 to 2006. The historic employment database provides micro data for all employees and benefit recipients (for example unemployed) in Germany, for instance demographic information, education, occupational codes, employment status, gross pay per day, unemployment benefits, place of work and place of residence (since 1999). The historic firm panel contains information for all firms with at least one employee, for instance industry codes, establishment size, location of the firm and so on. These two datasets can be linked via the firmspecific registration number of the social insurance system. Thereby, we obtain a linked employer–employee dataset which enables us to analyse job mobility at the individual, firm, regional and industrial level. We focus on the analysis of the movements of employees into businesses and regions

106

Emerging clusters

(labour inflows). A job mover is defined as a person who has changed his/ her workplace between two establishments within a one-year period. On the one hand, intra-regional mobility rates might be overestimated by our database. It is because the unit of analysis is plants, and several plants of one enterprise can be registered under different numbers within a municipality. Thus, intra-firm job changes between the plants of one enterprise may contribute artificially to increased intra-regional labour mobility. On the other hand, freelancers work very often for a certain period within a media firm. After their project is finished they start a new project in another firm. These job changes are to a certain extent not covered by our database. This results from the fact that freelancers are mostly self-employed and, thus, they are not obliged to be a member in the German social insurance system. Then these freelancers are not registered in the databases of the Federal Agency of Employment. The rates of labour mobility in Germany’s AVM industry may, therefore, be underestimated by our database. To capture long-term developments of regional clusters, the chapter takes into account the time period from 1980 to 2007. As pointed out in section 1, the analysis focuses on the emerging and growing stages of AVM clusters in Germany. The identification of these stages of the cluster life cycle is based on regional characteristics in three different sub-periods: 1980–89, 1990–98 and 1999–2007. Hence, the AVM cluster in one region can be identified three times in different stages of development. We focus the analysis on West German regions and Berlin in order to avoid specific reunification and East Germany labour market effects which could disturb the overall validity of the study. On the opposite side a focus on West Germany should lead to sound results, since most of the AVM activities are anyhow located in West Germany with only Leipzig and its surrounding areas as well as Potsdam/Babelsberg having a considerable amount of activities in the AVM sector in East Germany. The AVM industry consists of a range of sub-sectors. Table 5A.1 (Appendix) presents those sectors which are included in the subsequent analysis. Since the sectoral classification changed between 1980 and 2007, we applied three different classification systems. Figure 5.1 depicts the developments in the number of firms and employees from 1980 to 2007. While the number of firms stabilised after 1990, indicating that the life cycle of the industry is already in a relatively mature state without exhibiting a shake-out until now, the number of employees increased until the year 2000 and levelled out in the following years. The significant increase in the number of employees in the 1980s is due to the specific dual structure of Germany’s AVM industry. Only public television stations existed until 1984. After the Second World War the Allies established one public television station in each of West Germany’s eleven federal states. These

Origins of human capital in clusters

107 90000

6000

80000 5000 70000 60000

4000

50000 3000 40000 30000

2000

20000 1000 10000 0

Number of firms

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

1989

1988

1987

1980

0

Number of employees

Note: there is a lag in the data between 1980 and 1987.

Figure 5.1

Development of number of firms and employees in the audiovisual media industry

federal regional television stations together built the public television station ARD. The formation of the second public television station ZDF which broadcasts nationwide and is situated in Mainz is investigated by Dorenkamp and Mossig in this volume. Since 1984, the foundation of private television stations contributed significantly to the rise in Germany’s AVM employment (Mossig 2004). Regional cluster research did not manage to establish unique methods, either bottom-up or top-down approaches, in order to identify clusters. This is in part due to the fuzziness of this concept. We use a top-down approach which has in common with other studies the application of a concentration index, the Concentration Index (CI) of Sternberg and Litzenberger (2004), in order to identify industrial agglomerations. The emergence and growth stage are measured by developments in the numbers of employees and firms within the AVM industry. The starting points are the developments in the number of firms, employees and the concentration index (CI) in the three different periods of time in the 327 so-called ‘Kreise’ in West Germany for the AVM industry. The spatial unit ‘Kreis’ refers to the NUTS 3 level of the European Union. We denominate the Kreis as a local district. We calculated the average Birch index for the number of firms and

108

Emerging clusters

employees, which is a ‘balanced’ index considering both the relative as well as the absolute growth rates (Birch 1987) for each district and time period (1980–89, 1990–98, 1999–2007). The distribution of the Birch growth index was then divided into four quantiles with the fourth quantile containing those regions with the highest growth rates. Furthermore, we calculated the average concentration index for each district and time period. We selected the concentration index by Sternberg and Litzenberger because this index controls for several basic regional features such as the size and the number of inhabitants in the district. In order to identify the cluster stage we defined an emerging cluster if the concentration index was below 1, which indicates a non-clustered situation, in one time period, and above 1 in the subsequent period, indicating a certain degree of clustering (those regions are labelled with ‘E’ in the following). Hence, this procedure cannot identify an emerging cluster in the period between 1980 and 1989. A growing cluster is defined as a region with a concentration index above 1 and the Birch index for the number of employees as well as for the number of firms must be in the highest of the four quantiles (labelled with ‘G’). Those region which emerge and grow later on are labelled with ‘EG’. Although there might be other procedures on how to define stages of a cluster life cycle, until now none of these is known to us which provides less arbitrary borders. McGahan and Silverman (2001) have used a similar approach, only focusing on the decline in the growth rates of the number of firms in order to define stages of an industry life cycle. We identified 36 districts in the three periods of time as emerging or growing clusters. Table 5A.2 in the Appendix gives a cluster description with, for example, the number of firms/employees and the concentration index. Figure 5.2 shows the location of the selected districts. Averaged over the full period of time the identified regions cover 59 per cent of industry employment; around 45 000 of the 77 500 employees (average values over all years) work in the identified districts. The five largest identified districts in terms of AVM employees are Cologne, Hamburg, Munich, Mainz and BadenBaden. The districts Cologne and Rhein-Erft constitute one AVM cluster. However, the public television WDR is situated in Cologne whereas private television stations formed the second part of this cluster in the 1990s. In the largest clusters either a large public or private television station or both are located. On average the growing clusters have 2.71 per cent of total industry employment and the emerging clusters 0.14 per cent. Thus, the emerging clusters have a relatively low number of employees which increases over time, but this growth is relatively modest. To define the geographic reach of labour mobility we define local mobility as those employees originating from the same Kreis as the cluster is located in; the regional supply of employees comes from the same labour

Origins of human capital in clusters

109

Flensburg Kiel Hamburg

Region Hannover Berlin Osnabrück Münster Bochum Gelsenkirchen Hagen Duisburg Mülheim Krefeld Wuppertal Cologne Rhein-Erft-Kreis Bonn Offenbach

Hof

Aschaffenburg

Bayreuth

Mainz

Würzburg Erlangen Nuremberg

Trier

Schwabach Kaiserslautern

Karlsruhe

Pforzheim Baden-Baden

Cluster Stage Growing clusters Emerging clusters

Note:

Passau

Rosenheim

Augsburg Kempten Munich

Further large AVM firms in the clusters are listed in Table 5A.5 in the Appendix.

Figure 5.2

Location (Kreis) of the identified emerging and growing AVM clusters

market region, but from a different Kreis. One labour market region consists of several districts. Thus, flows of employees deriving from the same Kreis and labour market region in which the cluster is located are defined as intra-regional mobility. The extra-regional supply of employees

110

Emerging clusters

originates from all other West German labour market regions. This type of geographical mobility is defined as interregional mobility. We applied the definition of regional labour market regions in Germany conducted by Eckey, Koesfeld and Türck (2006). Table 5A.2 provides the information to which labour market regions the clusters are assigned.

4

EMERGING AND GROWING CLUSTERS: THE SOURCING OF HUMAN CAPITAL

In the following section the specialisation and development of the local labour pool in all selected AVM clusters is investigated. Then some hypotheses on the patterns of job mobility in emerging and growing clusters are introduced based on the review of theoretical and empirical approaches in section 2. These hypotheses are tested by our empirical analysis. The focus relies on the comparison of the outcomes between the clusters of these two different development stages. 4.1

Local Labour Market Pool and its Development

In the previous section we pointed to the fact that we also identify clusters by a high concentration of activities in the respective industry as the majority of cluster studies do. In many such cluster studies the existence of a strong industry-specific concentration is implicitly equated with the existence of a specialised labour market pool. However, a differentiation of the labour market pool into different cluster-related occupations has not been conducted so far. By doing so, our starting point is the definition of those occupations which are of utmost importance in the AVM industry. The occupations which exhibit the highest shares of employment in West Germany’s AVM industry averaged over the whole period of time (1980–2006) are selected as cluster-specific jobs. The ten most important jobs are displayed in Table 5.1. The three top occupations – office clerks, publicists and artistic and attributive occupations – account for 46 per cent of total AVM employment. Overall, the ten occupations listed in Table 5.1 sum up to nearly 70 per cent of the total AVM workforce. Most of these occupations only require a vocational education (mediumlevel qualification). For example, this holds true for office clerks, typists, photographers and salesmen. In particular, the group of artistic and attributive occupations comprises career changers seeking new opportunities in this creative industry. In comparison to the aforementioned occupations there are no high entry barriers in these creative occupations because they do not require formal degrees. Even employees without any

Origins of human capital in clusters

Table 5.1

111

Ranking of most important occupations in AVM industry

Name of job

Average no. of employees per year

Share (%)

Cumulative percentage

Office clerks Publicists Artistic and attributive occupations Shorthand writers, typists Entrepreneurs, directors, division managers Descriptive artists Photographers Electrical engineering technicians Salesmen Musicians

12 077.71 11 410.86 11 403.14 4 497.43 2 933.57

15.6 14.7 14.7 5.8 3.8

16 30 45 51 55

2 521.71 2 225.71 2 210.43 1 741.29 1 312.86

3.3 2.9 2.9 2.2 1.7

58 61 64 66 68

Others

25 182.43

32.5

100

education can enter this industry. In other industries, the possibilities for career changers are much more restricted. In recent years, specific occupations related to the AVM industry emerged and were institutionalised. However, most of these new jobs cannot be represented by the current, but older, occupational classification system of the Federal Agency of Employment (Nuremberg). Table 5A.3 presents the rank correlation of the occupations for those clusters which have more than 1 per cent of the total AVM employees in West Germany. On average, strong correlations can be observed between the occupational structures of the human capital base in these AVM clusters. Hence, the industry builds upon comparable human capital; no matter in which location. The correlation coefficient between the clusters Cologne and Munich is the highest one, while Munich and Bonn exhibit the lowest coefficient value. Although there are still some more differences in detail: the shares of occupations differ between the clusters and vary also between the overall industry and the different clusters. However, a closer look at the regional level reveals that the same occupations – as indicated by the ranking above – make up a crucial part of the clusterspecific labour pools (Table 5A.4). In the selected clusters the minimum and maximum employment share of the ten top AVM occupations in the total AVM workforce amounts to about 50 per cent and 80 per cent, respectively. For instance, roughly 20 per cent of the AVM employees in Munich are concentrated within only one occupational code: the artistic

112

Emerging clusters

and attributive occupations. Thus, it can be concluded that the labour pool in each of the identified AVM clusters relies to a large extent on a specialised occupational portfolio. 4.2

Labour Inflows into the Selected AVM Clusters

By aggregating the total number of AVM employees over the three time periods for the respective clusters we obtain a number of 466 000 workers (Table 5.2). We traced in which industries, regions and firms these workers were employed in the year before (t − 1). These annual comparisons at the micro-level are possible because the Historic Employment Database and the Historic Firm Panel provide for each employee and time period the registration number of the firm in which his workplace is located. We distinguish between three different groups of employees and thus between three forms of transitions when comparing their workplaces between t and t − 1 (Table 5.2): ●

●

●

If the former workplace in t − 1 corresponds to the present workplace in t, the employees did not change their job. The share of this group of non-job changers in all AVM employees amounts to 71 per cent (column d). If the employee changed his job and started at a new workplace between t and t − 1, he is identified as a job mover. These job movers account for 14 per cent (65 000) of all AVM employees (column e). The third group includes all persons who also started a new job between period t and t − 1 in firms of the AVM industry (around 15 per cent) (column f). However, these persons were not employed in period t − 1 in a firm. There are many reasons which explain the integration of this group in the labour market. These persons could have been unemployed or at a college or university, in maternity protection and so on. The Historic Employment Data base provides no detailed information about the disposition of this group of employees in t − 1.

Table 5.2 depicts the average shares of these three groups of employees in total AVM employment in the selected clusters and time periods. The shares of the non-job changers vary between 45 per cent (Flensburg in Period 3) and 85 per cent (Baden-Baden in Period 2). A range of clusters is dominated by large public television stations. The employees of these broadcasting stations obtain a tenured position after having worked for 15 years in their job. Hence, the shares of non-job changers in these clusters, for instance Cologne, Mainz and Baden-Baden, are relatively high. This holds also true for smaller clusters with tiny public broadcasting stations

113

G G G G G G G G G G G G G G

Cluster stage (a)

1 1 1 1 1 1 1 1 1 1 1 1 1 1

Time period (b)

380 1 170 624 3 268 967 965 726 32 897 1 061 439 168 2 381 507 684

Total number of AVM employees over the period (c) 53.4 66.3 60.9 72.8 63.4 56.9 78.5 80.0 71.0 69.9 63.7 74.2 69.2 57.9

Share of non-jobchanging employees (former workplace (t − 1) = present workplace (t)) (d) 21.6 16.8 16.3 15.7 18.1 18.4 11.4 11.6 13.0 14.6 14.9 9.2 10.7 19.0

Share of job movers who were employed the year before (t − 1) not in the same firm (e)

Shares of job-changers and non-job-changers in total AVM employment in clusters

Osnabrück Duisburg Wuppertal Bonn Münster Bochum Kaiserslautern Mainz Karlsruhe Pforzheim Rosenheim Nuremberg Würzburg Augsburg

Cluster

Table 5.2

25.0 16.9 22.8 11.6 18.5 24.7 10.1 8.3 16.0 15.5 21.4 16.5 20.1 23.1

Share of persons who were in the year before (t − 1) not employed in a firm for different reasons (f)

114

G G EG EG G EG G EG G G G G G G EG G G

Cluster stage (a)

(continued)

Kiel Hamburg Krefeld Mülheim Cologne Rhein-Erft-Kreis Gelsenkirchen Hagen Baden-Baden Munich Passau Bayreuth Hof Erlangen Schwabach Aschaffenburg Kempten

Cluster

Table 5.2

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

Time period (b)

5 644 73 469 520 420 100 563 3 395 775 480 23 503 40 653 407 370 359 771 766 452 175

Total number of AVM employees over the period (c) 78.1 73.3 56.2 52.4 78.9 48.5 64.1 68.8 85.3 64.7 69.5 67.0 63.2 48.5 77.4 61.1 58.3

Share of non-jobchanging employees (former workplace (t − 1) = present workplace (t)) (d) 9.1 12.1 15.6 19.8 8.3 20.2 20.0 10.8 5.8 16.9 13.8 11.9 17.3 18.9 15.4 15.9 14.9

Share of job movers who were employed the year before (t − 1) not in the same firm (e)

12.8 14.6 28.3 27.9 12.8 31.3 15.9 20.4 8.9 18.4 16.7 21.1 19.5 32.6 7.2 23.0 26.9

Share of persons who were in the year before (t − 1) not employed in a firm for different reasons (f)

115 168 100 563 460 061 5 581

Growing clusters Emerging clusters

502 12 332 2 278 418 696 59 737 626 90 094

Minimum Maximum

3 3 3 3 3 3 3 3 465 642

G G G G G G G G

Total/average

Flensburg Region Hanover Münster Offenbach Trier Munich Kempten Berlin

71.0 55.2

45.4 85.3

70.9

45.4 71.7 69.3 51.4 53.9 64.5 65.2 61.3

13.8 18.3

5.8 25.4

13.9

23.3 12.8 13.2 25.4 17.7 20.5 20.1 18.4

15.1 26.5

7.2 32.6

15.3

31.3 15.5 17.6 23.2 28.4 15.0 14.7 20.3

116

Emerging clusters

such as Kaiserslautern, Kiel or Schwabach. Apart from these clusters in most of the AVM clusters a considerable part of the employment stock is made up by labour inflows, that is to say by job movers and by persons who started a new job for different reasons in the firms of the AVM clusters. On average, the shares of these two groups of employees are higher in emerging than in growing clusters. In the following, we confine our empirical analysis to the group of the job movers. 4.3

Intra- and Interregional Labour Mobility

With respect to Marshall’s labour market pooling approach it is expected that inter-firm job mobility inside clusters is strong due to labour pooling effects which facilitate search and screening processes of labour demand and supply. Thus, strong local job mobility should be observed especially in growing clusters because there is already a specific labour pool available. In this case, the occurrence of labour poaching among firms is also more likely. According to Storper and Walker (1989) the relevance of interregional labour inflows should be more important in emerging clusters for the formation of a human capital base (Thesis 1). The labour inflow rate is defined as the share of job movers in the total number of AVM employees in the respective cluster and time period. Table 5.3 depicts the intra- and interregional rates of labour inflow. The clusters are ranked in Table 5.3 according to the values of the intraregional inflow rates. The intra-regional inflow rates are differentiated whether the employee originates from the same district (Kreis) or from the same labour market region, but not from the same Kreis. Only in six clusters are the interregional inflow rates larger than the intra-regional ones. Thus, human resources deriving from other labour market regions play an important role in satisfying the local needs for human capital in these clusters (for example Mainz, Kaiserslautern, Krefeld). Labour mobility inside the respective labour market region of the cluster is in most cases stronger than the inflows of labour originating from other labour market regions (interregional mobility). There is one striking difference between the emerging and growing clusters: on average, the local mobility rate (job changes within the same Kreis in which the cluster is located) of growing clusters amounts to 6.5 per cent and is, therefore, higher than in emerging clusters (5.1 per cent). In turn, the remaining labour market region in which the emerging cluster is located is much more important as regional labour pool. The interregional mobility rates are also slightly higher in emerging clusters. It can be concluded that a major part of the job changes takes place inside AVM clusters at the local level – mostly originating from the same Kreis in growing clusters and

117

3 3 2 1 2 2 3 2 3 1 1 2 1 2 2 1

Stage Time period

18.9 17.9 15.8 15.3 14.8 14.3 14.2 14.0 13.9 13.9 13.4 13.3 12.9 11.8 11.4 11.3

Intra-regional inflow-rate (a+b)

Average rates of labour inflows in per cent

Offenbach G Flensburg G Erlangen G Bochum G Hof G Gelsenkirchen G Munich G Rhein-Erft-Kreis EG Kempten G Augsburg G Pforzheim G Schwabach EG Osnabrück G Munich G Kempten G Rosenheim G

Clusters

Table 5.3

5.0 14.5 7.0 12.7 13.1 9.3 6.8 4.5 11.3 13.2 11.6 9.4 11.3 4.9 6.3 8.9

Same Kreis (a)

13.9 3.4 8.8 2.6 1.7 5.0 7.4 9.5 2.6 0.7 1.8 3.9 1.6 6.8 5.1 2.4

Same labour market region, but not the same Kreis (b) 6.5 5.4 2.9 3.1 2.5 5.5 6.0 6.1 6.2 5.1 1.1 2.1 8.7 5.0 3.4 3.6

Other labour market region (inter-regional)

The former workplace of the job movers was situated . . .

12.4 12.5 13.0 12.2 12.3 8.8 8.3 7.9 7.7 8.8 12.3 11.2 4.2 6.8 8.0 7.7

Difference between intraand interregional inflow rates

118

G G G G G G G G G G G EG G G G EG

Berlin Passau Duisburg Aschaffenburg Bonn Wuppertal Trier Karlsruhe Münster Münster Hamburg Hagen Region Hanover Bayreuth Nuremberg Mülheim

3 2 1 2 1 1 3 1 1 3 2 2 3 2 1 2

Stage Time period

(continued)

Clusters

Table 5.3

11.2 10.8 10.8 10.6 10.3 10.3 10.2 9.2 8.4 8.3 8.1 7.9 7.8 7.3 6.9 6.4

Intra-regional inflow-rate (a+b)

11.0 6.6 10.2 8.4 4.7 9.9 8.3 7.1 7.0 7.2 7.5 6.5 7.4 4.3 5.2 3.1

Same Kreis (a)

0.2 4.2 0.6 2.2 5.6 0.3 1.9 2.2 1.3 1.1 0.5 1.5 0.4 3.0 1.7 3.3

Same labour market region, but not the same Kreis (b) 6.7 2.9 6.0 5.3 5.4 6.1 7.5 3.8 9.7 4.8 4.0 2.9 4.8 4.3 2.3 13.3

Other labour market region (inter-regional)

The former workplace of the job movers was situated . . .

4.5 7.9 4.8 5.3 4.9 4.2 2.7 5.5 −1.3 3.5 4.1 5.0 3.0 3.0 4.6 −6.9

Difference between intraand interregional inflow rates

119 18.9 1.4 8.7 11.9

Growing clusters Emerging clusters

5.9 4.8 4.2 3.9 2.9 1.9 1.4

Maximum Minimum

1 2 2 2 2 1 1 8.7

G G EG G G G G

Total/average

Würzburg Cologne Krefeld Kiel Baden-Baden Kaiserslautern Mainz

6.5 5.1

14.5 1.1

6.5

4.9 3.9 3.5 3.2 2.0 1.8 1.1

2.2 6.8

13.9 0.1

2.2

1.0 0.9 0.8 0.7 0.9 0.1 0.3

5.3 6.3

13.3 1.1

5.3

4.7 3.4 11.3 5.2 2.9 9.5 10.2

3.3 5.6

13.0 −8.8

3.4

1.2 1.4 −7.1 −1.3 0.0 −7.6 −8.8

120

Emerging clusters

to a larger extent deriving from the surrounding labour market region in emerging clusters. One might conclude from these outcomes that emerging clusters in fact do not possess the labour market pool themselves but need to attract labour from other neighbouring regions. The selected clusters are obviously capable of attracting the required workers. 4.4

Geographical Mobility of Highly-qualified Human Capital

Storper and Walker (1989) point to the relevance of the adaptability of the local labour market and the recruitment of personnel from other regions. It is expected that clusters satisfy their demand for high-qualified labour to a large extent on the extra-regional labour market. An important motivation for knowledge-intensive and creative cluster firms of the AVM industry to hire and poach talents consists of absorbing their incorporated strategic knowledge. Thereby, firms may enlarge and renew their knowledge base and in the end improve their innovation activities and competitiveness. In particular, in the initial stage of a cluster, the recruitment of highly-skilled employees from other labour market regions than the one the cluster is located in is of utmost importance for building a competitive and specialised human capital base (Thesis 2). Besides the theoretical approach of Storper and Walker (1989) labour market theory also sustains this thesis. Highly-qualified employees can make better use of job changes than low-skilled workers because they are more likely to obtain higher wages and in the end higher rents for their human capital (human capital theory). In addition, the regional density of job offers declines while the degree of regional specialisation increases. Therefore, highly-qualified employees need to enlarge their spatial radius of search for a new job (job search theory) (Mertens and Haas 2006). The information on the qualification level of each employee in the Historical Employment Database is incomplete. It provides information on the respective qualification (college/university degree, universityentrance diploma, intermediate school degree) for only 45 per cent of all job movers. We use instead a best-if indicator for the qualification level which is based on a differentiation between academic and non-academic occupations.1 The database provides the occupational code for 96.1 per cent of all job movers (Table 5.4). To identify academic and non-academic occupations we computed the share of employees with a university or college degree in each single occupation for Germany in 2006. The 43 occupations (out of 337) with the highest shares were subsumed under the rubric of academic occupations. The minimum share of employees with a university or college degree in total employment of one occupation amounts to 30 per cent. Occupations with a lower share of highly-qualified

Origins of human capital in clusters

Table 5.4

121

Ranking of most important occupations in AVM industry (academic vs non-academic background)

Academic background

No.

Cumu- Non-academic lative (%) background

821 Publicists 11 410.9 751 Chief executive 2 933.6 officer

48.5 61.0

832 Descriptive artists 831 Musicians 602 Electrical engineers

2 521.7

71.7

1 312.9 1 114.0

77.3 82.1

823 Librarians, archivists, museum experts 762 Executive administrators

946.0

86.1

944.1

90.1

774 Data processing experts 607 Specific engineers 752 Consultants

845.7

No.

Cumulative (%)

781 Office clerks 12 077.7 835 Artistic and 11 403.1 attributive occupations 782 Shorthand 4 497.4 writers, typists 837 Photographers 2 225.7 622 Electrical 2 210.4 engineering technicians 682 Salesmen 1 741.3

22.7 44.2

1 161.9

66.5

93.7

706 Ticket sellers and supervisors 982 Trainees

993.0

68.4

306.3

95.0

628 Technicians

957.0

70.2

299.0

96.3

72.0

881 Economic and social scientists

173.3

97.0

Others

702.0

100.0

784 Non-skilled 943.7 office clerks 923.4 833 Visual artists, graphic artists, designers Others 4 0377.7

Total

23 509.4

Total

52.7 56.9 61.1

64.3

73.7

100.0

53 091.7

employees are classified as non-academic occupations. The 43 academic occupations identified account for 65 per cent of all high-qualified employees in Germany in 2006. Table 5.4 depicts the respective academic and non-academic occupations within the AVM industry in Germany (2006). The occupations with academic background are mainly influenced by publicists who account for nearly 50 per cent of all employees. Taking a closer look at the occupational distribution in the ten core

122

Emerging clusters

clusters we find that around 30 per cent of the occupations in these regions have an academic background while 70 per cent have not. There is, however, a strong variance, too: the share of academic occupations in total AVM employment in Baden-Baden (41 per cent) exceeds the value of Bonn (23 per cent) by a factor of nearly 2.0 (Table 5.5). This points to structural differences in the composition of the respective clusters. In the following, the figures refer to averaged values over all clusters and time periods because significant differences between the clusters could not be observed. All in all, the academics account for 20 per cent of all the job movers whereas 80 per cent of the job movers work in non-academic occupations. The large part of the former workplaces of the academics (48 per cent) was situated outside of the labour market region in which the respective AVM cluster is located (Table 5.6). In contrast, 66 per cent of the non-academic job movers have worked before within the same labour market region. Thus, the academics exhibit a higher geographical mobility. This is the case in emerging and particularly in growing clusters. These results confirm the assumption outlined above. Clusters in the emerging and growing phase rely to a large extent on the inflow of highly-qualified employees. 4.5

Industrial Origins of Labour Inflows

According to the literature the activities of emerging clusters frequently take place in new technologies, competence fields and industries (Menzel and Fornahl 2007). Thus, it is most likely that there is no specialised labour pool available in emerging clusters (Thesis 3). In this respect, it should be expected that a lot of employees in the emergent stage stem from other industries whereas growing clusters have better access to an already existing industry-specific workforce. The AVM industries, however, are not really new industries. Moreover, this industry experienced significant intra-industry changes due to substantial changes of the institutional, economic and technological environment (Mossig 2004). Thus, there are available labour pools in this industry to satisfy the demand of emerging and growing clusters. This is shown in Table 5.7 which depicts in which industries the workplaces of the job movers were situated one year before (t − 1). It can be observed that the major part of job changers originates in emerging as well in growing clusters from the AVM industries itself. Otherwise most of the remaining job movers did not originate from other related media industries, instead they stem from the remaining service industries. In emerging clusters the role played by employees originating from manufacturing industries is more important than in growing clusters. These industrial transitions, therefore, reveal only slight differences between the growing and emerging phase.

123 27

73

33

67

Share of employees with academic background (%) Share of employees with non-academic background (%)

2

Hamburg

2

Kiel

70

30

3

Region Hanover

77

23

1

Bonn

65

35

2

Cologne

71

29

2

RheinErft-Kreis

AVM clusters

72

28

1

Mainz

Average shares of academic and non-academic occupations in core AVM clusters

Time period

Table 5.5

59

4

2

BadenBaden

72

28

3

Munich

73

27

3

Berlin

124

Emerging clusters

Table 5.6

Geographic origins of the former workplaces of academic and non-academic job movers*

Qualification

Emerging clusters Academic professions Non-academic professions Growing clusters Academic professions Non-academic professions Total Academic professions Non-academic professions Note:

4.6

Geographic origin of the former workplace of the job movers Same Kreis

Same labour market region, but not the same Kreis

Other labour market region (interregional)

No data information available

Total

16.3

43.3

39.5

0.8

100.0

45.3

28.2

26.4

0.2

100.0

38.3

11.2

48.5

2.0

100.0

48.5

17.0

33.5

1.0

100.0

38.0

11.7

48.4

2.0

100.0

48.5

17.3

33.3

1.0

100.0

* Average percentage values for all selected clusters over all time periods.

Mobility Patterns and Cluster Structure

The internal structure of a cluster determines the patterns of inter-firm labour mobility. Markusen (1996) introduced three different types of an industrial district which are accepted and applied in the literature as emblematic examples for the varying structures of clusters. In this respect, a hub-and-spoke district and a satellite platform are rather fragmented clusters whereas a Marshallian district represents a more integrated type due to mutual and dense relationships between the prevalent SME’s. The hub-and-spoke district corresponds to Storper and Walker’s (1989) vertically integrated district with a large focal enterprise. This district type seems to be most appropriate in order to characterise the majority of our selected AVM clusters. This is because many clusters are dominated by

Origins of human capital in clusters

Table 5.7

125

Industrial origins of job movers in AVM clusters

Average shares in %

Total (all clusters and time periods, (n = 65 000 job movers) Minimum of all clusters Maximum of all clusters Emerging clusters (average) Growing clusters (average)

Audiovisual media industries

Remaining media industries

Remaining services

Manufac- Primary turing sector industries

48.1

5.7

37.8

7.0

0.1

9.5

0.0

11.0

1.2

0.0

72.3

20.4

70.8

51.6

4.0

47.3

3.7

35.7

12.5

0.6

48.1

5.8

37.9

6.9

0.1

either a public or a private television station. Table 5A.5 in the Appendix shows whether a large television and broadcasting station is located within the selected clusters. These stations form the hub within the respective clusters. It is expected that labour flows are directed towards the hub (core) of a cluster (Thesis 4). There are several reasons for this thesis which are reasonably based on the difference in firm size between the hub (core) and the periphery of the cluster. Large firms and institutions offer more favourable job conditions and prospective job opportunities than small- and mediumsized firms. For instance, large firms provide higher wages, better career prospects, a higher job security, better possibilities for education and so on. Hence, larger firms have more power to attract adequate and highly-skilled people. In the following, the directions of labour inflows are differentiated between small- and medium-sized firms and large enterprises. The biggest firms, namely all firms with more than 250 employees, are the most important employers within the selected AVM clusters. They account for 56 per cent of all AVM employment when the numbers of AVM employees are summed up over all clusters and time periods (Table 5.8). The employment shares of the small- and medium-sized firms amount to 22 per cent and 20 per cent, respectively. On average, the inflow rate of labour amounts to 19 per cent for small- and for medium-sized firms. In other words, 19 per cent of the total workforce within these firms (in year t) have worked in the year before (t − 1) in another firm. The labour inflow-rate for the large firms is lower (9 per cent). This result confirms the

126

261 812 61 470 49 452 150 890

6 095

Large firms 251–500 employees 501–1000 employees 1001 employees and more

No data information available

465 642

94 006 40 358 53 648

Medium-sized firms 51–100 employees 101–250 employees

Total

103 729 22 765 16 407 23 509 41 048

Small firms 1–5 employees 6–10 employees 11–20 employees 21–50 employees

100.0

1.3

56.2 13.2 10.6 32.4

20.3 8.7 11.5

22.3 4.9 3.5 5.0 8.8

64 557

1 574

18 397 5 757 6 913 5 727

18 252 7 773 10 479

26 334 6 835 4 286 5 459 9 754

100.0

2.4

28.5 8.9 10.7 8.9

26.9 12.0 16.2

35.9 10.6 6.6 8.5 15.1

%

Abs.

Abs.

%

Labour inflow – wherefrom?

Number of all employees in the respective year

64 557

1 504

23 329 8 835 6 352 8 142

18 121 7 453 10 668

21 603 4 507 3 315 4 665 9 116

Abs.

100.0

2.3

36.1 13.7 9.8 12.6

26.8 11.5 16.5

29.4 7.0 5.1 7.2 14.1

%

Labour inflow – whereto?

0.0

−0.1

4.9 3.1 −0.6 2.4

−0.1 −0.3 0.2

−4.4 −2.3 −1.0 −0.8 −0.6

Abs.

0.0

−0.1

7.6 4.8 −0.9 3.7

−0.1 −0.5 0.3

−6.4 −3.6 −1.5 −1.2 −1.0

%

Difference between whereto and wherefrom

13.9

24.7

8.9 14.4 12.8 5.4

19.3 18.47 19.89

19.4 19.8 20.2 19.8 22.2

Labour inflow rate = (number of job movers/all employees) × 100

Employment, inflow rates of labour and directions of labour flows between small, medium and large firms over all years and time periods in all selected AVM clusters

Firm size

Table 5.8

Origins of human capital in clusters

127

aforementioned expectation that in-house job changes are more likely to occur in large enterprises by virtue of more favourable job conditions. A closer look at the directions of the labour flows in Table 5.8 reveals that 36 per cent of all job movers have begun their new job in a large enterprise (in year t) whereas only 29 per cent of them have worked in the year before (t − 1) in a large firm. The reverse holds especially true for the small firms: 36 per cent of the job movers worked beforehand in a small firm. Only 29 per cent of the job movers started in a small firm after having changed their workplace. The labour flows are apparently oriented towards the hubs of the AVM clusters. All in all, it seems that the hub-and-spoke cluster of most of the AVM clusters shapes in part the patterns of labour mobility. A comparison of the high-tech regions Cambridgeshire and Oxfordshire revealed that local job turnover of high-qualified employees is low due to in-house job changes in the hubs (large firms and public institutions) of Oxfordshire (Lawton Smith and Waters 2005). In contrast, this study shows a high level of job turnover in Cambridgeshire which is characterized by large numbers of SMEs. Job hopping in the Motorsport Valley is fostered by a high local density of small firms, too (Henry and Pinch 2000). The study of Longhi (1999) evidenced that a local labour pool which could be mobilized did not develop within the high-tech cluster of Sofia Antipolis. It is because cluster firms internalized human resources. Besides, large firms have always tried to capture the highly-skilled employees when they started their business in the high-tech park. This led to a reverse spin-off process. Small firms particularly might profit from the access to the regional labour pool which mainly consists of experienced workers with clusterspecific knowledge and competencies. In contrast to large firms, they are more likely to concentrate their search processes for new employees onto the regional and not onto the national or international labour market because of their limited financial resources. Additionally, search processes in small firms are normally unsystematic because there is no extra staff available to focus on these activities (Otto 2005). Yet another advantage of being located within a cluster is the access to high numbers of already trained employees. It is because in-house training, especially of job beginners and inexperienced workers, is too costly for small firms. Training on the job is, therefore, very important for small firms in order to satisfy their qualification-specific needs in terms of human capital (Angel 1991). All in all, small firms are expected to hire employees primarily from the local labour market pool whereas large firms may employ to a larger extent extra-regional personnel (Thesis 5). Table 5.9 shows the mean shares of the region-specific origins of job movers differentiated by firm size. As expected, the labour inflows into small firms originate to a large extent from the intra-regional labour

128

68.3 74.0 72.2 64.0 66.2 59.3 59.3 59.4 54.5 66.5 64.2 33.8 80.4

Medium-sized firms 51–100 employees 101–250 employees

Large firms 251–500 employees 501–1000 employees 1001 employees and more

No data information available

intra-regional (a+b) (in per cent)

70.7

36.4 43.7 39.6 26.1

43.8 47.5 41.3

54.5 59.9 56.7 50.3 53.2

same Kreis (a) (in per cent)

9.6

18.0 22.8 24.6 7.8

15.5 11.7 18.1

13.8 14.1 15.5 13.7 13.1

Same labour market region, but not the same ‘Kreis’ (b) (in per cent)

19.4

44.0 31.7 33.5 65.6

39.4 39.4 39.4

30.7 25.3 26.8 35.2 32.5

Inter-regional: other labour market region (in per cent)

0.2

1.5 1.8 2.4 0.5

1.3 1.3 1.2

1.0 0.7 0.9 0.8 1.3

No data information available (in per cent)

Average shares of the region-specific origins of job movers differentiated by firm size over all years and time periods in all selected AVM clusters

Small firms 1–5 employees 6–10 employees 11–20 employees 21–50 employees

Firm size

Table 5.9

Origins of human capital in clusters

129

market. The interregional labour market is more important for the large firms. On average, 44 per cent of all job movers to the large firms derive from the extra-regional labour pool while only 31 per cent of the labour inflows into small firms stem from outside of the labour market region.

5

CONCLUSIONS

The main results reveal that apparently the mobility patterns of job movers are shaped by different factors which are situated at the firm, regional and industrial level. All in all, we were able to confirm most of our hypothesis. There are some outstanding results: in most of the AVM clusters a significant part of the employment stock is made up by job movers and persons which started a new job in the AVM clusters for different reasons (motherhood protection, education, unemployed). Thus, the emerging and growing phase is characterized by intense inflows of labour into the local labour markets. This phenomenon is even more apparent in emerging clusters. Interregional mobility rates are in most of the AVM clusters lower than intra-regional mobility rates. Moreover, it is local mobility which is the strongest within clusters. The labour inflows into emerging clusters stem to a larger extent from the same labour market region in which the cluster is situated, but not from the Kreis (local level) itself. This holds especially true for growing clusters which – in comparison to emerging clusters – source a large share of employees from the region itself. But there are extra-regional flows as well: especially inflows of high-qualified human capital originate to a large extent from extra-regional labour pools. With regard to the sectoral origin we found that labour inflows into emerging and growing clusters stem either from the AVM industries themselves or from other services industries. The local industrial structure affects labour mobility, too. The empirical results indicate that many labour flows are oriented towards the large enterprises within AVM clusters. These hubs are mainly local broadcasting and television stations. Employees in small firms are more likely to change their job. Thus, small firms suffer particularly from ‘losing’ embodied knowledge via job mobility to other firms. The internal labour market of large enterprises fosters in-house changes of the workplace. Small firms focus on the regional labour market when recruiting personnel, whereas large firms hire more frequently new workers deriving from the extra-regional labour pool. Thus, the structure of a cluster whether it is fragmented (for example hub and spoke-cluster) or more integrated, shapes the patterns and intensity of job mobility. It can be observed that the level and portfolio of labour inflows vary significantly among the selected AVM clusters. In this respect, the analysis

130

Emerging clusters

of the relevance of region-specific characteristics, apart from the cluster structure, is expected to contribute to the explanation of the variance of these mobility patterns.

NOTES 1. The construction of this best-if indicator is based on another set of employment statistics, the so-called quarterly employment data base, which covers to a high degree the information about the qualification of the employees. The difference between these two employment statistics is in the procedure of counting jobs. The historic employment statistics refer only to a person’s main job whereas the quarterly statistics count all the part-time jobs of a person.

REFERENCES Amend, E. and P. Herbst (2008): Labor Market Pooling and Human Capital Investment Decisions. IAB Discussion Paper, 04/2008, Nuremberg. Angel, D.P. (1991): High-technology agglomeration and the labor market: the case of Silicon Valley. Environment and Planning A, 23, 1501–16. Birch, D.L. (1987): Job Creation in America: How our smallest companies put the most people to work. New York: Macmillan, Free Press. Breschi, S. and F. Lissoni (2001): Knowledge Spillovers and Local Innovation Systems: a Critical Survey. Industrial and Corporate Change, 10(4), 975–1005. Carnoy, M., M. Castells and C. Benner (1997): Labour market and employment practices in the age of flexibility: a case study of Silicon Valley. International Labour Review, 136, 27–48. Combes, P.-P. and G. Duranton (2001): Labor Pooling, Labor Poaching, and Spatial Clustering. University of London, Centre for Economic Policy Research (CEPR), Discussion Paper No. 2975. Dahl, M.S. (2002): Embedded Knowledge Flows through Labour Mobility in Regional Clusters in Denmark. Working Paper at the DRUID Summer Conference. Aalborg. De Blasio, G. and S.L. Di Addario (2005): Do workers benefit from industrial agglomeration? Journal of Regional Science, 45, 797–827. Eckey, H.-F., R. Kosfeld and M. Türck (2006): Abgrenzung deutscher Arbeitsmarktregionen. Discussion Papers in Economics from University of Kassel, Institute of Economics, Nr 81/06. Fallick, B., Fleischmann, C.A. and J.B. Rebitzer (2005): Job-Hopping in Silicon Valley: Some Evidence Concerning the Micro-Foundations of a High Technology 23 Cluster. Finance and Economic Discussion Series Division of Research & Statistics and Monetary Affairs Federal Reserve Board, Washington, DC. Feldman, M.P. (2001): The entrepreneurial event revisited: firm formation in a regional context. Industrial and Corporate Change, 10(4), 861–91. Frank, B. (2008): Location decisions in a changing labour market environment. Jahrbuch für Regionalwissenschaft, 28, 31–42.

Origins of human capital in clusters

131

Franz, W. (2006): Arbeitsmarktökonomik. Heidelberg: Springer. Gordon, I. and P. McCann (2000): Industrial clusters: Complexes, agglomeration and/or social networks? Urban Studies, 37, 513–32. Henry, N. and S. Pinch (2000): (The) industrial agglomeration (of Motor Sport Valley): a knowledge, space, economy approach. In J. Bryson, P. Daniels, N. Henry and J. Pollard (eds): Knowledge Space Economy. London: Routledge, pp. 120–41. Klöpper, C. (2005): Zwischen kontinuität und transformation: die evolution des Paderborner It-clusters. Geographica Helvetica, 60(2), 105–14. Krugman, P. (1991): Geography and Trade. Leuven: Leuven University Press. Lawton Smith, H. and R. Waters (2005): Rates of turnover in high-technology agglomerations: knowledge transfer in Oxfordshire and Cambridgeshire. Area, 37, 189–98. Longhi, C. (1999): Networks, collective learning and technology development in innovative high technology regions: the case of Sophia Antipolis. Regional Studies, 33(4), 333–42. Markusen, A. (1996): Sticky places in slippery space: a typology of industrial districts. Economic Geography, 72(2), 294–314. Marshall, A. (1890): Principles of Economics. London: Macmillan. McGahan, A.M. and B.S. Silverman (2001): How does innovative activity change as industries mature? International Journal of Industrial Organization, 19, 1141–60. Menzel, M.-P. and D. Fornahl (2007): Cluster Life Cycles – Dimensions and Rationales of Cluster Development. Jena Economic Research Papers #2007-076. Mertens, A. and A. Haas (2006): Regionale arbeitslosigkeit und arbeitsplatzwechsel in Deutschland. Eine analyse auf kreisebene. Jahrbuch für Regionalwissenschaft, 26, 147–69. Mossig, I. (2004): The networks producing television programmes in the Cologne Media Cluster (Germany): new firm foundation, flexible specialisation and efficient decision-making structures. European Planning Studies, 12, 155–71. Otto, A. (2005): Wissens- und technologieintensive dienstleistungsgründungen in West- und Ostdeutschland. Zeitschrift für Wirtschaftsgeographie, 49, 200–218. Porter, M. (1998): Clusters and the new economics of competition. Harvard Business Review, No. 11–12, 77–90. Power, D. and M. Lundmark (2004): Working through knowledge pools: labour market dynamics, the transference of knowledge and ideas, and industrial clusters. Urban Studies, 41, 1025–44. Sternberg, R. and T. Litzenberger (2004): Regional clusters in Germany – their geography and their relevance for entrepreneurial activities. European Planning Studies, 12, 767–91. Storper, M. and R. Walker (1989): The Capitalist Imperative: Territory, Technology, and Industrial Growth. Cambridge, MA: Basil Blackwell. Weber, A. (1909): Theory of the Location of Industries. 2nd edition. Chicago: University of Chicago Press. Zellner, C. and D. Fornahl (2002): Scientific knowledge and implications for its diffusion, Journal of Knowledge Management, 6, 190–98. Zucker, L.G., M.R. Darby and J. Armstrong (1998): Geographically localised knowledge spillovers or markets? Economic Inquiry, 36, 65–86. Zucker, L.G., M.R. Darby and M. Brewer (1998): Intellectual human capital and the birth of US biotechnology enterprises. American Economic Review, 88, 290–306.

132

Emerging clusters

APPENDIX 5A Table 5A.1

Definition of the audio-visual media industry

WZ 73

WZ 93*

WZ 2003*

763 Cinemas, film production and distribution 764 Radio and television broadcasting corporations 765 Own-account artists and performers

921 Motion picture and video activities

921 Motion picture and video activities

922 Radio and television activities

922 Radio and television activities

92317 Activities of ownaccount stage, motion picture, radio and television artists 92325 Technical activities in support of cultural and entertaining services

92317 Activities of ownaccount stage, motion picture, radio and television artists

Note:

* Corresponding to ISIC Rev. 3.1.

92325 Technical activities in support of cultural and entertaining services

133

Growth – – – Growth – – Emergence – Growth – Growth – Growth – Growth Growth – – – Emergence – Growth

5314 Bonn 5315 Cologne 5362 Rhein-ErftKreis 5513 Gelsenkirchen 5515 Münster 5911 Bochum 5914 Hagen

5124 Wuppertal

5117 Mülheim

– Growth Growth – Growth – – Growth – – – – Emergence – Growth – Emergence – Growth Growth – –

Flensburg Kiel Hamburg Hanover Osnabrück Duisburg Krefeld

– – – Growth Growth Growth –

1999– 2007

1001 1002 2000 3241 3404 5112 5114

1980–89 1990–98

Cluster stage

Name

Flensburg Kiel Hamburg Hanover Osnabrück Duisburg Krefeld 15.8

41.0 354.5 5 721.8 875.5 48.3 137.0 31.0

1980– 89

1999– 1980– 1990– 2007 89 98

44.1

88.2

5.5

9.4

24 30 24 26

Essen Münster Essen Wuppertal/ Hagen

49.0 150.3 150.0 38.0

66.6 179.8 231.2 45.6

65.2 259.4 887.8 40.3

13.0 26.0 20.0 10.8

12.3 15.2 23.6 11.4

CI (Mean over time frame)

6.8 19.8 19.9 5.4

41.2 329.3 42.3

16.1

10.7

4.4 29.1 343.8 72.1 9.0 13.8 12.9

1.5 3.6 3.8 0.9

34.9 209.0 0.1

2.1

0.4

3.7 14.4 89.0 1.8 2.2 2.0 0.6

1.9 2.4 6.5 1.1

38.3 346.6 2.3

1.8

1.8

1.9 30.9 127.1 2.3 2.5 2.7 1.1

0.9 3.5 17.0 0.4

59.5 623.7 11.0

2.4

3.4

1.8 39.6 131.3 2.1 1.9 0.9 2.1

1999– 1980– 1990– 1999– 2007 89 98 2007

AVM companies (Mean over time frame)

24.0 34.6 6.0 5.4 583.8 687.6 16.0 22.1 7490.0 9 145.9 256.0 311.2 1143.0 1 366.4 69.3 72.6 66.0 78.0 12.0 10.8 173.8 133.0 25.5 27.2 51.3 97.0 7.5 10.0

1990– 98

AVM employees (Mean over time frame)

26 Wuppertal/ 91.0 103.3 163.3 20.0 16.8 Hagen 28 Cologne/Bonn 418.5 452.3 1 228.9 46.5 51.7 28 Cologne/Bonn 7 878.8 10 284.3 13 708.1 138.0 187.0 28 Cologne/Bonn 25.3 341.7 1 537.9 17.5 24.7

24 Essen

1 2 6 12 16 23 25

No. Name

Labour market region

Selected industrial clusters in the audio-visual media industry

No.

Table 5A.2

134

Aschaffenburg Würzburg Augsburg Kempten Berlin

9661 9663 9761 9763 11000

– Growth Growth – –

– Growth Growth – Growth Growth Growth – – – – – Growth –

Trier Kaiserslautern Mainz Baden-Baden Karlsruhe Pforzheim Rosenheim Munich Passau Bayreuth Hof Erlangen Nuremberg Schwabach

7211 7312 7315 8211 8212 8231 9163 9184 9262 9462 9464 9562 9564 9565

Growth

No. Name

Labour market region 1980– 89 44.6

4.2

45.7 112.3 4.3 7.4 108.1 168.4 11.8 11.3 136.1 208.6 22.3 20.0 18.1 57.1 5.3 4.8 7 307.1 11 502.4 329.3 369.8

13.3 11.1 12.3 12.1 23.3 18.7 10.0 14.3 21.8 21.2 9.8 4.7 8.5 9.1 94.3 121.1 5.8 6.7 3.3 3.0 5.0 6.8 6.5 8.2 42.0 38.9 1.3 1.6

8.5

CI (Mean over time frame)

7.1 13.3 22.9 9.4 655.2

7.1 5.7 28.9 19.2 21.7 4.3 6.3 169.3 6.7 4.4 7.6 5.2 39.4 1.9

8.8

2.0 6.7 5.2 2.4 691.7

4.6 4.9 385.0 224.6 4.8 3.3 8.9 101.3 2.1 1.1 2.7 1.1 10.3 0.1

3.6

5.8 7.5 4.9 1.5 297.5

2.8 5.6 366.5 314.5 5.2 1.1 12.2 196.5 5.8 1.6 5.8 6.4 11.0 5.9

1.2

10.1 10.7 6.9 7.7 169.0

2.4 2.8 565.7 483.4 5.3 0.7 9.2 360.4 5.9 3.5 10.8 3.0 17.1 0.3

4.1

1999– 1980– 1990– 1999– 2007 89 98 2007

AVM companies (Mean over time frame)

1999– 1980– 1990– 2007 89 98

42.0 71.0 93.3 122.9 5 211.2 6 626.1 2 331.3 3 465.6 174.0 215.0 30.7 31.6 41.9 60.3 4 164.3 7 869.3 42.2 56.6 34.3 70.6 32.0 74.9 86.1 71.9 385.9 731.2 58.4 3.7

20.3

1990– 98

AVM employees (Mean over time frame)

41 Frankfurt am 31.0 Main – Growth 51 Trier 53.0 – – 53 Kaiserslautern 75.3 – – 55 Mainz 4 023.3 Growth – 63 Karlsruhe 2 093.8 – – 63 Karlsruhe 141.0 – – 65 Pforzheim 51.0 – – 84 Rosenheim 28.0 Growth Growth 83 Munich 2 499.5 Growth – 89 Passau 16.3 Growth – 98 Bayreuth 23.3 Growth – 100 Hof 21.8 Growth – 102 Nuremberg 17.8 – – 102 Nuremberg 307.5 Emergence – 102 Nuremberg 2.0 Growth Growth – 103 Aschaffenburg 23.5 – – 105 Würzburg 86.3 – – 106 Augsburg 119.5 Growth Growth 108 Kempten 22.5 – Growth 113 Berlin 5243.0

–

1999– 2007

Cluster stage

1980–89 1990–98

–

Name

(continued)

6413 Offenbach

No.

Table 5A.2

135

Note:

1 0.6048* 0.5595* 0.7442* 0.6122* 0.6440* 0.6636* 0.7356* 0.6911*

Hamburg

1 0.5450* 0.6211* 0.6134* 0.5894* 0.6025* 0.6070* 0.6293*

Region Hanover

1 0.5415* 0.5421* 0.5673* 0.6075* 0.5114* 0.5247*

Bonn

1 0.6138* 0.6887* 0.6923* 0.7655* 0.6860*

Cologne

1 0.5522* 0.6714* 0.6168* 0.6096*

RhineErft

1 0.7126* 0.6426* 0.6055*

Mainz

* Spearman rank correlation coefficients reported in cells; all correlations significant on the 0.01 level.

1 0.6072* 0.6886* 0.5469* 0.5684* 0.6148* 0.6167* 0.5360* 0.5663* 0.5437*

Kiel

Correlation of job rankings between prominent cluster regions

Kiel Hamburg Region Hanover Bonn Cologne Rhine-Erft Mainz Baden-Baden Munich Berlin

Table 5A.3

1 0.6370* 0.6021*

BadenBaden

1 0.7350*

Munich

1

Berlin

136

Emerging clusters

Table 5A.4

Number and share of most important AVM jobs in core clusters

Region Office clerks

Publicists

Artistic Shorthand Entrepreand writers, neurs, attributive typists Directors, occupations Division managers

Germany

Number Share (%)

12 077.71 15.58

11 410.86 14.72

11 403.14 14.71

4 497.43 5.80

2 933.57 3.78

01002 Kiel

Number Share (%)

75.5 13.28

137 24.10

71.5 12.58

49 8.62

11.5 2.02

02000 Hamburg

Number Share (%)

1 217.5 16.84

856.5 11.85

1 301.5 18.00

624.5 8.64

337.5 4.67

03241 Hanover

Number Share (%)

150.67 11.84

235.00 18.46

178.33 14.01

83.33 6.55

25.67 2.02

05314 Bonn

Number Share (%)

41 10.65

53.5 13.90

73.5 19.09

36 9.35

17 4.42

05315 Cologne

Number Share (%)

1 588.5 15.48

1 965.5 19.15

1 224.5 11.93

1 143.5 11.14

307.5 3.00

05362 RheinErft-Kreis

Number Share (%)

69 15.05

42 9.16

81.5 17.78

15 3.27

23.5 5.13

07315 Mainz

Number Share (%)

717.5 19.93

472 13.11

485 13.47

378.5 10.51

68.5 1.90

08211 BadenBaden

Number Share (%)

275.5 11.95

434.5 18.84

152.5 6.61

267.5 11.60

122 5.29

09184 Munich

Number Share (%)

1 322.4 21.27

628.2 10.10

1 223.2 19.67

141.2 2.27

281.2 4.52

11000 Berlin

Number Share (%)

1 686.00 15.20

1 542.33 13.91

407.82 3.68

362.33 3.27

501.33 4.52

Note:

* Numbers in bold indicate above average importance.

Origins of human capital in clusters

137

Jobs Descriptive artists

Photographers

Electrical engineering technicians

Salesmen

Musicians

Others

2 521.71 3.25

2 225.71 2.87

2 210.43 2.85

1 741.29 2.25

1 312.86 1.69

25 182.43 32.49

3.5 0.62

51 8.97

23 4.05

27.5 4.84

0 0.00

119 20.93

166.5 2.30

217 3.00

149 2.06

77.5 1.07

131.5 1.82

2 150 29.74

4.67 0.37

59.00 4.63

17.67 1.39

15.00 1.18

76.00 5.97

427.67 33.60

0 0.00

34.5 8.96

13.5 3.51

5 1.30

0 0.00

111 28.76

206.5 2.01

216 2.10

575 5.60

38.5 0.38

187.5 1.83

2 810 27.38

30 6.54

17.5 3.82

4 0.87

21 4.58

0 0

26.5 0.74

197 5.47

188 5.22

5.5 0.15

2 0.06

1 060 29.44

40 1.73

98 4.25

215 9.32

1 0.04

100 4.34

600 26.02

302.8 4.87

145.6 2.34

84.4 1.36

11.6 0.19

12.4 0.20

2,065.2 33.21

599.33 5.40

286.00 2.58

132.00 1.19

327.67 2.95

14.67 0.13

5 229.18 47.16

155 33.81

138

Emerging clusters

Table 5A.5 No.

Large AVM firms in clusters

Name

Large AVM firms

1001

Flensburg

1002

Kiel

2000

Hamburg

3241 3404 5112 5117

Hanover Osnabrück Duisburg Mülheim

5124

Wuppertal

5314

Bonn

5315

Cologne

5515 5911 6413 7312 7315

Münster Bochum Offenbach Kaiserslautern Mainz

8212 8231 9163

Karlsruhe Pforzheim Rosenheim

9184

Munich

NDR Studio Flensburg; OK TV Flensburg; Videomaxx NDR Landesfunkhaus S.-H.; RTL Nord GmbH; TVN-Televisions-, Programmu. Nachrichtengesellschaft mbH; Sat.1 Norddeutschland GmbH Landesstudio S.-H. ZDF; moreTV Broadcasting GmbH; LTV Live Television GmbH; Valdor Frank Promotions Musikwerbung KG; Premiere Fernsehen GmbH & Co KG Serviceline; Cinemaxx AG TVPlus GmbH BFBS Radio und TV-Broadcasting Studio 47 – Stadtfernsehen Duisburg Kabel-TV Medien Service Gesellschaft für Breitbandkabel mbH WDR Studio Wuppertal; Arte Media Medienproduktion GmbH Bonnavista TV GmbH; Deutsche Fernsehnachrichten Agentur GmbH; PHOENIX (TV) Faction-TV GmbH; WDR Cologne; Park Fernsehen GmbH; RTL Television GmbH WDR UCI Multiple GmbH; WDR Team HR-Service GmbH SWR ZDF Newmedia GmbH; SWR Funkhaus Mainz; ZDF Enterprises GmbH; News and Pictures Fernsehen GmbH & Co KG; ZDF ERB Medien GmbH; SWR SWR Korrespondentenbüro BR Korrespondentenbüro; RFO Regional Fernsehen Oberbayern Silverline Television AG; Telepool GmbH; Max TV Media GmbH; Bayrischer Rundfunk (BR); Eurosport Media GmbH; RTL-Plus Deutschland Fernsehen GmbH; Team 72 Film- und Fernsehservice GmbH; Faktor 3 TV; Medienpool TV GmbH; Sunset Television GmbH; Deutsches Wetterfernsehen;Beate Uhse TV GmbH; Sat.1 Regionalprogram Bayern

Origins of human capital in clusters

Table 5A.5 No.

139

(continued)

Name

Large AVM firms

9262 9462

Passau Bayreuth

9464 9564 9661 9663 9761

Hof Nuremberg Aschaffenburg Würzburg Augsburg

Tele Regional Passau 1 GmbH BR Korrespondentenbüro; TV Oberfranken GmbH & Co KG TV Oberfranken; Medienhaus Zentrale BR Shop; Studio Franken Main-TV BR; TV Touring Fernsehgesellschaft mbH BR Fuggerstadtcenter; Augsburger Fernsehfenster GmbH & Co. Studiobetriebs-KG BR; Outpro TV GmbH; TV Allgäu Fuji TV; CineCentrum Berlin; Berlin 1 Fernseh-Beteiligungs GmbH; OK Berlin; Beate Uhse TV GmbH & Co. KG Fernsehsender; Schmidt u. Paetzel Fernsehfilme; SCALA Fernsehprogrammproduktion; Job TV-24 GmbH;

9763 11000

Kempten Berlin

6.

The co-evolution of ICT, VC and policy in Israel during the 1990s Gil Avnimelech and Morris Teubal

1

INTRODUCTION AND OBJECTIVES

The economic success of well-known innovative clusters such as Silicon Valley has fostered attempts to transform local economies into successful clusters (Feldman et al., 2005). In the last decade many studies (such as Feldman, 2001; Avnimelech and Teubal, 2004, 2006; Brenner, 2004; Bresnahan and Gambardella, 2004; Feldman et al., 2005; Breznitz, 2008; Menzel and Fornahl, 2009) attempt to explain how an innovative regional agglomeration may develop into a global leading cluster. However, there is still a need to extend our understanding of this process. This is mainly because many studies in the field often analyse the processes taking place within established clusters and based on that analysis attempt to draw conclusions about cluster emergence processes (Maskell, 2001). While such an approach is useful for characterizing the operation of mature clusters the analysis of cluster development and emergence requires a more dynamic approach (Breschi and Malerba 2001; Bresnahan et al., 2001; Feldman, 2001). On the other hand, many studies which do explore into the initial development stages of clusters often tend to be quite descriptive in nature. Recently, a few researchers introduced dynamic cluster development models (such as Bresnahan et al., 2001; Feldman, 2001; Feldman et al., 2005; Avnimelech and Teubal, 2006; Menzel and Fornahl, 2009; and others). However, more analytical case studies are needed in order to enable the development of more generalized theories and to expand the scope of profiles of cluster development that are identified. This will enable more case-specific policy implications. This chapter focused on the third – emergence – phase of Avnimelech and Teubal’s (2006) four-phase start-up-intensive high-tech cluster life cycle model. While there are many regions that can be defined as latent clusters (regions that fulfilled the so-called background conditions for eventual cluster emergence) and a non-insignificant number of regions also present temporary rapid growth in economic activities in specific 140

The co-evolution of ICT, VC and policy in Israel

141

technological fields (usually based on external growth waves of various technological areas), only few regions succeeded in the full cluster emergence process. This chapter attempts to contribute to the analysis of the transition from pre-emergence to emergence. Following Avnimelech and Teubal’s analysis of the Israeli ICT cluster development, which covers the 1969–2008 period, the background phase begins with the appearance of R&D-performing firms and high-tech activities in the business sector. It is a period in which capabilities and other relevant resources are gradually being accumulated by various types of agents. During the pre-emergence phase start-up companies make their appearance. The experience accumulated enables selection of start-uprelated features and identification of specific systems failures related to start-ups’ foundation, operation and growth. This phase involves extensive qualitative changes including new start-up forms of organization, new technological competencies and cultural changes toward more acceptance of entrepreneurship. Also the domestic and global environments were becoming increasingly favorable to the development of ICT start-ups and related clusters. These ‘pre-emergence’ processes and conditions facilitated the cumulative process of emergence in the next phase. During emergence the venture capital (VC) industry emerged, the ICT cluster grew rapidly (overcoming a critical mass of activity) and supporting institutions were developed. The post-emergence phase is characterized by more moderated growth, increase of the cluster diversity and attempts to open new subclusters in related technological areas (such as cleantech and biotech or new sub-sectors of the ICT industry). The chapter adopts a broad co-evolutionary framework linking innovation and technology policy (ITP), regional ICT sectors, the start-up segment and innovation finance (VC). It combines two approaches which appear in the evolutionary economics literature: that which searches for mutual links between aggregates such as the co-evolution between technology and institutions pioneered by Nelson (1994, 2001, 2007) and links among agents in the complexity literature (Allen, 2004). The focus is on the role of policy and the VC industry in cluster emergence. While it strongly suggests VC emergence as a triggering factor during cluster emergence, the relevant co-evolutionary processes started at least two and a half decades earlier, with the creation of a specific government agency in charge of direct support of R&D in firms (The Office of the Chief Scientist, Ministry of Industry and Trade – OCS). It suggests that in addition to the direct contribution of VC to their portfolio companies the VC industry made also a significant indirect contribution to cluster development such as stimulating start-up creation, generating cluster reputation, accessing global networks, acting as a focusing device and enhancing coordination

142

Emerging clusters

activities (due to its position in a ‘structural hole’ of the innovation process – see Florida and Kenney, 1988). Several of these links are co-evolutionary, some new and others an acceleration of processes already initiated in the early 1970s. All of these were crucial in pushing the cluster above a critical mass level that triggers the self-reinforcing process of growth. The chapter is structured as follows: section 2 present a literature review, section 3 presents the general co-evolutionary patterns in the Israeli ICT cluster development, sections 4–6 present these co-evolutionary patterns in the various stages of development, and section 7 summarizes the chapter.

2 2.1

LITERATURE REVIEW Industrial Clusters

The phenomenon of regional clusters has been an integral part of economic development and economic geography theories since the works of Marshall (1890). Marshall (1890) defined an industrial district as ‘a geographic area containing a number of firms producing similar products, including firms operating at different stages of a production process that gain advantages through co-location’. Industrial clusters were established in many regions during the 1990s and provided, at least for a while, economic benefits for firms operating within them (Porter, 1998). The traditional explanation for the existence of regional clusters focuses on cost saving through agglomeration economics. Marshall (1890) proposed three main advantages of industrial clustering: (1) the development of a local pool of specialized labor and the availability of specialized input factors; (2) sharing the cost of investments in infrastructure and economy of scale in production, and (3) efficient information flows. A significant consequence of the process of globalization is a sharp reduction in shipping, transportation and communication costs, and elimination of most trade barriers (Maskell, 2001). Therefore, we might expect that clusters would become less relevant, mainly in knowledgeintensive industries where shipping costs are negligible. In practice, the role of clusters in knowledge-intensive industries seems to be even more significant than in traditional industrial clusters. Many explanations are given for this phenomenon. Ellison and Glaeser (1999) argue that the knowledge-intensive activities are more dependent on frequent face-to-face contacts due to the complexity of the knowledge and its tacit nature. Pinch and Henry (1999) show in their empirical study that clustering in knowledge-intensive industries is

The co-evolution of ICT, VC and policy in Israel

143

related to knowledge spillovers and learning rather than to operational cost saving. Roelandt et al. (2000) suggest three incentives for the clustering of knowledge-intensive industries: access to complementary technologies and operational synergies, joint R&D with suppliers and users, and lower transaction costs. A subgroup of the knowledge-intensive cluster is the start-up-intensive high-tech clusters, which is dominated by a large portion of innovative start-ups. These start-ups are the leading force in cluster dynamics and technological progress. Saxenian (1994) describes such clusters as agglomerations of start-up firms, pools of technical skills, specialized suppliers and supportive institutions. 2.2

Cluster Development and Entrepreneurship

Acs and Audretsch (1990) argue that entrepreneurs and start-ups play an important role in the economy, serving as agents of change, being a considerable source of innovation activity, stimulating industry evolution and becoming the main source of job creation. Entrepreneurship is also a fundamental characteristic of successful clusters (Saxenian, 1994; Feldman, 2004). Entrepreneurs play a special role in the cluster formation process (Brenner, 2004). They start new firms that exploit technological opportunities, start new economic activities and create new markets (Feldman et al., 2005). Klepper (2001, 2006) argues that cluster development is primarily the outcome of the nature of spin-off processes. His key assumptions are that key entrepreneurs may appear randomly in a certain region, leading to the growth of some leading firms within that region, and as a consequence of the cumulative nature of innovation, these firms will generate more and more efficient spin-offs. The eventual outcome of this process will be the development of a regional cluster. However, many scholars suggest that entrepreneurship often cannot be viewed as the only trigger of cluster formation, but as an essential element that co-evolves with the cluster after the same background condition has been generated in the latent cluster (Feldman, 2001; Avnimelech and Teubal, 2006; Menzel and Fornahl, 2009). Firm founding and cluster development have a strong interrelationship. On the one hand, firm founding is a significant element in cluster emergence, and on the other hand, the level of the cluster development has significant influence on the rate of new firm founding (Menzel and Fornahl, 2009). Firm founding has both quantitative and qualitative effects on cluster development. The quantitative effects are related to the number of firms and employees within the cluster. The qualitative ones are related to knowledge and competencies possessed in the cluster (Menzel and Fornahl, 2009).

144

2.3

Emerging clusters

Cluster Emergence

The functions of a new cluster are to promote division of labor/specialization and coordination, start-up formation and subsequent support, and economic growth. The relevant entity must be of a certain size (as Menzel and Fornahl, 2009 suggested ‘achieved a critical mass around some focal points’) that will enable it to exploit economies of scale and scope. In Avnimelech and Teubal (2006), we posit that a new cluster is an emergent structure, the outcome of ‘collective behavior’ of pre-existing agents. The cluster emergence process is characterized as a self-reinforcing cumulative process. This self-reinforcing, non-‘regular’ growth process is based on collective learning processes and other positive feedback cycles. Alternatively it can be stated that the process of emergence is characterized by dynamic economies of scale in which past growth accelerates future growth; and that it involves events that both generate and benefit from externalities. Not any trajectory with positive externalities will do: what is required is a trajectory with high positive externalities which induces significant resource inflows which exploit these externalities and which create new ones, and so on, thereby leading to a strong cumulative process. The cumulative momentum does not end with creation of the new cluster; rather it continues afterwards at least for a period of time in a selfreinforcing fashion. The new cluster, once emerged, will positively further stimulate start-ups and other components, at least for a while. Prior to cluster emergence the level of interconnections and collective learning between different agents in the cluster is quite low. To progress towards an emergence, some coordination between agents’ activities should be created, such as collective learning that will foster the accumulation of capabilities or selection of some technological focus that will enable the region to reach critical mass faster. In some cases, this (informal) coordination could be done by significant early market agents, that is through spin-off and imitation of the successful agent as Klepper (2006) suggests. But in most cases coordination issues should be addressed directly by public or semi-public authorities (Feldman et al., 2005; Menzel and Fornahl, 2009). Often the crucial push toward cluster emergence stems from radical technological change which creates new business opportunities, and is enhanced by individual entrepreneurs’ activities that spark collective action in the cluster and co-evolve with financial organizations and other institutions’ development (Feldman et al., 2005). The cluster, once established, acts as a selection device, attracting specific kinds of activities while repelling others and reducing the costs facing local entrepreneurs (Maskell, 2001). An initial step toward cluster emergence is the creation of a critical

The co-evolution of ICT, VC and policy in Israel

145

mass of start-ups, financial resources, skills and activities. The size of this critical mass depends on the thickness of the technological field, the size of the region, the strength of local networks, the business culture (mostly the tendency for communication and cooperation between different market agents) and the dynamism of the external environment. Once a critical mass is achieved, supporting organizations such as VC companies, specialized suppliers and business supporting services are attracted. Through this cumulative process, the cluster’s resources are built, supporting institutions emerge and networks are created. A significant characteristic of the emergence process is its co-evolutionary feature. Successful emergence requires simultaneous development of both demand-side (technological capabilities and entrepreneurs) and supplyside (risk capital and specialized intermediations) conditions (including resources, skills and agents). Often the government can accompany and stimulate this process and assist at critical points in the development of demand or supply.

3

CO-EVOLUTIONARY LINKS IN THE EVOLUTION OF ISRAEL’S ICT CLUSTER

Avnimelech and Teubal (2006) describe the development of the Israeli ICT cluster in four phases. We will analyse the first three by emphasizing the potential co-evolutionary links between three sets of variables: finance of Innovation; ICT start-ups; and Innovation and Technology Policy (ITP).1 The analysis will sometimes focus on ‘aggregates’ like the ‘VC industry’ and sometimes on individual agents. Following Nelson (2007) it could be argued that VC is an ‘institution’ which co-evolved both with the development, diffusion and application of ICT sectors in Israel and with the emergence of start-ups as specialized ‘inventor’ companies. Concerning individual agents’ links with other agents it is worth noting Allen’s view that ‘the creative interaction of multiple agents is naturally described by co-evolutionary, complex systems models in which both the agents, the structure of their interactions and the products and services that they exchange evolve qualitatively’ (Allen et al., 2007, p. 402). We will see that the qualitative dimension pervades all three phases: in phases 1 and 2 (1969–84, 1985–92) it refers, among other things, to the types of financial institutions supporting start-ups and R&D in firms; and in phase 3 (1993– 2000) it refers to the emergence of the multi-agent structure itself; that is Israel’s VC industry (and market) and the associated entrepreneurial ICT cluster.

146

Emerging clusters

The types of co-evolutionary links we focus on are: (a) ITP–Innovation co-evolution; (b) ITP–Finance co-evolution; and (c) Innovation–Finance co-evolution. These co-evolution processes result when ITP succeeds in expanding innovation in firms, in developing innovation capabilities, and in creating new R&D-performing firms and new sectors. This in response created new demands and challenges to the ITP in two directions. First was the increase of the scope of the ITP budget due to growing demand and adaptation of the ITP to new system failures identified in later stages of the cluster development – type (a) co-evolution. Second, some of the new system failures identified require the emergence of new supporting institutions in the cluster. ITP may be involved in enhancing the creation of such supporting institutions including innovation financing institutions – type (b) co-evolutions. Type (c) co-evolution results from search and adoption of non-government mechanisms of financing R&D in firms and the response by the start-ups to these new finance mechanisms. 3.1

ITP: The ‘Industrial R&D Fund’ and Other OCS Programs

The Israeli Government’s ITP towards the business sector, which played an important role in the abovementioned co-evolutionary processes, began in 1969 with the creation of the OCS (The Office of the Chief Scientist, Ministry of Industry and Trade) and of the ‘regular’ R&D support program. It supports the R&D of individual companies whose objective was the creation of new or improved products (or processes) directed to the export market. This program was horizontal, that is directed to the business sector as a whole and open in principle to all firms (Teubal, 1997). Moreover, the 50 percent subsidy to every R&D project submitted to the OCS was common to projects in all sectors and technologies (‘neutrality’ of incentives). The ‘R&D legislation’ of 1984 further consolidated Israel’s support of business sector R&D. The objective was to support knowledge intensive industries, through expansion of the STE infrastructure, exploitation of existing human resources and focusing on export markets. The outcome was significant increases in R&D grants from US$2.5m in 1969, to US$39m in 1980, to US$133m in 1990 and to US$337m in 2000. This program was the first of the set of programs comprising Israel’s program portfolio (Box 6.1) and it was, and to some extent continues to be, the backbone of Israel’s innovation and technology strategy as far as the business sector is concerned.2 A framework for understanding the impact of this R&D grant program is to adopt an evolutionary perspective (Teubal, 1997). Under such a perspective, the major obstacle in early

The co-evolution of ICT, VC and policy in Israel

147

BOX 6.1 NEW ITP PROGRAMS IN ISRAEL DURING THE 1990s (1) Inbal (1992–98) – a government-owned insurance company, which gave partial (70 percent) guarantees to traded VC funds. Four VC companies were established under Inbal regulations. This was a ‘failed’ government program, which paved the way for the successful implementation of Yozma. (2) Yozma (1993–98) – a $100m government-owned VC company, which invested in ten private early-phase oriented VC funds which operated in Israel ($8m per fund). (3) Magnet Program (1992–) – a horizontal program supporting cooperative, generic R&D involving a group of firms and at least one University (annual budget $40–60m). (4) Technological Incubators’ Program (1992–) – a program supporting start-ups during the first three years of their operation. The incubators are privately managed and get financial support from the government (annual total program budget $25–30m).

development of R&D-intensive sectors is the absence of R&D capabilities (and correspondingly, absence of ‘demand’ for the R&D functions within firms). Therefore, the central objective is penetration and diffusion of R&D throughout the business sector. This involves much more than ‘R&D’ additionality: it also involves aiming at generating capabilities (largely through the promotion of collective learning about R&D/innovation) and promoting high-tech entrepreneurship. After the so-called ‘infant’ period of implementation the R&D-performing segment of the business sector may be characterized by: exhaustion of learning opportunities and associated externalities (in certain areas); opportunities for complex types of R&D projects (based on capabilities generated); and identification of possible areas of competitive advantage (Teubal, 1997). The policy response to this would ideally involve: (a) reductions in average R&D subsidies together with greater selectivity; and (b) the implementation of a number of targeted programs for specific areas/technologies or specific system failures. Israel’s successful implementation of the regular ‘Grants to Business Sector R&D’ program created conditions for the subsequent implementation of a set of new programs in the early 1990s including the ‘targeting’ of VC through the Yozma program.

148

3.2

Emerging clusters

Financing Innovation and Venture Capital

The identification of new forms of intermediation lies at the heart of the creation of new industries and clusters. Gompers and Lerner (1999) and others have been very clear that VC (a new ‘supply agent’) mediates between capital and start-ups in ways that the traditional banking system (old ‘supply agent’) did not, thereby overcoming ‘market failures’ with start-up and innovation financing. Thus, the VC industry has an important role in creation of new industries and clusters. According to Florida and Kenney (1988), the VC industry played a central role in coordinating various high-tech agents: entrepreneurs, managers, professional employees, specialized suppliers, investors and capital markets, and customers and product markets. In our research we consider VC as an industry and market, which coevolves with the start-up segment of the high-tech cluster. We use the notion VC industry in its broad sense to describe not only the formal, independent VC sector but also other types of specialized providers of finance to technological start-ups such as business angels, technological incubators, and specialized investment companies. We consider these agents as a new industry due to the fact that they created a new eco-system that generated the financial and other support required for start-ups’ creation and development. This new eco-system includes new supply agents and corresponding adjusted demand agents (start-ups), comprehensive channels of communication and interaction, common contracts, specialized labor and other features common to industries.

4

BACKGROUND CONDITIONS PHASE (1969–84)

In this phase, the start-up-intensive ICT cluster did not yet exist. The science, technology and higher education infrastructure (STE) was strong. Stimulated by the OCS’s grants to business sector R&D scheme which started in 1969, a segment of R&D-performing firms and a specialized labor force were created. Throughout, OCS’s proactive policy promoted ‘demand’ for R&D by firms both through a ‘collective learning’ process and through enhanced awareness of the commercial potential of R&D (Teubal 1997). In the early 1980s, in response to experience and changed circumstances, the OCS adjusted the R&D grants program and was instrumental in creating new ‘incentives’ programs (for example BIRD, see p. 150). In this phase, a number of critical events took place. These include the development and diffusion of R&D/innovation capabilities throughout the business sector, the beginning of global product and

The co-evolution of ICT, VC and policy in Israel

149

capital market links, creation of a favorable environment for foreign investment, the beginning of involvement of business sector agents such as banks in high-tech activity financing, and the very first steps of technological entrepreneurship activity. Phase I Co-evolutionary Processes Type (a) co-evolution links started with the implementation of the regular R&D fund. The OCS support for firms’ R&D was in the form of a 50 percent subsidy of R&D expenses. This required from the beginning the involvement of the private sector in financing innovation. In the early years this support was used mostly by established firms which conducted small R&D projects and complemented the OCS grants with internal funds. The initial successes of OCS-supported R&D projects and collective learning of R&D performing companies led to a strong rise in ‘demand’ for government subsidies to R&D. There were a couple of firms in the early 1970s that presented significant increasing demand for R&D budget. Due to this significant increase in the R&D budget these firms start looking for private external sources of innovation finance. Later on (late 1970s) this became an increasing problem for the whole R&D-performing business sector of Israel, namely that the limited funds available from the OCS to promote innovation were constraining the rapid increase in R&D projects desired and proposed by firms. This led to a continued search for an alternative mechanism to finance R&D and innovation during the subsequent decade of the 1980s. The immediate government response was the implementation of various limited solutions such as the Elscint Law, that is tax benefits based on R&D expenditures. However, the search for a more comprehensive solution continued in the next phases of the ICT cluster development. As mentioned before, the regular R&D grant program led to an expansion of capabilities and the enhanced willingness of entrepreneurs to harness R&D for company growth. This in response led both to a continued increase in the budgets of the Israeli agency in charge of ITP (OCS) and to the addition of two new programs to deal with the new market and systems failure identified. These programs were the ‘Projects of National Importance’ program and the BIRD program. The former program was financed by the World Bank and implemented through the OCS during the second half of the 1970s. It supported the R&D of ‘large projects’ involving not only a firm but also a university at an 80 percent R&D subsidy rate rather than the normal 50 percent. By promoting a higher degree of sophistication and complexity of R&D/innovation projects and therefore facilitating future company growth and diversification, this program

150

Emerging clusters

benefited those companies who had accumulated substantial innovation capabilities during the first years of OCS support. Through its success and by considerably enhancing ‘demand’ for R&D by firms it also indirectly spurred not only the growth of the OCS budget but also the search for alternative sources of support for R&D as mentioned above. Another new program was the BIRD (Bi-national Industrial R&D) program which supported joint innovation programs implemented by a US and an Israeli company. It supported innovation more broadly speaking (rather than the R&D function exclusively as was the case with regular OCS grants), with the Israeli company focusing on R&D and the US company on accessing the complementary assets required for effective commercialization of the invention. As with the first program, BIRD (whose implementation started in the early 1980s) depended on the innovation capabilities developed during the 1970s as a result of the OCS’s grants to the R&D program. It had an enormous influence in developing personal and business links between Israeli and US companies. Both programs directly and indirectly influenced innovation capabilities in firms and contributed to the emergence of Israel’s ICT cluster of the 1990s. Another important co-evolutionary effect of the early 1980s coincided with the beginning (worldwide and in Israel) of a separately identifiable software industry during the early 1980s. The ample opportunities for new high-tech start-ups that this entailed (partly also due to the significant R&D and innovation capabilities accumulated until then) spurred the appearance of a new set of entrepreneurs and a new set of start-up companies. These, which included Rad Group, Comverse (or its precursor, Efrat), Amdox and others, were not linked to existing leading economic groups and did not have internal financial resources. The OCS was active, no less than directly financing these start-ups, in finding alternative mechanisms of support as well. This led to the first and highly idiosyncratic form of VC financing in Israel, an important precursor to the high impact VC industry 15 years later. VCs then were pools of money fed from investments by private individuals in the US that were oriented to specific projects (rather than to purchasing an equity stake in a company). Israel’s government policy was instrumental in implementing the new financial opportunities by helping investors, through loans, to take advantage of potential income tax breaks. The outcome was the beginning of a pool of domestic start-up companies during the first half of the 1980s. Phase 1 co-evolution summary Our analysis up to now suggests that, through its impact on innovation, ITP implemented during the 1970s and 1980s co-evolved with innovation capabilities, R&D-performing firms and, very importantly, with the

The co-evolution of ICT, VC and policy in Israel

151

emerging set of ICT-oriented start-ups. Moreover, the co-evolution processes also led to the appearance of alternative methods of finance of R&D/ innovation in the business sector (over and beyond government subsidies), all of them precursors to the major role played by VC during the 1990s.

5

PRE-EMERGENCE PHASE (1969–92)

A central feature of this phase is the appearance of significant start-up activity and the gradual acceptance of technological entrepreneurship. A critical mass of ICT start-ups was accumulated towards the end of this phase and, correspondingly, a measure of ‘demand’ for the services of VC. During this phase a process of experimentation within the high-tech population of firms took place. This was followed by a measure of development of software and other ICT areas and of critical characteristics of the start-ups. Also, a process of experimentation took place in connection with types of VC-related activities and with related ITPs. Phase II Co-evolutionary Processes During this phase disbursements for ITP continued to grow. This factor together with other policies, for example the restructuring of the defense industries starting in the mid 1980s (a fact leading both to a significant transfer of scientists and engineers and to an increase in the flows of new S&T graduates to civilian industry) also led, directly and indirectly, to the creation of new ICT start-ups. In addition, since the late 1980s, new national priorities emerged with the arrival of the massive immigration from the former Soviet Union. The government of Israel began searching for means to employ the thousands of engineers that came to this country. One of the new objectives of the OCS was enhancement of start-up formation, survival and growth. The growth in the number of start-up companies urged the OCS to react to other systems failures identified in the innovation process. Until the 1990s, the percentage of successful young companies was extremely low and the accepted view was that this resulted from weak management and marketing skills and lack of complementary assets and institutions in the Israeli ICT sectors. Experts in the field realized that despite massive government support for R&D there still was a clear ‘market failure’, which blocked the successful creation and development of start-ups. The basic problem was lack of capability to grow after the product development phase. The head of OCS at the time, Yigal Erlich, thought that the way to get it was to foster a VC industry. The response was a search for policy

152

Emerging clusters

processes that could enhance VC activity in Israel. These searches eventually included Targeted programs supporting VC (Inbal and Yozma), and complementary programs raising the demand for VC services (for example the Technological Incubators’ Program, and the increasing regular R&D grant program budget). The growth of R&D-performing companies which resulted from successful OCS programs, and the appearance of what could be termed a ‘civilian’ high-tech sector comprising companies mostly oriented to the US and European markets, was also accompanied by the private sector search for new mechanisms for financing innovation. In parallel, a new phase in the globalization process began during the late 1980s: foreign start-ups could more easily be floated on NASDAQ, provided the economy had adapted to the new opportunities of the ICT revolution. Thus, an important ‘new’ source for the ‘mature’, non-start-up companies that operated in the mid 1980s (Scitex, Tadiran, Optrotec, Orbot, and others) was undertaking IPOs (Initial Public Offerings) in NASDAQ. New links with US underwriters and investment banks were formed, with the latter beginning to appreciate the innovative and associated capital market potential of Israeli companies. These early links were two-way links, in the sense that the new financial possibilities further re-enforced innovation capabilities and the growth of companies, both ‘mature’ and start-up. Lehman Brothers, which was active as an underwriter of ‘mature’ companies in the mid 1980s, also became an underwriter for some of the IPOs of start-ups during the early 1990s. This provided new opportunities for start-up companies and must have been a factor explaining the acceleration of start-up foundations in the early 1990s. The above effects further contributed to create favorable conditions for the strong VC start-up co-evolutionary process of phase 3. The growing importance of start-ups towards the late 1980s drew the attention of foreign investors (for example Jay Morrison of California, who was co-founder of a VC company in Israel which later metamorphosed into Jerusalem Venture Partners) and foreign financial institutions that came to Israel (such as Advent Venture Partners and Walden International) to search for start-up investment opportunities. To this we must add the creation of the first two independent Israeli VC funds, Athena in 1985 and Star Ventures in 1989. Phase II co-evolution summary A major underpinning of the strong VC start-up co-evolution of Phase 3 was the mutual adaptation of the finance demand agents and the finance supply agents during Phase 2. Qualitative changes in the start-up

The co-evolution of ICT, VC and policy in Israel

153

organization and strategy which developed in parallel to the quantitative changes were strongly influenced by the eventual identification of and awareness about technological areas having a potential competitive advantage in the Israeli context (‘software and communications-related innovations’). Continued experimentation and learning about R&D and innovation (enhanced by monotonic expansion of government R&D grants) generated variety and eventual selection of adequate business models and organizational forms that fitted the local context. Enhanced experimentation also resulted from the restructuring of large civilian and defense-oriented companies. The combined effect of both led to the appearance and ‘selection’ of the new start-up business model, a ‘born global’ outlook, and a strategy and organization which increasingly became oriented not only to product markets but also to knowledge and capital markets (both private and public). A parallel process of experimentation of various forms of ‘finance supply’ took place. The process could be described as leading – for the selected ICT areas – to a clear preference for a VC-based financial intermediation ‘model’ over other forms of innovation finance. It also led to selection of one particular organizational form – limited partnerships (the dominant VC form in the US and in Israel) – with a very clear technological and strategic investment focus.

6

EMERGENCE OF THE HIGH TECH CLUSTER (1993–2000)

The rapid quantitative growth of start-up and VC activity that took place during the emergence process led to a well defined VC industry (and market) and start-up-intensive ICT cluster. Emergence, in our context, is the outcome of an evolutionary process leading to structural change (Foster and Metcalfe 2001). Emergence began with a fluid sub-phase (1993–95) followed by rapid growth (1996–98) and ended with the bubble (1999–2000). During the fluid sub-phase a continuation of the experimentation and collective learning of pre-emergence took place leading to further post-selection specification of start-up and VC strategies and organization. It ended with a specific technological focus and with a specific set of business models and exit mechanisms of the companies in the cluster. During the rapid growth sub-phase accelerated entry of start-ups and VC companies occurred and a domestic VC industry was created. It is then that the cluster attained a size which enabled it to sustain a large number of supporting services such as local investment banks and underwriters, specialized attorneys and

154

Emerging clusters

lawyers, and other start-up-oriented services. This ‘cluster effect’ induced entry of additional leading multinational companies in the ICT areas and of additional domestic and foreign VCs, followed by rapid creation of new start-ups and rapid growth in the acquisition activities of multinationals. As long as external and internal conditions remained unchanged, the dynamic increasing returns to scale (or positive feedback cumulative) process of creation of large numbers of start-ups continued and the startup-intensive ICT cluster continued to grow. As the global ICT market entered the bubble and bust period the local Israeli ICT cluster followed it. At this point the accumulated strength and stability of the cluster was crucial for future survival and development. Phase III Co-evolutionary Processes The expansion and diversification of the ‘direct’ support of R&D/ innovation in firms as represented by the increased budgets to the R&D grants program and the new Technological Incubators and Magnet Programs of the early 1990s (see Box 6.1), further reinforced innovation and technological entrepreneurship, start-up foundations, and growth of established R&D-performing companies. Eventually, the number of start-ups in operation by 1992 was approximatly 300, some of them of high quality including some (Lannet, Lanoptics, Magic and The Third Dimension) having successfully undertaken IPOs in NASDAQ. These positive impacts and other factors, like the immigration of large numbers of engineers from the former Soviet Union, the rapid growth of global ICT markets and the taking off of the Internet which led to the foundation of new firms like CheckPoint, led to further expansion of these programs. The significant increase in start-up activity and the prospects of further increases in response to the new opportunities in the early 1990s created an ‘excess demand’ for the mechanisms in support of start-ups. This led also to the two VC-directed programs mentioned above: Inbal in 1992, a failure; and Yozma in 1993, a big success. Both targeted private, independent VC funds. Since Yozma was strongly related to future VC and to innovation and start-ups, it linked to all types of co-evolutionary process. The outcome was a new mechanism for financing and supporting innovation which co-evolved with the growth of start-up and, at another level, the emergence of the entrepreneurial ICT cluster as a whole. As mentioned, the trigger of VC emergence was implementation of Yozma in 1993 which led to $250m capital raised (see Table 6.1) and significant amounts invested up to 1996 (Avnimelech and Teubal 2006). This and a few successful exits during the mid 1990s induced imitation by other private VCs which in turn stimulated further start-up entry as well

The co-evolution of ICT, VC and policy in Israel

155

as a strong collective learning processes. This further stimulated entry both of new start-ups and of new VC companies, and accelerated creation of second and third funds both by existing Yozma VC companies and by other VC companies Beyond ‘entry’ of new VC organizations and funds, cluster emergence in Israel involved three other important component sub-processes: collective learning,3 networking effects and exploitation of economies of scale. Also emergence was associated with strong VC-start-up co-evolution which became a central part of the overall process. Yozma’s implementation signals the onset of a more intense, largely ‘quantitative’ process of VC–start-up co-evolution (see Table 6.2). The direct and indirect links among variables become increasingly regular and frequent.4 First we observe a quantum jump in VC capital and a temporary excess supply of VC services, followed by a relatively rapid entrepreneurial response. As a result we observe not only an increased share of start-ups that were VC-backed but also significant increases in gross additions to start-ups during 1995–96. These were either a reaction to the ‘excess supply of VC services’ or the expectation that new start-up foundations would easily find new VC sources of finance if required. Then we observe increasingly synchronous growth: starting in 1996 start-up demand for VC services and VC ‘supply’ become increasingly synchronous, that is there was rapid mutual adjustment. During the rest of the decade, the share of start-ups that are VC-backed increases (up to about 50 percent). Enhanced high-tech activity enabled a better exploitation of economies of scale in the domestic generation of non-traded (or partially non-traded) inputs to high technology, for example accountants, lawyers, investment bankers, consultants, providers of knowledge inputs, and suppliers of production and marketing inputs. This factor, together with the existence and expansion of networks of personal and professional links, contributed to reductions in transaction costs and facilitated international cooperation. Through time Israel’s ICT cluster became increasingly capable of providing effective services to new start-ups, both through VC and through other input and service suppliers. Emergence Outcome The VC industry After 1993, there was a rapid growth of the Israeli VC industry both in terms of capital raised and in terms of number of funds (see tables 6.1 and 6.2). We mentioned that prior to Yozma there was some VC activity in Israel, but it did not reach a critical mass. The big jump occurred in 1993 when Yozma and private VCs raised

156

Emerging clusters

Table 6.1

Capital raised by PE (Private Equity)/VC organization in Israel* 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Private VCs Yozma VC Inbal VC Other PE Total

40

49

27

33

82

93

287

0

0

0

149

40

15

30

0 5 45

0 9 58

54 79 160

22 168 372

0 262 384

0 25 133

0 620 937

595 19* 0 134 777

653 0 0 33 686

1160 2712 0

0

0 0 258 66 1418 2778

Note: * Table 6.1 presents the growth of different segments of the Israeli VC industry during 1993–2000, in which period the LP VC fundraising average annual growth rate was 85 percent. At the same time, the other PE agents experienced moderate growth rates. The time trends of VC/PE fundraising growth rates during 1991–2000 suggest that while underlying capital market trends influences both VC and PE, the Yozma program, which triggered VC emergence, crowded out PE activity for a while.

Table 6.2

VC invested and high-tech start-ups foundation, IPOs and M&A*

Year

VC raised (Total PE)

VC invested (% Foreign)

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

40 (49) 49 (58) 81 (160) 204 (372) 112 (374) 135 (166) 317 (937) 643 (777) 653 (686) 1160 (1418) 2521 (2778)

NA NA NA NA NA NA 293 (32%) 436 (40%) 589 (43%) 1011 (57%) 3233 (59%)

IPOs in Significant Start-ups IPOs in M&As foundation NASDAQ Europe (VC(VC(VC-backed) (VCbacked) backed) backed) 53 (4) 51 (5) 94 (12) 124 (73) 140 (85) 175 (87) 231 (117) 260 (119) 312 (152) 573 (208) 642 (372)

1 (1) 4 (1) 9 (1) 11 (4) 8 (4) 9 (4) 16 (10) 12 (4) 7 (4) 12 (9) 19 (12)

0 (0) 0 (0) 0 (0) 0 (0) 1 (0) 2 (0) 3 (1) 0 (0) 6 (1) 6 (1) 13 (3)

1 (0) 0 (0) 1 (0) 1 (0) 2 (2) 7 (3) 11 (3) 7 (3) 16 (6) 15 (9) 32 (11)

Note: * Only of high-tech start-ups; ** Not including fire sales (at least $25m or ROI (Return on Investment) of 25 percent). Source:

IVC (2008) and author’s calculations.

The co-evolution of ICT, VC and policy in Israel

157

$162m compared to $27m in 1992. After 1993 the VC industry presented a rapid growth in capital and activity. During the second half of the 1990s, the Israeli VC industry became one of the most significant VC industries in the world with significant influence on Israel’s ICT industry. It was then that the first foreign VC companies began to invest directly in Israeli start-ups. Later on, a few of them (for example, Benchmark, Sequoia, Intel Capital and others) established Israeli offices. The cumulative process was also fueled by favorable changes in the external environment in particular the rise of the NASDAQ Index and deregulation of global communication markets. Additional internal factors were the Oslo peace agreements, the immigration from the former Soviet Union countries and domestic regulatory changes.5 IPOs and M&As Only small numbers of Israeli companies undertook an IPO in NASDAQ (or in other global markets) until the late 1980s. In the early 1990s, there was a wave of Israeli IPOs in NASDAQ (13 during 1991–92). This in part reflects the fact that high quality start-ups operated in the early 1990s. This number jumped in 1996 with the first exits from Yozma funds (16 IPOs, 8 backed by Yozma funds). IPOs in NASDAQ also increased considerably during 1996-2000 (66 IPOs, 39 VC-backed) compared to 1991–95 (41 IPOs, 14 VC-backed). A related point is the increase in the share of VC-backed issues from roughly 35 percent before 1996 to over 60 percent in 1999–2000.6 The picture that emerges is one of increasing maturity of Israel’s ICT cluster, a process that accompanied the increase in the NASDAQ index (which by itself would also induce an increase in IPOs).7 Today Israeli (or Israeli-related) companies traded in the US (over 150 companies) are the third largest group after the U.S. and Canada. Moreover, many Israeli (or Israeli-related) high-tech companies are also traded in European stock markets such as the London-based AIM stock market. Mergers and acquisitions (M&A) is one of the main mechanisms of exit for VC investors and for start-up entrepreneurs. Since there is no clear ‘market place’ where M&A transactions are negotiated and implemented, the conditions for an emerging cluster to facilitate M&A activity on a continuous basis differ from those required to provide access to public capital markets. Clearly reputation effects are required in order to trigger MNEs to undertake a costly search for technological opportunities in a specific cluster. The Israeli case suggests that a critical mass of IPOs might play a crucial role in creating the conditions for cluster emergence and that M&A only came in increasingly large numbers later on (becoming significant in Israel only since 1996).

158

Table 6.3

Emerging clusters

Israel’s high-tech cluster – selected structural elements (1969–2000)

A: Accumulated figures for a period Number of high-tech start-up creations (VCbacked*) Israeli VC fundraised/VC invested in Israeli startups ($m) Number of IPOs at US (at EU & TASE**) Number of significant trade sales (M&As) Amount raised – public markets and M&As ($b)

1969–1984

1985–1992

1993–2000

136 (0)

349 (23)

2 436 (855)

0/0

~85 / ~50

7 480 / ~5 600

14 (7)

19 (15)

101 (75)

0

2

91

0.3

0.8

36.7

B: Figures for an actual year Share of ICT in manufacturing exports ICT exports $m (as % of ICT sales) Software development exports $m (as % of software sales) ICT professional employees (000s) Patents issued in the US (ICT patents issued) R&D as % of GDP (OCS R&D grants $m)

1984 14%

1992 28%

2000 53%

~900 (50%)

2 711 (50%)

12 893 (59%)

5 (4%)

135 (23%)

2 600 (70%)

~42.9

61.7

152.4

193 (44)

355 (89)

969 (417)

2.4% (97)

2.6% (199)

4.5% (440)

Notes: Notice that in Part A, the second column represents 16 years while columns three and four represent eight years. * by investment year. ** TASE = Tel Aviv Stock Exchange. Sources: Avnimelech and Teubal (2008a), IVC (2008) – row 2–6, CBS (2008) – row 8–11 (estimates from IAEI), USPTO (2009) – row 12, and OCS (2008) row 13.

Start-up intensive ICT cluster The ICT cluster emergence period runs from 1993 until 2000 (see Table 6.3). The ICT cluster that emerged during the 1990s was very different from the military industries-oriented high-tech sector of the 1980s. Its basic feature is that of being a start-up-intensive and export-oriented

The co-evolution of ICT, VC and policy in Israel

159

cluster closely following the Silicon Valley model, where both start-ups and VCs play important roles. It is also much more integrated and linked with the US and with ICT clusters in Silicon Valley, the Boston area and elsewhere. The number of active start-ups increased from approximately 300 in 1991 to approximately 3000 towards the end of the decade, and the number of VC companies increased from two in 1990 (Star and AthenaVeritas) to over 60 towards the end of the decade (IVC 2001). In parallel to this the economy continued the high-tech biased structural change initiated in the 1970s but in an accelerated mode (Justman and Zuscovitch 2002). ICT industries (manufacturing and services) revenues increased from $4.6b in 1990 to $23.9b in 2000 (CBS 2001, in 2005 prices). ICT exports increased from $2.1b in 1990 to $12.9b in 2000 (ibid.). The share of these industries in the total employment increased from 3.9 percent in 1990 to 6.9 percent in 2000 while the share of ICT in total manufacturing exports has increased from only 27 percent in 1990 to 53 percent in 2000 (CBS, 2001).

7

CONCLUSIONS

Israel’s success in creating a VC industry (and market) and entrepreneurial ICT cluster during the 1990s depended on conditions (1)–(5) below, some of which also reflect previous work by the authors. Following their presentation we end by summarizing their partial ‘integration’ into a broad co-evolutionary framework which is the main contribution of this chapter. The main conditions are: 1.

2.

3.

existence of favorable background conditions (1969–85), particularly the strong Science, Technology and Higher Education infrastructure (STE) including in the military, the significant diffusion of R&D capabilities throughout the business sector, and the creation of specific high-tech segments; A pre-emergence period (1986–92) during which wide experimentation and qualitative selection took place concerning both desirable areas of R&D/innovation and start-up and VC organizational forms and strategy; where policy capabilities were developed and new strategic priorities identified; and where a critical mass of start-ups was generated; A cumulative and autocatalytic emergence process that led to a reconfiguration of Israel’s prior, military-dominated, high-tech sector towards a start-up-intensive ICT cluster with unprecedented dynamics and significant quantitative growth;8

160

4.

5.

Emerging clusters

A targeted VC policy (Yozma, 1993–98), which triggered the abovementioned rapid process of cumulative growth during 1993–2000 including, among other sub-processes, a strong process of VC start-up co-evolution; Creation of strong global capital market links (through IPOs and M&A) particularly with NASDAQ and the US; and a growing presence of R&D labs in leading ICT multinational companies in Israel.

The main contribution of this chapter lies in the analysis of the coevolution of three groups of variables which contributed to the emergence process mentioned above: ITP; ICT innovation and the organizations involved; and mechanisms of finance of innovation. The co-evolutionary analysis, which is largely ‘qualitative’ in this chapter, is based on considering three types of co-evolutionary processes along each one of the three phases of evolution between 1969 and 2000. Co-evolution of type (a) links the two first variables (ITP–innovation co-evolution); co-evolution of type (b) links the first and the third variables (ITP–finance co-evolution); and co-evolution of type (c) links the second and third variable (innovation– finance co-evolution). Introducing co-evolutionary processes in a much more extensive way than in previous papers by the authors considerably adds to the analysis of emergence of Israel’s VC industry (and market) and entrepreneurial ICT cluster. Despite numerous pieces of research on the topic of co-evolution and considerable advances made (see Nelson 1994, 2007; Murmann 2002, 2003; Breznitz 2007, 2008; Avnimelech and Teubal 2008a, 2008b; among others) the ‘Schumpeterian’ literature on co-evolution including this chapter lacks a basic analytical framework for incorporating such processes into a broader evolutionary process. This chapter suggests that an early, rather random and haphazard, largely qualitative co-evolutionary process among broad sets of variables (‘R&D performing companies’ and ‘mechanisms of innovation finance’) should precede a subsequent, more focused and synchronized process among a sharper and more reduced set of variables (start-ups and VC). The former qualitative co-evolution corresponds to our pre-emergence phase while the latter rather quantitative systemic co-evolution corresponds to the emergence phase. Systemic VC start-up co-evolution would be the backbone of the rather complex process of emergence of the new, higher order organizations namely the VC market/industry, and start-up-intensive ICT cluster. These two phases corresponded to Utterback and Abernathy’s (1975) Fluid and Growth Phase respectively. In our analysis Qualitative coevolution is a process of mutual adaptation of the organization/strategy

The co-evolution of ICT, VC and policy in Israel

161

of two classes of agents, in our case innovation finance supply agents (mechanisms of finance first followed by VC organizations and funds), and innovation finance demand agents (R&D-performing companies first followed by start-up companies), when the quantities of at least one of the more specific agents (VC or start-up) is not large. In our analysis this led to the limited partnership form of VC organization; and to a start-up business model involving ‘born global’ companies and an orientation not only to the product but also to the global capital market. Finding a good organizational/strategic fit which reflects the underlying environment in which these agents operate (a fact that may require also adaptations of the institutional context) is the main outcome of a virtuous process of qualitative co-evolution. Provided that like in Israel during the 1990s, other exogenous and endogenous variables do not deteriorate dramatically, and provided that a ‘dominant (product) design’ of the corresponding industry/market (VC in Israel) has also been arrived at, the abovementioned process could represent the qualitative underpinning for a subsequent, quantitative and more synchronized and systemic coevolutionary process. In the case analysed in the chapter, the trigger for systemic co-evolution was achieved through a critical mass of each one of the ‘narrowly defined’ variables: start-up and of VC (both with ‘appropriate’ organization/strategy’ configurations). The critical mass of start-ups was achieved in 1992 by an innovation policy involving direct grants for R&D in firms starting 24 years before (which together with other public policy measures generated a number of related co-evolutionary processes); while the critical mass of VC was achieved by the Yozma program in 1993 which triggered ‘emergence’.

NOTES 1. For our purpose, ITP will focus (although not exclusively) on direct support of innovation in firms through incentive programs (for example OCS subsidies to R&D projects of firms) and on support of VC. Other indirect support of innovation through support of the S&T infrastructure will only briefly be considered since the bulk of their funding was exogenous to the processes being discussed. 2. Until the early 1990s more than 90 percent of government disbursements to civilian R&D came out of this program. 3. VC learning was related to screening deal flow, due diligence, selection of investments, investing in start-ups, monitoring start-ups, providing value added, and in the exiting process. In the start-up area, learning was related to structuring, raising capital, conducting effective R&D, marketing and alliance skills, penetrating global markets and other relevant skills. Collective learning stimulated entry of new market agents, and growth of existing market agents. Moreover the learning process involved a significant component of learning from foreign agents: in the VC area particularly from the foreign LPs (Limited Partnerships) of Yozma funds, and in the start-up area particularly from foreign multinational enterprises.

162

Emerging clusters

4. Frequently entrepreneurs from VC portfolio companies subsequently become consultants to the VC or even partners (VC interviews during the years 1999–2001). 5. The distinctive characteristics of Israel’s VC industry during the 1990s include (see OECD 2004 and Avnimelech and Teubal, 2004, 2006): (i) highest VC investments as a share of GNP; (ii) high share of VC investments in Early Stages; (iii) high share of VC investments in ICT; (iv) more than 90 percent of funds coming from foreign sources; and (v) the highest number of IPOs in NASDAQ after the US and Canada. 6. It is expected to see exits of VCs about 5–7 years after the fund establishment. 7. Both factors were at work here. However, the increases in the NASDAQ Index did not induce other clusters to float more companies in that ‘global’ capital market for technology companies. 8. It is also noteworthy to mention that, during the 1990s when the above emergence processes took place, the stage of globalization was such that it enabled the strong inflows of foreign resources that underpinned that process (over 95 percent of funds of domestic VCs were foreign-owned).

REFERENCES Acs, Z.J. and D.B. Audretsch (1990), Innovation and Small Firms, Cambridge, MA: MIT Press. Allen, P. (2004), ‘The Complexity of Structure, Strategy and Decision Making’, Chapter 5 in J.S. Metcalfe and J. Foster (eds), Evolution and Economic Complexity, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Allen, P.M., M. Strathern and J.S. Baldwin (2007), ‘Complexity and the Limits to Learning’, Journal of Evolutionary Economics, 17, 401–31. Avnimelech, G. and M. Teubal (2004), ‘Venture Capital–Start-up Co-Evolution and the Emergence and Development of Israel’s New High Tech Cluster’, Economics of Innovation and New Technology, 13(1), 33–60. Avnimelech, G. and M. Teubal (2006), ‘Creating VC Industries which Coevolve with High Tech: Insights from an Extended Industry Life Cycle (ILC) Perspective to the Israeli Experience’, Research Policy, 35(10), 1477–98. Avnimelech, G. and M. Teubal (2008a), ‘From Direct Support of Business Sector R&D/Innovation to Targeting Venture Capital/Private Equity: A CatchingUp Innovation And Technology Policy Life Cycle Perspective’, Economics of Innovation and New Technology, 17(1), 153–72. Avnimelech, G. and M. Teubal (2008b), ‘Evolutionary Targeting’ Journal of Evolutionary Economics, 18(2), 151–66. Brenner, T. (2004), Local Industrial Clusters: Existence, Emergence, and Evolution, London: Routledge. Breschi, S. and F. Malerba (2001), ‘The Geography of Innovation and Economic Clustering: Some Introductory Notes’, Industrial and Corporate Change, 11(4), 817–33. Bresnahan, T. and A. Gambardella (eds) (2004), Building High Tech Clusters: Silicon Valley and Beyond, Cambridge: Cambridge University Press. Bresnahan, T., A. Gambardella and A. Saxenian (2001), ‘Old Economy Inputs for New Economy Outputs: Cluster Formation in the New Silicon Valleys’, Industrial Corporate Change, 10(4), 835–60. Breznitz, D. (2007), ‘Industrial R&D as a National Policy: Horizontal Technology

The co-evolution of ICT, VC and policy in Israel

163

Policies and Industry–State Co-evolution in the Growth of the Israeli Software Industry’, Research Policy, 30, 1465–82. Breznitz, D. (2008), Innovation and the State, Political Choice and Strategies for Growth in Israel, Taiwan and Ireland, New Haven and London: Yale University Press. CBS (2001), Development of Information and Communications Technologies in the Last Decade, Central Bureau of Statistics, Israel. CBS (2008), Central Bureau of Statistics, Israel, available at http://www.cbs.gov.il, accessed 10 April 2009. Ellison, G. and E.L. Glaeser (1999), ‘The Geographic Concentration of Industry: Does Natural Advantage Explain Agglomeration?’, American Economic Review, 89(2), 311–16. Feldman, M.P. (2001), ‘The Entrepreneurial Event Revisited: An Examination of New Firm Formation in the Regional Context’, Industrial and Corporate Change, 10, 861–91. Feldman M.P., J.L. Francis and J. Bercovitz (2005), ‘Creating a Cluster While Building a Firm: Entrepreneurs and the Formation of Industrial Clusters’, Regional Studies, 39(1), 129–41. Florida, R. and M. Kenney (1988), ‘Venture Capital-Financing Innovation and Technological Change in the U.S.’, Research Policy, 17, 119–37. Foster J. and J.S. Metcalfe (2001), Frontiers of Evolutionary Economics: Competition, Self-Organization and Innovation Policy, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Gompers, P. and J. Lerner (1999), The Venture Capital Cycle, Cambridge, MA and London: The MIT Press. IAEI (2008), Israel Association of Electronics and Information Industries, available at http://www.iaei.org.il, accessed 10 April 2009. IVC (2001), IVC Yearbook 2001: A Survey of Venture Capital and Private Equity in Israel, Israel Venture Capital Research Center. IVC (2008), Israel Venture Capital Research Center Database, available at http:// www.ivc-online.co.il, accessed 10 April 2009. Klepper, S. (2001), ‘Employee Start-ups in High-Tech industries’, Industrial Corporate Change, 10(3), 639–74. Klepper, S. (2006), ‘The Evolution of Geographic Structure in New Industries’, Revue OFCE, June 2006. Marshall, A. (1890), Principles of Economics, 8th Edition, New York: Macmillan, 1948. Maskell, P. (2001), ‘Toward a Knowledge-Based Theory of the Geographical Cluster’, Industrial and Corporate Change, 10(4), 921–43. Menzel M.P. and D. Fornahl (2009), ‘Cluster Life Cycles – Dimensions and Rationales of Cluster Evolution’, Industrial and Corporate Change, 18(6), 1–34. Metcalfe J.S. and J. Foster (eds) (2004), Evolution and Economic Complexity, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Murmann, J.P. (2002), ‘The Coevolution of Industries and National Institutions: Theory and Evidence’, working paper, FSIV02.14, Social Science Research Center Berlin. Murmann, J.P. (2003), ‘The Coevolution of Industries and Academic Disciplines’, working paper, WP03-1, Kellogg School of Management, Northwestern University.

164

Emerging clusters

Nelson, R.R. (1994), ‘The Co-evolution of Technology, Industrial Structure and Supporting Institutions’, Industrial and Corporate Change, 3(1), 47–63 Nelson, R.R. (2001), ‘The Coevolution of Technology and Institutions as the Driver of Economic Growth’, Chapter 2 in J. Foster and J.S. Metcalfe (eds), Frontiers of Evolutionary Economics: Competition, Self-Organization and Innovation Policy, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Nelson, R.R. (2007), ‘What Enables Rapid Economic Progress: What are the Needed Institutions?’, Research Policy, 37(1), 1–11. OCS (2008), Office of Chief Science, Ministry of Industry, Trade and Labor (www. moit.gov.il), Israel, accessed 10 April 2009. OECD (2004), Venture Capital: Trends and Policy Recommendations, DSTI/DOC (2004), STI Working Papers, Paris. Pinch, S. and N. Henry (1999), ‘Paul Krugman’s Geographical Economics, Industrial Clustering and the British Motor Sport Industry’, Regional Studies, 33, 815–27. Porter, M. (1998), ‘Clusters and the New Economics of Competition’, Harvard Business Review, Nov–Dec, 77–92. Roelandt, J., V. Gilsing and J. Sinderen (2000), ‘Cluster-Based Innovation Policy: International Experience’, Erasmus University Rotterdam, Research Memorandum 0012. Saxenian, A. (1994), Regional Development: Silicon Valley and Route 128, Harvard University Press. Teubal, M. (1997), ‘A Catalytic and Evolutionary Approach to Horizontal Technological Policies’, Research Policy, 25, 1161–88. USPTO (United States Patent and Trademark Office) (2009), www.uspto.gov, accessed 10 April 2009. Utterback, J.M. and W.J Abernathy (1975), ‘A dynamic model of process and product innovation’, Omega, 3(6), 639–56.

7.

Standards as institutions supporting the cluster emergence process: the case of aquaculture in Chile Paola Perez-Aleman

1

INTRODUCTION

This chapter focuses on how institutions that foster collective learning interact with the process of cluster emergence. The argument is that standards, which are norms and agreements about product characteristics and production processes, constitute crucial institutions in the emergence and development of a cluster. Existing work highlights two elements as crucial in the early stages of cluster development: connecting to sources of market demand and firm building (Bresnahan et al. 2001). Standards interact in important ways with these two elements. First, the ability to compete in foreign markets depends on meeting the quality performance of prevailing international competitors. Second, standards encompass technological knowledge (Jacobsson 2000), which involve specific actions in production (Leblebici et al. 1991; Brunsson and Jacobsson 2000), and therefore can guide the firm’s capability building process in a new economic activity. They can assist in the individual and collective firm learning about how to produce products and improve production processes to take advantage of major market opportunities. Regional clusters, defined as geographical and sectoral concentration of firms operating in the same or related industries, are common in the developing world (Pietrobelli and Rabelloti 2006). Often, internationally competitive developing country firms grow within, along or tied to clusters (McKendrick et al. 2001; Saxenian and Hsu 2001; Giuliani et al. 2005; Perez-Aleman 2005; McDermott 2007). For example, successful Indian companies such as Wipro and Infosys are in Bangalore’s information technology cluster (Balasubramanyam and Balasubramanyam 1997; Ramamurti 2004). Similarly, Brazil’s Embraer (Goldstein 2002) and its global footwear companies (Schmitz 1999) are embedded in clusters of firms. The miraculous growth of the Taiwanese information technology 165

166

Emerging clusters

industry is concentrated in the Hsinchu region (Saxenian and Hsu 2001). In some cases, multinationals encourage clustering around local production sites by promoting the co-location of suppliers (Altenburg and MeyerStamer 1999; McKendrick et al. 2001; Patibandla and Petersen 2002). Research shows different factors affecting the emergence of clusters: the role of the state, concentration of skills in a geographical area, new technological and market opportunities, entrepreneurial activity, the role of venture capital, presence of universities, firm building, and market building, among others (Breschi and Malerba 2005; Perez-Aleman 2005; Feldman and Braunerhjelm 2006; Saxenian 2006). The argument here is not that these are irrelevant in the emergence of clusters, but rather that there is a need to understand better the institutional mechanisms at the local level in the early phases of a cluster that contribute to its subsequent growth and success. Existing studies show the many ways in which firms derive benefits from their location in clusters. One thing that is highlighted in the literature relates to the important role of knowledge flows among firms through interactions that are supported by institutions that foster collective learning (Breschi and Malerba 2005). Collective learning is defined here as new knowledge that a group of enterprises develops and that changes existing routines and production practices (Nelson and Winter 1982; Amin and Cohendet 2004). The literature shows the crucial role of networks for learning among firms by facilitating knowledge acquisition and innovation (Piore and Sabel 1984; Powell 1990; Saxenian 1996; Uzzi 1996; MacDuffie and Helper 1997). The existing literature highlights that developing country firms face initial technological and market disadvantages compared to advanced country firms that are established in international markets (Amsden 1989, 2001). Therefore, to compete internationally as latecomers, they must build their capabilities by fostering collective knowledge flows and creating institutions that support learning, that is, improvements in products and processes that meet the quality and productivity performance levels of established competitors. Interactive learning processes at the local and global levels are important for knowledge building and circulation (Bathelt et al. 2004). How do the interactions work? How does knowledge flow through these interactions? How do they work in the early stages of cluster formation? In the context of developing countries, standards can play a central role in facilitating diverse knowledge flows among firms. They act as institutions that support interactions among local firms where the cluster is emerging, as well as with foreign firms important for market demand. The discussion in this chapter builds on a case study from Chile,

Standards supporting the cluster emergence process

167

specifically the salmon aquaculture cluster. Compared to other Latin American countries, Chile is often cited as the top performer in the region. Its dynamic export performance was particularly outstanding in natural resource-based sectors, such as forestry, agroindustry, fishing and mining. The growth of many of Chile’s successful natural resource-based industries formed new clusters, as firms located in specific geographical regions within the country. The analysis of Chilean firms’ growth and their international expansion illustrates the interaction between the creation of local institutions that help build competitiveness and the emergence and growth of a local cluster. The Chilean firms’ ability to compete internationally was dependent on achieving products and processes that could match the level of established international competitors. Creating standards was a central dynamic in the process of fostering collective learning to help firms compete internationally. Standards are institutions that emerge through a dynamic interaction between firms, associations, and other public and private organizations. The analysis presented in this chapter gives insights about organizational and institutional mechanisms that enable not just one firm, but many firms to compete effectively in global markets leading to the emergence of a cluster of firms in the same related industry. Starting from specific problems and challenges that developing country firms face, their ability to compete globally lies in their collective capability to improve volume, product quality, processes and costs to establish a market advantage. The cluster level reveals the ways in which firms collectively cope with local institutional challenges or initial disadvantages compared to advanced country firms. In addition, this chapter highlights the active role of firms in creating local institutions to face the challenge of competing globally.

2

CLUSTER EMERGENCE, MARKETS AND INSTITUTIONS

The cluster concept has been subject to many debates. Several reviews have shown the different points of theoretical emphasis (Porter 2000; Martin and Sunley 2003; Breschi and Malerba 2005; Cortright 2005). The focus of this chapter is not to review the multiple ways of defining clusters, but instead to add to existing understanding on the emergence process of spatial agglomeration of economic activity. Most of the research has focused on well established, mature clusters, such as Silicon Valley. Yet, it is now widely accepted that the forces and conditions accounting for the emergence of new clusters differ from those involved in sustaining existing

168

Emerging clusters

clusters (Bresnahan et al. 2001). In the early stages, the elements associated with increasing returns due to agglomeration are not present. Two aspects seem particularly important in the early phases of cluster formation. First is the ability of firms to access major markets outside their cluster (Bresnahan et al. 2001). Second, a distinguishing feature of clusters that grow from those that fail to develop is the combination of entrepreneurial activity and the active building of institutions (Feldman and Francis 2004). The internal dynamics and institutions make a difference in the ability of firms to build their competitiveness in the early stages. The discussion of each of these aspects in the existing literature is developed next. Studies have identified that one of the regularities involved in starting and sustaining a cluster is the ability to connect with leading centers of demand (Bresnahan et al. 2001; Breschi and Malerba 2005).1 The transition from an early phase to a well-developed cluster involves a long process of market building. Connecting to these markets, however, may not be so easy, particularly because of the technological capabilities and market connections of the incumbents. Moreover, international markets may endorse product standards that may make entry difficult for newcomers. While it is obvious that demand is important for cluster growth, compliance is not so simple for firms trying to enter into a new economic activity in the early stages of cluster formation. This implies that access to major markets depends on firms developing new capabilities. For clusters to develop, however, many firms have to build capabilities to compete in such markets. In the early stage search for markets, how do new enterprises without foreign experience achieve such access? Several perspectives emphasize the role of institutions in cluster growth dynamics, both in the initial and subsequent growth phases. Institutions are norms, conventions, codes and rules that show certain regularity (Wagner 1994; Storper and Salais 1997; Biggart and Beamish 2003). In particular, the literature on industrial districts (Piore and Sabel 1984) focuses on the norms of reciprocity supporting technological dynamism. Cooperation and competition in the relations among firms sustain innovation. Similarly, Saxenian (1996) shows that external economies due to spatial proximity alone cannot explain the innovation and growth dynamic in Silicon Valley, emphasizing the relevant role of local institutions to coordinate decentralized production. At the collective level, the existence of common norms and conventions shapes the local interactions that are favorable to new activities (Storper and Salais 1997; PerezAleman 2005). While most of the literature focuses on established clusters, some authors also link institutions to the initial cluster formation stage, when both entrepreneurial activity and interactions increase (Feldman

Standards supporting the cluster emergence process

169

and Francis 2004; Feldman and Braunerhjelm 2006). During this phase, the institutions that support the development of firms and their products can stimulate innovation and localized learning. However, the types of institutions remain underspecified. For example, what institutions support the building of firms’ capabilities to access major markets? The organizational literature highlights that knowledge building and circulation at the collective level affects the ability of a group of firms to develop capabilities, survive, and enhance their competitiveness (Miner and Anderson 1999). Knowledge is intricately related to learning (Brown and Duguid 1991). This is an interactive process that builds on knowledge from a wide variety of sources. Learning implies a change in the firm’s practices and strategies, and it can occur at many levels: individual, group, industry or population of organizations (Miner and Anderson 1999). When a whole group or population of firms is able to discover, build and share new knowledge, this is called collective learning. For example, when firms develop norms related to performance standards or other coordination routines after observing another group of firms represents collective learning. The growth of a cluster depends on how effectively institutions foster learning among the collective of firms entering a given economic activity in the early stage of cluster formation. Firms actively construct frameworks, agreements and expectations that can create conditions for cluster emergence and development when these foster learning (Sabel 1994; Storper and Salais 1997). The argument presented in this chapter is that cluster emergence depends on institutions that encourage and support collective learning processes among firms that contribute to build their technological and organizational capabilities. The work on clusters in developing countries has shown how firms overcome many constraints by drawing on ‘collective efficiencies’ (joint action) (Nadvi 1999; Schmitz 1999), which enable many firms, especially small and medium-sized ones, to enter global markets, especially as exporters or as suppliers to global chains. These joint efforts make it possible to pool resources, increase productivity, and build individual and collective capabilities. Though the cluster literature has noted the relevance of institutions at the regional or cluster level, it still has not sufficiently explored how these institutions encourage collective learning among firms that contributes to building their capabilities, and that leads to cluster emergence and growth. In particular, how do institutions facilitate the flow of ideas, and support the firm and market building that is so crucial for firms to be viable in the marketplace and for the rise of a cluster? One way to support the ability of firms to connect to market demand is to build institutions that enable them to produce products and improve production processes that meet foreign requirements. This is where

170

Emerging clusters

standards can act as institutions supporting the local and global interactions that facilitate connections to demand and knowledge sources. Standards are explicit conventions, rules, norms, shared expectations, agreements and regulations. A standard concerns production practices, and different rules or expectations governing production affect how they are done (Storper and Salais 1997; Piore 2003). A quality standard refers to final product characteristics as well as production methods designed to meet certain expectations (Brunsson and Jacobsson 2000). These can be generic (that is, ISO 9000), or sector-specific (that is, Hazard Analysis of Critical Control Points (HACCP)). As knowledge creation and learning are the result of interactive processes between actors at the local and global levels, standards play a role in such dynamics. One of the frameworks in the cluster literature highlights that localized learning has a decisive role in clusters as co-located firms within the same industry and region can benefit from face-to-face contacts, meetings and communication between enterprises as they engage in problem-solving (Bathelt et al. 2004). Co-location can stimulate the creation of institutions as firms attempt to address common problems and build on shared experiences. At the same time, studies point out that learning is not solely based on local knowledge, as distant or global knowledge is as important (Bresnahan et al. 2001; Bathelt et al. 2004; Giuliani et al. 2005; PerezAleman 2005; Owen-Smith and Powell 2006). Knowledge flows from local and global interactions seem crucial for clusters to develop. In this sense, the success of entrepreneurial endeavors and the rise of a cluster depend on the ability to connect not only to foreign markets, but also to foreign knowledge. Standards can act as institutions that facilitate knowledge flows from foreign to local contexts as well as within localities. Connecting to global knowledge has particular relevance for latedeveloping firms. In the context of late development, compared to advanced country firms, developing country enterprises do not have the initial asset of pioneering technology (Amsden 1989, 2001; Hikino and Amsden 1994. Late-developing firms enter global markets facing intense pressure to improve quality and productivity, and reduce costs to the levels of advanced country multinationals who are established global competitors. They face the challenge of increasing their competitiveness by improving skills, managerial expertise, products, production processes and organization. As latecomers in global markets, they are building their competitiveness under very different circumstances than North American and European firms. The technological challenges that developing country enterprises face to compete globally means that learning processes are at the center of the

Standards supporting the cluster emergence process

171

firms’ journey to become competitive (Amsden 1989, 2001; Sabel 1994; Hausmann and Rodrik 2003; Perez-Aleman 2005). From the firm standpoint, capabilities emerge from processing and circulating knowledge (Zander and Kogut 1995). A key issue then is how and by what mechanisms developing country firms acquire, build and circulate the local and global knowledge needed to upgrade and compete internationally and gain access to major markets. The next section presents an analysis of Chilean firms’ growth and their international expansion to demonstrate how collective knowledge flows and the creation of institutions contributed to build the firms’ capabilities, access to foreign markets, and to support the emergence of a successful cluster.

3

THE CASE OF THE SALMON AQUACULTURE IN CHILE

Traditionally a mining country, copper products accounted for half of Chile’s exports in 1980 (Meller and Saez 1995). By 1998, copper represented only one third of Chile’s exports (Montero et al. 2000). The leading new export products included wood and pulp, farmed salmon, fruits and wine. In salmon farming, Chile experienced remarkable growth, becoming the largest exporter of farmed salmon in the world in nearly two decades (APSTC 2000–2004). The industry was export-oriented from the start, and over time the product’s value-added component grew significantly, and reached diversified foreign markets (Montero, 2004). More than 95 percent of the salmon cluster production is exported to more than 60 countries (Maggi 2006). In the 1970s, the idea of becoming a salmon exporter was just an experimental project of the Chilean government. When Chile entered the farmed salmon business in the 1980s, the world’s top producers included Norway, the UK, Scotland, Canada and the USA. In 1987, Chile produced a mere 1.5 percent of total world production (Montero 2004). By the mid-1990s, Chile became a major global producer of farmed salmon. By 2000, it accounted for 27 percent of the world production of farmed salmon, ranking second after Norway (AISC 2003). As shown in Figures 7.1 and 7.2, by 2003, Chile was on a par with Norway, producing 35 percent of total global production (AISC 2003).2 In one decade between 1990 and 2000, Chilean farmed salmon (including trout) production for export increased twelvefold, from 23 800 to 303 000 net tons (AISC 2003). Currently, Chile has moved into first place. Aquaculture represents one of the most important exporting sectors in Chile, as revenues more than doubled in a decade. They accounted for 56

172

Emerging clusters

Chile Norway Rest of World

Source:

AISC (2006).

Figure 7.1

Chile’s share of global salmon production, 2006

Chile Prod

World Prod ('000 tons)

2000 1500 1000 500 0 1985 Source:

1990

1995

2000

2003

AISC (2003).

Figure 7.2

Evolution of Chilean and world salmon production (in thousand tons)

percent of total fishing exports in 2000, compared to only 28 percent in 1990 (Aquanoticias 2001). Within Chilean aquaculture, salmon accounts for 95 percent of total export product volume. As shown in Figure 7.3, Chile’s annual salmon exports increased from US$38 million in 1989, to more than a billion dollars in 2003 (APSTC 2004).3 Chile’s salmon industry is concentrated in a territory known as the Tenth Region, located 1000 kilometers south of Santiago. Of registered

Standards supporting the cluster emergence process

173

Total Exports US$ Millions

1200 1000 800 600 400 200 0 1985 Source:

1990

1995

2000

2003

AISC (2003).

Figure 7.3

Evolution of Chile’s salmon exports, 1985–2003

fish farming centers in Chile, 81 percent (or 324 sites) are in the Tenth Region (Aquanoticias 2001). The Tenth Region accounts for 87 percent of total volume of salmon exported. In 2004, the Chilean salmon cluster was populated by some 140 enterprises involved in direct production, and another 100 involved in supporting services (Montero 2004). Some firms integrate the cultivation and production phase in fish farms with processing. Cultivation combines freshwater and saltwater farms, and processing involves plants to produce the final products for exports. Some 30 enterprises concentrate on egg hatchery production, while another 22 focus on producing smolts, the early stage of salmon life. The industry employs 28 000 people directly , and some 12 500 indirectly (Maggi 2006). The industry’s initial development in Chile was accompanied by the entrance and growth of mostly indigenous firms. From the late 1970s, when this activity was nonexistent, the number of Chilean firms engaged in the salmon farming industry grew to become what is now a cluster of firms (Maggi 2002; Montero 2004). Since the late 1990s, foreign investment from advanced country multinationals has been changing the local landscape of mainly indigenous firms. By 2004, foreign firms appeared among the top 15 exporters of Chilean farmed salmon. This case study of the Chilean salmon cluster is based on primary data collection at the firm and cluster levels. Most of the data discussed in this chapter draws on the author’s fieldwork in Chile’s Southern region during

174

Emerging clusters

2003. As the author has conducted other research projects in Chile, prior to and after 2003, there is a research familiarity with the Southern region and Chile in general. The information was collected from personal interviews and observations by visits to production sites in a sample of eight leading exporting firms in the cluster located in Chile’s Tenth Region, including Puerto Montt and Chiloe areas. The managers interviewed at each firm included: general, production, commercial and processing plant levels. Personal visits to plants and cultivation installations also provided important observations. Several interviews were also with the technical directors of the firms, such as the engineers involved with quality control and cultivation. Additionally, the author conducted interviews with one firm producing feed, with people identified as key historical informants, and with representatives from related government agencies (CORFO, PROCHILE, Fundacion Chile, SERNAPESCA, regional governments) and associations (APSTC). The personal interviews were based on openended question guides. In addition to collecting information via interviews, other sources include bibliographical material from companies, associations, government, and published sources.

4

THE ROLE OF GOVERNMENT IN THE AQUACULTURE CLUSTER’S EMERGENCE

Unlike other current exporters (Norway, the UK, Canada and the USA), salmon is not a native species in Chile. It was introduced after much experimentation by the Chilean government in the 1960s and 1970s.4 Government research and experiments uncovered new possibilities and business opportunities for local entrepreneurs (Perez-Aleman 2005). These experiments established that salmon could grow domestically in Chile. Until this point, Chile had a fishing industry based on extraction of existing fishing stock for the domestic market. In 1969, a joint venture between Chile’s National Fisheries Service (Servicio Nacional de Pesca or SERNAP) and the Japan International Cooperation Agency (or JICA), sparked the beginning of a serious salmon farming program (Fundación Chile 2000).5 This was followed by another cooperation agreement begun in 1988 that involved Chile’s Economic Development Corporation (Corporacion de Fomento, CORFO), through its affiliate Fisheries Development Institute (Instituto de Fomento Pesquero, IFOP) and JICA.6 They identified suitable river and ocean sites for salmon and trout farming activity in the Tenth Region.7 Also, these programs created opportunities to acquire experience with ocean ranching and cultivation techniques, particularly those related to nutrition, disease

Standards supporting the cluster emergence process

175

control and fish transport. While the initial natural resource conditions explain why the industry concentrated in this region, this advantage did not assure automatic economic success. Farming presented several technological challenges due to the salmon’s lifecycle, its nutrition requirements, disease susceptibility and environmental management that Chilean producers were required to learn.8 Rather than centralize investment and ideas, the state nurtured their quick flow to private firms. The public research program became an important source for independent firm start-ups by new groups of entrepreneurs, who drew on the knowledge and skills developed in the public sector. In 1974, professionals that had worked in the government programs of SERNAP and IFOP formed their own aquaculture companies for commercial purposes (Fundación Chile 2000). This move of professionals from the public to the private sector helped to transfer skills and information between the two. With a government loan from CORFO in 1975, for example, the Sociedad de Pesqueria Llanquihue started as the first commercial farming venture in Chile. It was eventually the first to export to Europe. At the same time, the government’s venture capital agency, the Fundación Chile, invested capital to create a firm that transferred foreign technology and developed local know-how.9 During the early 1980s, the technical development input of Fundación Chile and its operations wing, Salmones Antártica, contributed to the take-off of salmon aquaculture. Fundación Chile broke new ground when it began to put hatchery-reared smolts (juvenile salmon) into cages in the sea for their main phase of growth.10 It facilitated the transfer of aquaculture technology, such as the floating net pen used in Scotland and Norway, to Chile. It also adopted new techniques, such as rearing in tanks instead of the open river (Achurra 1995). The Fundación acted as technical consultant to private firms interested in entering this activity, and conducted research continuously. These initial ventures uncovered new business possibilities, and had significant demonstrative effects to local firms. The idea of producing farmed salmon became attractive as two commercial ventures demonstrated success. The number of domestic firms in salmon farming grew significantly in the 1980s. The first generation of investors was mostly local entrepreneurs or Chilean economic groups from other sectors (industry, construction, forestry and fishing) (Montero et al. 2000). In 1980, there were three private enterprises. By 1985, there were 36 enterprises. Many of these were started by professionals who had worked previously in the public salmon research program. By 1987, some 120 firms were involved in ocean ranching; of these, about 42 enterprises accounted for 85 percent of total production (Achurra 1995). By 1997, there were 219 firms exporting salmon (Bjorndal and Aarland 1999). In addition to firms engaged in production,

176

Emerging clusters

other firms entered the industry in the Tenth Region, to become involved in egg hatcheries, feed production, cage manufacturing, product processing, refrigerated containers and transport services (Maggi 2002). All of these firms concentrated in the Tenth Region. Whereas in 1993 there were only 21 firms in services, by 2002 there were more than 100 enterprises providing services to salmon producers and processors, including labs, nets producers, containers and transportation. Government financing, as well as internally generated private capital from existing firms, supported the entrance of new firms into salmon aquaculture activities.

5

CREATING STANDARDS TO SUPPORT COLLECTIVE KNOWLEDGE FLOWS DURING CLUSTER EMERGENCE

This section presents a discussion of crucial challenges that Chilean firms faced to establish themselves in advanced country markets during the emergence phase: (1) the pressure to improve quality and volumes to the level of established international competitors; (2) mastering production capability; (3) connecting to markets in advanced countries. I focus on standards created during the emergence phase to highlight how they foster knowledge flows and support cluster development. Standards for Quality Control At the beginning, the firms were small enterprises trying to produce for export markets. Both volume and reputation were necessary factors to sell their products abroad. In the early cluster formation stage, firms wanted to connect with diverse foreign markets where Chilean salmon was unknown (Maggi 2002). The first generation of investors in the salmon industry (between 1982 and 1987) included new local entrepreneurs, independent firms involved in extractive fishing and Chilean firms from other sectors (agro-industry, forestry and fishing) that invested their capital in this new activity (Montero et al. 2000). Many were started by professionals who had worked previously in the government’s salmon research program (Perez-Aleman 2005). It was a diverse set of firms that entered the salmon business activity. A major challenge for Chilean firms was how to supply highly discriminating international market demand when the country had little experience producing salmon products. Chilean firms entered the salmon business without having the competitive position of leading established producers like Norway, Scotland and other European and North American

Standards supporting the cluster emergence process

177

companies. In particular, the novice Chilean firms faced intensive pressure to achieve the quality standards and volumes that advanced countries were already capable of meeting. From the beginning, Chilean firms targeted the advanced country markets of Japan, the USA and Europe. Attracting foreign clients depends on producing at prevailing international standards. This is in contrast to the assumed pattern of expansion from domestic to foreign markets (Aulakh et al. 2000). In the early 1980s, Chilean salmon was not a known product compared to Norwegian or Scottish products (Maggi 2002). As they had no brand name like the established advanced country multinationals, Chilean firms set out to develop their ‘country name’ as their collective brand (Author Interviews # 7, 17 and 33, 2003). Not having an established presence in world markets, their first goal became to convince customers in Japan and the USA that their product could meet the stringent quality standards of advanced countries. Volumes and quality necessary to be globally competitive required that the firms coordinate their collective productive activity. To expand internationally, Chilean firms’ strategy focused on developing product quality standards. Some aquaculture firms had previous experience in exporting fresh fruits, and had learned that meeting international norms was crucial to export to the USA (Author Interview # 7, 2003). In addition, meetings with Japanese and American buyer firms provided important knowledge flows on how to define norms (Author Interview # 34, 2003). As a Chilean entrepreneur said: ‘we learned a culture of quality under the wings of the Japanese customers’ (Author Interview # 34, 2003).11 As well, the Fundacion Chile, a quasi-government agency, acted as a key advisor in the process of developing new norms for salmon aquaculture (Author Interview #6, 2003). In 1986, 17 firms formed the Association of Salmon and Trout Producers of Chile (Achurra 1995; APSTC 2002). The APSTC currently has 42 affiliated enterprises that account for 85 percent of salmon production (APSTC 2002).12 The APSTC’s goal was to insure that foreign customers would see Chile as a permanent source, with the capacity to insure sufficient volumes of quality product, thereby creating a collective reputation for Chilean salmon abroad. Through the Association, Chilean firms established an institutional framework to coordinate their activity. They established a ‘quality seal’ certification through compliance with self-monitored processing and product standards. As firms strategized to establish a reputation in this early stage, the newly developed standards classification and certification guided each firm’s product improvements to develop exports and attract foreign clients. The firms associated in APSTC voluntarily developed strict and detailed export product standards as well as processing

178

Emerging clusters

standards. The ‘Code of Standards for Chilean Salmon’ that emerged provided detailed information on how to evaluate each stage of farming and processing (Achurra 1995). The Association established different product classifications (Premium, Grade I and Industrial Quality) and a seal in the exterior of each box indicates which category the product meets (Author Interview # 33, 2003). This self-imposed quality certification seal became a key mechanism for fostering knowledge flows between firms. The APSTC standards applied to members and non-members to rigorously control the quality of the fresh and frozen salmon exports. The quality standard was more than ‘putting a seal’ as it entailed adopting a set of practices at the firm level to insure final product characteristics. As such, their adoption affected production routines and required organizational capability at the firm and collective levels. For example, the adoption of Hazard Analysis and Critical Control Points (HACCP) has a set of routines that influence the way firms conduct their production practices. As well, HACCP implies developing the capacity to monitor each step of the production process, rather than simply evaluating the final product. This requires capabilities for analyzing bacterial, chemical, and supply dimensions and for avoiding contamination. Adopting this set of routines also increased meetings, face-to-face interactions, and shared problemsolving among and within firms. The APSTC also set up a quality control unit in charge of providing the product quality certification seal to those firms that met the high standards of the association. The quality seal was given after an inspection conducted by private independent certifying companies. Fundación Chile acted as the certifying company. In the mid-1980s, it had 40 inspectors visiting each plant for quality control (Author Interview # 12, 2003). In this process, the APSTC’s quality control unit and the Fundación Chile inspectors diffused information to all firms about how to improve performance. In the 1980s, the quality certification process would take several days as it included training while inspecting (Author Interview # 12, 2003). As firms advanced their knowledge and improved their product, Chilean salmon exports began to grow significantly. The government supported the private firms’ collective initiatives by financing the implementation of product quality certification (Maggi 2002). It also used the firms’ agreement as the basis for making national regulation inspired by the standards set by the industry. The standards developed by the APSTC were later adopted by the National Fisheries Service (SERNAPESCA), the public authority in charge of fishing regulation, evolving into mandatory quality norms for any plant operating in Chile (Montero et al. 2000). Prior to this point, there was no state regulation or enforcement of salmon production standards.

Standards supporting the cluster emergence process

179

Regulations for Improving the Production Process Parallel to the issue of quality control to achieve a world standard product, the Chilean firms faced the challenge of mastering production capability. Initially, Chilean firms imitated the most obvious aspects of salmon cultivation, such as the system of ocean cages used in advanced countries. To be competitive, however, they directed major efforts at process improvements: increasing productivity during incubation and growth phases, reducing fish mortality rates and improving feed conversion rates. Salmon cultivation is a complex process: managing live species, dispersed sites, sensitive reproduction and growth stages; highly demanding logistics in terms of movement intensity, coordination and timing; and complex feed technology and delivery. At the beginning, Chilean firms had high fish mortality rates due to fish diseases, causing major economic losses (SalmonChile 2003). The firms and the APSTC made the issue of fish health (and mortality reduction) a priority. The sustainability of the industry, particularly the problems of environmental contamination from feed and chemical treatments was a related issue (Barton 1997). Water quality and disease transmission are critical for the demanding requirements of salmon farming. The firms jointly developed important regulations, such as the minimum distance between concessions for farming in the ocean or bay. Firms addressed jointly the problem of disease control. All of this occurred in a context where there was no national legislation, nor legislation to regulate aquaculture industry activity of any kind (Author Interviews # 20 and 21, 2003). The Association established the Salmon Technology Institute (INTESAL) in 1993 with 45 percent financial assistance from the Chilean Development Corporation (CORFO) (Barton 1997). INTESAL provided information on how to improve production practices to reduce disease. The Association uses this information to influence its member and nonmember companies to establish the latest disease management and sustainable production strategies. The firms initially established agreements on general sanitary management norms in hatcheries and cultivation, but later agreed on more detailed sanitary practices for farm sites: pathogen screening, rotation of cultivation centers, monitoring programs, how to control breakouts and feeding techniques (SalmonChile 2003). Government financing supported joint research projects between Chilean universities and firms to investigate the causes of fish mortality and strategies to reduce it (Author Interviews # 3, 20, 2003; and Maggi 2002).

180

Emerging clusters

Connecting to Advanced Country Markets through Joint Marketing As Chilean firms attempted to gain access to foreign markets, there were other important collaborative experiences that helped them to face challenges together. Given their small volumes in the incipient business years of the 1980s, their ability to meet large volume orders was weak, and limited by the individual production capacity of each company. By 1990, the majority of Chilean enterprises still had volumes that did not surpass 1000 tons annually. In the beginning, occasional buyers would visit Chilean producers who sold their product as a commodity and would negotiate a small purchase contract (SalmonChile 2003). Many Chilean firms felt that given their distant location from the advanced country markets, buyers would not come to them in search of a highly perishable product (Author interview # 7, 2003). The Japanese, US and Norwegian firms were closer in distance to the advanced country markets. Moreover, the Chilean firms had little experience in managing distribution channels. In 1990, 13 Chilean enterprises formed an association to commercialize salmon, as they felt too dependent on buyers from their two major markets, the USA and Japan. Salmocorp became the joint venture to commercialize and market their product together. Jointly, the 13 firms accounted for 30 percent of salmon production in 1990 (Achurra 1995). At this point, because Chilean firms had little product variety, they could negotiate joint foreign sales contracts. For example, Invertec, which was a member of Salmocorp, had only three types of products in the 1980s (Author Interview # 7, 2003). Through Salmocorp, the companies sold 100 percent of their export production. Besides negotiating sales contracts together, this alliance aimed to improve the international market positioning of Chilean salmon, as Chilean firms felt at a disadvantage compared to foreign firms. Foreign producers had their own marketing departments that were part of the multidivisional structure of their parent companies. Salmocorp gave visibility to Chilean producers, and enabled connection to new customers in the USA and Japan, as well as new markets in Asia (Taiwan, Singapore, China), Europe and Latin America, further diversifying demand for their products (Maggi, 2002). Also, by joining their production, Chilean firms could deal with large volume clients in their main markets, the USA and Japan. Significantly, Chilean firms began to establish direct relations with large food supermarket chains, rather than going through traders (SalmonChile 2003). These linkages provided ideas for new products, which Chilean firms used to strengthen their competitiveness in value-added products. The Salmocorp alliance lasted six years and dissolved in 1996, when firms felt they had accomplished the goal of diversifying their foreign markets. The dissolution of this alliance was the

Standards supporting the cluster emergence process

181

result of their marketing success, increased production volumes and diversified products in the context of growing foreign demand. This collective effort increased production coordination among the firms involved as they tried to sell and promote their products together. This provided another channel for collective learning about global buyers, product requirements, new product ideas, distributions and foreign marketing activities. As Chilean enterprises began to diversify their products, they moved away from their initial commodity sales toward more elaboration, new cuts and new packaging that allowed them to produce more value-added products. In this process, Chilean firms expanded and improved local standards focused on managing product safety and quality control of processed products. Going from the focus on producing whole fish to produce fillet, boneless, smoked and ready-to-serve products facilitated the subsequent growth of the firms and the cluster.

6

FIRM BUILDING AND CLUSTER EMERGENCE: LESSONS FROM THE CHILEAN CASE

Evaluating the Chilean salmon cluster experience raises several issues. First, the Chilean firms’ experience suggests that inter-firm networks play a crucial role in their ability to compete and expand in global markets. The emergence of networks characterize the process of joint problemsolving to face multiple international challenges, such as: the Association of Chilean Salmon Producers, the joint alliance through Salmocorp and Salmofood, the ties with local university research departments, the links with government agencies such as CORFO and Fundación Chile, and the ties to large foreign buyers, are just some of the examples noted in the case discussion. This finding supports a large literature highlighting networks in achieving dynamic growth (Piore and Sabel, 1984; Saxenian, 1996; Enright, 1999; Schmitz, 1999, Breschi and Malerba, 2001; Saxenian and Hsu, 2001; Pietrobelli and Rabellotti, 2003; Giuliani and Bell, 2005; among others). As important, the Chilean firms’ experience indicates that their ability to compete globally depends on improving technological and productive capabilities. The case illustrates the kinds of challenges these firms faced to improve product quality, and master process organization. Building their capabilities depended on fostering collective knowledge flows to improve quality and productivity to the level of established international competitors. Foreign firms played a role as source of foreign demand, and as suppliers of crucial inputs. As global customers and global suppliers of technology, foreign multinationals were an important source of ideas.

182

Emerging clusters

The Chilean case indicates that institutions such as standards fostered knowledge flows across firms, and between local and distant locations. Chilean firms and government actively created institutions through agreements on product quality standards, sanitary norms in plant processing and cultivation sites, fish disease control standards, setting strict targets for fish mortality rates, dissemination of university research, rules for maritime concessions location, waste management standards, environmental sustainability, and so on. These collective agreements and norms at the level of the industry or group, elaborated between firms, or between them and the government, contributed to knowledge generation and distribution, and skill-building as firms interacted in the emerging cluster. Chilean firms actively created institutions through dynamic interaction that facilitated knowledge diffusion across many enterprises enabling many firms to grow and compete, and to successfully expand in international markets. The Chilean aquaculture experience gives insights on the kinds of institutional mechanisms that support cluster emergence. These institutions emerge from the dynamic interaction between private and public actors as they attempt to improve the firms’ productive capabilities. Cluster level institutions pushed product and process improvements and facilitated the flow of ideas among many firms. Standards are an important institutional mechanism that can foster knowledge flows for enhancing competitive performance. Cluster emergence depends on creating institutions that foster collective learning. The cluster develops as firms transform products, improve processes and organization, and increase connections to markets. This study has shown how the local and global knowledge flows through standards can contribute to build crucial capabilities of firms in the early stages of cluster formation, when tapping into external market demand can be decisive for the cluster to develop and grow. Existing literature has made it clear that an active search for large markets is crucial for understanding the rise of clusters, and that institutions play a role in the early cluster stages. This study contributes to our understanding of cluster formation processes by highlighting that standards are one type of institution facilitating local and global interactions that support collective learning among firms to access major global markets.

NOTES 1.

2.

Other regularities highlighted during the early phase of cluster formation include: a plentiful supply of skilled labor, either from local sources or from foreign-educated people or large incumbent firms (Breschi and Malerba 2001) and external shocks (Feldman and Braunerhjelm 2006). Globally, the farmed salmon or aquaculture industry has grown significantly. Farmed

Standards supporting the cluster emergence process

3.

4. 5. 6. 7.

8. 9. 10. 11. 12.

183

salmon accounts today for 68 percent of total world salmon production, compared to a mere 20 percent in 1987 (AISC 2003). With respect to dollar value, salmon exports accounted for 1.8 percent of total Chilean exports in 1991, and increased to 5.4 percent in 2000 (APSTC 2001). Excluding copper exports, salmon exports represented 8 percent of total non-copper exports in 2000. For more on the initial stage of emergence of Chile’s salmon cluster and the role of the Chilean government, see Perez-Aleman (2005). The Chilean government initially focused on open fishing development. It tried to populate the area with salmon for future open fishing (that is, artificial wild), not ‘farming’. Ocean ranching came later. IFOP (Fisheries Development Institute) is the fishing industry development agency established in 1965 as part of CORFO, the government development corporation. IFOP focused on technological research for production. Salmon farming is complex, requiring hatchery production in sweet water with later growth in salt water. The industry has been limited to geographic locations that meet the suitable conditions, such as North America, Norway and Scotland. Chile is the most recent addition to the list of producing countries. During the 1950s and 1960s salmon aquaculture emerged for the first time in Japan and the USA. Fish farming was consolidated in the late 1970s (Barton 1997, p. 313). On the role of the Fundación Chile, a quasi-government development agency created with mixed public–private funding to support local entrepreneurial development, particularly in the fruit, vegetable and fish industries, see Huss (1991). Cages are floating structures, like cage boats developed in Europe, Japan and North America. The Fundación Chile conducted experiments to cultivate both Pacific and Atlantic salmon. Currently, the USA and Japan together account for 80 percent of the value of Chilean exports (Maggi 2006). The Asociacion de Productores de Salmon y Truchas (APSTC) changed its name recently to Asociacion de la Industria del Salmon (AISC), or SalmonChile.

REFERENCES Achurra, M. (1995), ‘La experiencia de un nuevo producto de exportacion: los salmones’, in P. Meller and R. Saez (eds), Auge exportador chileno: lecciones y desafios futuros, Santiago: CIEPLAN, DOLMEN. AISC (2003), ‘Statistics of the Chilean Salmon Industry Association’, Santiago: AISC, Asociacion de la Industria del Salmon de Chile. AISC (2006), ‘Statistics of the Chilean Salmon Industry Association’, Santiago: AISC, Asociacion de la Industria del Salmon de Chile. Altenburg, T. and J. Meyer-Stamer (1999), ‘How to Promote Clusters: Policy Experiences from Latin America’, World Development, 27(9), 1693. Amin, A. and P. Cohendet (2004), Architectures of Knowledge: Firms, Capabilities, and Communities, Oxford: Oxford University Press. Amsden, A.H. (1989), Asia’s Next Giant: South Korea and Late Industrialization, New York: Oxford University Press. Amsden, A.H. (2001), The Rise of ‘the Rest’: Challenges to the West from LateIndustrializing Economies, Oxford: Oxford University Press. APSTC (2000–2004), Annual Report, Santiago: APSTC. Aquanoticias (2001), Revision Aauanoticias, Santiago: Fundación Chile. Aulakh, P.S., M. Kotabe and H. Teegen (2000), ‘Export Strategies and Performance

184

Emerging clusters

of Firms from Emerging Economies: Evidence from Brazil, Chile, and Mexico’, Academy of Management Journal, 43(3), 342–61. Balasubramanyam, V.N. and A. Balasubramanyam (1997), ‘International Trade in Services: The Case of India’s Computer Software’, The World Economy, 20(6), 829–43. Barton, J.R. (1997), ‘Environment, Sustainability and Regulation in Commercial Aquaculture: The Case of Chilean Salmonid Production’, Geoforum, 28(3), 313–28. Bathelt, H., A. Malmberg and P. Maskell (2004), ‘Clusters and Knowledge: Local Buzz, Global Pipelines and the Process of Knowledge Creation’, Progress in Human Geography, 28(1), 31–56. Biggart, N.W. and T.D. Beamish (2003), ‘The Economic Sociology of Conventions: Habit, Custom, Practice, and Routine in Market Order’, Annual Review of Sociology, 29, 443–64. Bjorndal, T. and K. Aarland (1999), ‘Salmon aquaculture in Chile’, Aquaculture Economics and Management (Blackwell Science), 3(3), 238. Breschi, S. and F. Malerba (2001), ‘The Geography of Innovation and Economic Clustering: Some Introductory Notes’, Industrial and Corporate Change, 10(4), 817. Breschi, S. and F. Malerba (2005), ‘Clusters, Networks, and Innovation: Research Results and New Directions’, in S. Breschi and F. Malerba (eds), Clusters, Networks, and Innovation, Oxford: Oxford University Press. Bresnahan, T., A. Gambardella and A. Saxenian (2001), ‘‘‘Old economy” inputs for “new economy” outcomes: Cluster formation in the new Silicon Valleys’, Industrial and Corporate Change, 4. Brown, J.S. and P. Duguid (1991), ‘Organizational Learning and Communitiesof-Practice: Toward a Unified View of Working, Learning, and Innovation’, Organization Science, 2(1), 40–56. Brunsson, N. and B. Jacobsson (2000), ‘The Contemporary Expansion of Standardization’, in N. Brunsson and B. Jacobsson (eds), A World of Standards, Oxford: Oxford University Press, pp. 1–20. Cortright, J. (2005), ‘Making Sense of Clusters’, The Brookings Institution Metropolitan Policy Program, Washington, DC: The Brookings Institution. Enright, M.J. (1999), ‘Regional Clusters and Firm Strategy’, in A. Chandler, J.P. Hagstrom and O. Solvell (eds), The Dynamic Firm: The Role of Technology, Strategy, Organization, and Regions, New York: Oxford University Press. Feldman, M.P. and P. Braunerhjelm (2006), ‘The Genesis of Industrial Clusters’, in P. Braunerhjelm and M. Feldman, Cluster Genesis: Technology Based Industrial Development, Oxford: Oxford University Press, pp. 1–15. Feldman, M.P. and J.L. Francis (2004), ‘Homegrown Solutions: Fostering Cluster Formation’, Economic Development Quarterly, 18(2), 127–37. Fundación Chile (2000), El Libró del Salmón, Santiago: Fundación Chile. Giuliani, E. and M. Bell (2005), ‘The Micro-determinants of Meso-level Learning and Innovation: Evidence from a Chilean Wine Cluster, Research Policy, 34, 47–68. Giuliani, E., R. Rabellotti and M. Pieter van Dijk (eds) (2005), Clusters Facing Competition: The Importance of External Linkages, Aldershot: Ashgate Publishing. Goldstein, A. (2002), ‘Embraer: From National Champion to Global Player’, Cepal Review, 77.

Standards supporting the cluster emergence process

185

Hausmann, R. and D. Rodrik (2003), ‘Economic Development as Self-discovery’, Journal of Development Economics, 72(2), 603–33. Hikino, T. and A. Amsden (1994), ‘Staying Behind, Stumbling Back, Sneaking Up, Soaring Ahead: Late Industrialization in Historical Perspective’, in W.J. Baumol, R.R. Nelson and E.N. Wolff (eds), Convergence of Productivity: Cross-National Studies and Historical Evidence, New York: Oxford University Press. Huss, T. (1991), ‘Transfer of Technology: The case of the Chile Foundation’, Cepal Review, 97–115. Jacobsson, B. (2000), ‘Standardization and Expert Knowledge’, in N. Brunsson and B. Jacobsson (eds), A World of Standards, Oxford: Oxford University Press, pp. 40–50. Leblebici, H., G.R. Salancik, A. Copay and T. King (1991), ‘Institutional Change and the Transformation of Interorganizational Fields – An Organizational History of the United States Radio Broadcasting Industry’, Administrative Science Quarterly, 36(3), 333–63. MacDuffie, J.P. and S. Helper (1997), ‘Creating Lean Suppliers: Diffusing Lean Production Throughout the Supply Chain’, California Management Review, 39(4), 118. Maggi, C. (2002), ‘Cadenas Productivas: Lecciones de la Experiencia Internacional y Regional: El Cluster del Cultivo y Procesamiento del Salmon en la Region Sur-Austral del Chile’, Paper based on the Proyecto Cooperacion Tecnica: BIDFOMIN Trust Fund Italiano. Maggi, C. (2006), ‘The Salmon Farming and Processing Cluster in Southern Chile’, in C. Pietrobelli and R. Rabellotti (eds) , Upgrading to Compete: Global Value Chains, Clusters, and SMEs in Latin America, Washington, DC: Inter-American Development Bank and David Rockefeller Center for Latin American Studies, Harvard University, pp. 109–40. Martin, R. and P. Sunley (2003), ‘Deconstructing Clusters: Chaotic Concept or Policy Panacea?’, Journal of Economic Geography, 3, 5–35. McDermott, G. (2007), ‘The Politics of Institutional Renovation and Economic Upgrading: Recombining the Vines that Bind in Argentina’, Politics and Society, 35(1), 103–43. McKendrick, D.G., R.F. Doner and S. Haggard (2001), From Silicon Valley to Singapore: Location and Competitive Advantage in the Hard Disk Drive Industry, Stanford, CA: Stanford University Press. Meller, P. and R. Saez (1995), Auge exportador chileno: lecciones y desafios futuros, Santiago: CIEPLAN: Dolmen Ediciones. Miner, A. and P. Anderson (1999), ‘Industry and Population-level Learning: Organization, Interorganizational and Collective Learning Processes’, Advances in Strategic Management, 16, 1–30. Montero, C. (2004), ‘Formación y desarrollo de un cluster globalizado: del salmón en Chile’, Desarrollo productivo, No. 145, Santiago: ECLAC. Montero, C., C. Maggi and C. Parra (2000), ‘La industria del salmon en la X region: un cluster globalizado’, Santiago: ECLAC. Nadvi, K. (1999), ‘Collective Efficiency and Collective Failure: The Response of the Sialkot Surgical Instrument Cluster to Global Quality Pressures’, World Development, 27(9), 1605–26. Nelson, R. and S. Winter (1982), An Evolutionary Theory of Economic Change, Cambridge, MA: Harvard University Press.

186

Emerging clusters

Owen-Smith, J. and W.W. Powell (2006), ‘Accounting for Emergence and Novelty in Boston and Bay Area Biotechnology’, in P. Braunerhjelm and M. Feldman, Cluster Genesis: Technology Based Industrial Development, Oxford: Oxford University Press, pp. 61–83. Patibandla, M. and B. Petersen (2002), ‘Role of Transnational Corporations in the Evolution of a High-Tech Industry: The Case of India’s Software Industry’, World Development, 30(9), 1561. Perez-Aleman, P. (2005), ‘Cluster Formation, Institutions and Learning: the Emergence of Clusters and Development in Chile’, Industrial & Corporate Change, 14(4), 651–77. Pietrobelli, C. and R. Rabelloti (2003), ‘Upgrading in Clusters and Value Chains in Latin America: the Role of Public Policies’, report for Agorá 2000 and the Micro and SME Division, Department of Sustainable Development (SD/ MSM), Inter-American Development Bank, Washington, DC. Pietrobelli, C. and R. Rabelloti (eds) (2006), Upgrading to Compete: Global Value Chains, Clusters and SMEs in Latin America, Washington, DC: Inter-American Development Bank (IADB) and David Rockefeller Center for Latin American Studies, Harvard University. Piore, M.J. (2003), ‘Stability and Flexibility in the Economy: Reason and Interpretation in Economic Behavior’, Paper presented at colloque Conventions et Institutions: Approfondissements Theoriques et Contributions au Debate Politique, Paris, December, available at httpp://econ-www.mit.edu/files/ 1117. Piore, M.J. and C.F. Sabel (1984), The Second Industrial Divide: Possibilities for Prosperity, New York: Basic Books. Porter, M.E. (2000), ‘Location, Competition, and Economic Development: Local Clusters in a Global Economy’, Economic Development Quarterly, 14(1), 15. Powell, W. (1990), ‘Neither Market nor Hierarchy: Network Forms of Organization’, Research in Organizational Behavior, 12, 295–336. Ramamurti, R. (2004), ‘Internationally Competitive Clusters in Developing Countries: India’s Information Technology Industry’, Paper presented at the Conference on Multinational Corporations and Global Poverty Reduction, Storrs, CT: University of Connecticut. Sabel, C.F. (1994), ‘Learning by Monitoring: The Institutions of Economic Development’, in N.J. Smelser and R. Swedberg (eds), The Handbook of Economic Sociology, New York: Princeton University Press and Russell Sage Foundation, pp. 137–65. SalmonChile (2003), Aquaculture in Chile, Santiago: Technopress. Saxenian, A. (1996), Regional Advantage: Culture and Competition in Silicon Valley and Route 128, Cambridge, MA: Harvard University Press. Saxenian, A. (2006), The New Argonauts: Regional Advantage in a Global Economy, Cambridge, MA: Harvard University Press. Saxenian, A. and J.-Y. Hsu (2001), ‘The Silicon Valley-Hsinchu Connection: Technical Communities and Industrial Upgrading’, Industrial and Corporate Change, 10(4), 893. Schmitz, H. (1999), ‘Global Competition and Local Cooperation: Success and Failure in the Sinos Valley, Brazil’, World Development, 27(9), 1627. Storper, M. and R. Salais (1997), Worlds of Production: The Action Frameworks of the Economy, Cambridge, MA: Harvard University Press. Uzzi, B. (1996), ‘The Sources and Consequences of Embeddedness for the

Standards supporting the cluster emergence process

187

Economic Performance of Organizations: The Network Effect’, American Sociological Review, 61(4), 674–98. Wagner, P. (1994), ‘Dispute, Uncertainty and Institution in Recent French Debates’, Journal of Political Philosophy, 2(3), 270–89. Zander, U. and B. Kogut (1995), ‘Knowledge and the Speed of the Transfer and Imitation of Organizational Capabilities: An Empirical Test’, Organization Science, 6(1), 76.

PART III

Patterns of emergence and growth

8.

The evolution of the banking cluster in Amsterdam, 1850–1993: a survival analysis Ron Boschma and Floris Ledder

1.

INTRODUCTION

A research challenge in economic geography is how clusters evolve over time (Audretsch and Feldman, 1996; Feldman and Schreuder, 1996; Maggioni, 2002; Brenner, 2004; Feldman and Francis, 2004; Visser and Boschma, 2004; Feldman et al., 2005; Iammarino and McCann, 2005; Menzel and Fornahl, 2007; Ter Wal and Boschma, 2009). While there is quite some understanding of how clusters develop once they are in place, there is as yet still little understanding of how clusters emerge, and where. Do clusters arise from scratch, or are there preconditions that sustain the rise and development of clusters? This chapter elaborates on the window of locational opportunity concept developed in the 1990s (Storper and Walker, 1989; Boschma, 1997). This concept will be integrated in the literature on industrial dynamics, especially with respect to ideas developed by Arthur (1994) and Klepper (2007). The key question is through which mechanisms these routines diffuse and cluster spatially when a new industry emerges and grows (Boschma and Frenken, 2003). Two mechanisms have drawn special attention in the literature, that is, spinoff dynamics and agglomeration economies. Both may act as vehicles through which knowledge and routines are created and diffused among a growing population of firms within a territory. Spinoff dynamics is considered a driving force behind the growth of industries in space, because it transfers and diffuses relevant knowledge from incumbent firms to new firms (Helfat and Lieberman, 2002). Once spatial clustering occurs, agglomeration economies may also become manifest. Local knowledge spillovers will become increasingly available, which will cause a further spatial concentration of that industry. As such, both mechanisms provide (alternative or complementary) explanations for why an industry develops and concentrates in space. 191

192

Emerging clusters

These ideas have been tested on manufacturing industries, like the automobile and tyre industries (Boschma and Wenting, 2007; Klepper, 2007). Only a couple of studies have focused on non-manufacturing industries from an industry life cycle perspective (for example, Fein, 1998; Carree, 2003; Grote, 2007). We investigate the evolution of a service sector, that is the banking sector, and see whether the same findings apply. To our knowledge, no study has investigated the spatial evolution of a service sector, an exception being Wenting (2008) who investigated the long-term development of the fashion industry. Since we analyse data on the headquarters of banks (we have no data at the level of offices/branches of banks), we account for the most knowledge-intensive part of this service industry where firm-specific routines are formed. This enables us to investigate how (new) routines in a growing population of firms diffuse in space. The aim of this chapter is to describe and explain the rise of the banking cluster of Amsterdam since 1850 from an evolutionary perspective. This analysis is based on a unique database of all entries and exits of the banking sector in the Netherlands during the period 1850–1993 collected by the authors. We apply a hazard model to determine which factors explain the spatial formation of the Dutch banking industry. We examine the extent to which spinoff dynamics, location and time of entry had a significant effect on the survival rate of banks for a period of almost 150 years. Doing so, we make a first tentative step in providing an evolutionary explanation for why Amsterdam became the leading banking cluster of the Netherlands. The chapter is structured as follows. In the next section, we briefly outline the window of locational opportunity approach, and link that approach to two types of explanations for the spatial evolution of a new industry. In Section 3, we explain which data sources have been used to describe the spatial formation of the Dutch banking sector during the period 1850–1993, and we present some descriptive results. In Section 4, we briefly explain the estimation techniques employed. Then, we present the empirical findings. Finally, some conclusions are drawn.

2

SPATIAL FORMATION OF INDUSTRIES FROM AN EVOLUTIONARY PERSPECTIVE

In Section 2.1, we set out the main outlines of the window of locational opportunity (WLO) framework. This has been constructed in the early 1990s to provide an evolutionary explanation for the spatial formation of new industries. In Section 2.2, we will connect this WLO approach to the literature of industrial dynamics, that applies a population perspective

The banking cluster in Amsterdam, 1850–1993

193

to the industry life cycle (Klepper, 1997). This literature is interested in the mechanisms that drive the evolution of a population of firm-specific routines in an industry, and which make fitter routines more dominant in an industry (Nelson and Winter, 1982; Klepper, 2007). Following Arthur (1994), we distinguish between two mechanisms through which interorganizational learning (that is, the diffusion of ‘fitter’ routines from one firm to the other) may take place in space. The first is spinoff dynamics, in which the transfer of knowledge occurs between a parent firm and its spinoffs. The other one will be associated with agglomeration economies, in which knowledge spills over between firms that are geographically proximate. 2.1

Windows of Locational Opportunity

In the 1980s, it was quite common to follow an industry life cycle approach in economic geography (see for example, Markusen, 1985; Hall and Preston, 1988; Marshall, 1987; Scott, 1988). The ILC was used as a background to explain why new growth regions had emerged in the 1980s, like the Sunbelt states in the US (Norton, 1979). New industries developed in new regions, while old industries declined in regions that were once the main centres of economic growth. In order to explain why, the WLO framework was proposed by American scholars like Allen Scott, Michael Storper and Richard Walker (Scott and Storper, 1987; Storper and Walker, 1989). The WLO concept explains why it is unpredictable where new industries will emerge and develop in space. New industries require new types of knowledge, skills, inputs and institutions which existing organizations and institutions cannot provide, since these are orientated towards old technologies and routines (Boschma and Lambooy, 1999). Therefore, new industries will invest in their own research to develop new knowledge, their employees will acquire the necessary skills through on-the-job-learning, and they will accumulate their own capital because established capital suppliers cannot assess possible returns in markets that are unknown to them. As soon as a new industry reaches a critical mass in a region, growing demands will transform the region into a supportive local environment, with new institutions and organizations like research institutes, educational facilities, venture capitalists and specialized suppliers. The more firms in the new industry locate in a region, the more diversified the local labour market becomes, the more local suppliers can specialize, the higher the local demand, the better the infrastructure, and the more attractive the region becomes for newcomers, leading to even more local firms, and so on (Myrdal, 1957). This may set in motion a self-reinforcing and

194

Emerging clusters

path-dependent process, sustained by Marshallian agglomeration economies. In other words, new industries produce their own space. This is not to say that new industries can develop everywhere and the role of place is completely overlooked (Martin, 1999). Despite the newness, new industries will build on generic resources like knowledge and skills, that is, resources that are not yet specific to support the new industry, but may still favour their development (Boschma, 1997). This is especially relevant when new sectors grow out of related industries, like the automobile industry emerged out of cycle- and coach-making industries (Klepper and Simon, 2000). When some regions are well endowed with these generic resources, this may increase their probability to develop that new industry. Depending on how many regions have access to those generic resources, we can determine the extent to which the windows were open when a new industry emerges. Boschma (1997) has shown that this differs from industry to industry. So there are good reasons to go beyond a static and deterministic approach, and instead employ a dynamic and evolutionary perspective to explain fully the spatial clustering of an industry (Brenner, 2004). As set out above, agglomeration economies may come into being when a new industry concentrates in space, sustaining its further development. However, spinoff dynamics may be another driving force, which by itself may provide an alternative explanation. We will explain below why. 2.2

Spinoff Dynamics and Agglomeration Economies

The spatial formation of a new industry can be described in terms of spinoff dynamics (Arthur, 1994). With spinoffs we mean new firms founded by former employees of incumbent firms in the same industry. There is increasing evidence that spinoffs play a crucial role in the spatial concentration of industries (Dahl et al, 2003; Koster, 2006). In Arthur’s spinoff model, a new industry grows firm by firm, and each new entrant is a spinoff of an incumbent firm in the same sector. Since almost all spinoff firms locate in the same region as their parent company, the region that is lucky to generate many spinoffs at the early stage of the life cycle of the new industry will most likely dominate the industry. This is because the probability of giving birth to a new spinoff is dependent on the amount of firms that is already present in that region. In other words, the spatial emergence of an industry is a path-dependent process. Small events may be decisive, which are in this case the lucky occurrence of many spinoffs at a very early stage in a region, and not in other regions. These small events will be followed by positive feedbacks: the more spinoffs enter the region, the higher the probability of generating even more spinoffs.

The banking cluster in Amsterdam, 1850–1993

195

This spinoff model has been extended by Klepper (2007). Klepper views the spinoff process as a mechanism through which tacit knowledge is transferred or diffused from parent to offspring, and which positively affects the performance of spinoffs. Klepper claims that entrepreneurs with a techno-economic background in the same or related industries will perform better than entrants that lack that kind of experience. In addition, Klepper argues that success breeds success. Doing a study on the American car industry, he provided evidence that spinoff firms have a higher probability to survive when they originate from very successful parents. The spinoff models of Arthur and Klepper assume that the spinoff process is basically a local phenomenon. In other words, spinoff dynamics in itself may be a sufficient evolutionary explanation for the spatial concentration of an industry. This means we can explain the spatial clustering of a new industry without referring to location-specific features. But in order to know for sure, we have to test at least whether this is true or not. Besides spinoff dynamics, the spatial clustering of an industry may be affected by agglomeration economies. It is common to distinguish between urbanization and localization economies. Urbanization economies are externalities available to local firms irrespective of the industry they belong to. Localization economies arise from a spatial clustering of firms in the same sector. When taking an evolutionary perspective, we have to make the impact of agglomeration economies more dynamic. Following Myrdal (1957), we claim the higher the number of local firms, the stronger the impact of agglomeration economies becomes. Like in the spinoff model, we can then describe the spatial evolution of a new industry as a self-reinforcing and path-dependent process. Small events are associated with the fact that a region is just lucky to develop many new entrants at a very early stage of the life cycle of an industry. After having passed a critical threshold, increasing returns come into being, that is, the more entrants locate in the region, the stronger the impact of agglomeration economies becomes. Knowledge will accumulate and become increasingly available in a region through the formation of local networks as an industry grows. As a consequence, agglomeration economies can cause an industry to concentrate in a region. Accordingly, spinoff dynamics and agglomeration economies dynamics provide different evolutionary explanations for the spatial clustering of an industry. This is not to deny that the spinoff mechanism and agglomeration economies may play a role simultaneously (Boschma and Frenken, 2003). On the contrary, it seems plausible that a high rate of spinoff activity in a region will strengthen agglomeration forces, which, in turn, enhance spinoff creation and increase the survival rate of spinoff companies.

196

Emerging clusters

Doing a long-term empirical study on the British automobile industry, Boschma and Wenting (2007) found that both effects play a role, although in different stages of the life cycle of an industry. They found evidence that spinoff companies did not show a higher survival rate during the first stage of the industry life cycle, which comes as no surprise because at that stage there is still not much to be learnt from the parent, due to the lack of a dominant design. Contrary to spinoffs, however, new entrants that had working experience in related industries did very well at this stage. Only in a later stage of the industry life cycle did spinoffs perform better, because the pre-entry working experience in the same industry appeared to be of much higher value at that stage. The effect of the location on firms’ survival also differed between the life cycle stages. First, start-ups in the British car industry that were founded in regions with related industries during the first stage of the life cycle had a higher survival rate (see also Buenstorf and Klepper, 2009). Second, localization economies did not matter at that stage, as expected. While one would expect a positive effect from localization economies on firms’ survival at a later stage of the industry life cycle, Boschma and Wenting (2007) found the opposite result. The more spatially concentrated the automobile industry became, the harder it was to survive for a new entrant in a cluster, probably due to more intense local competition (see for similar results Staber, 2001; Otto and Kohler, 2008).

3

DATA AND DESCRIPTIVES

Since spinoff dynamics and agglomeration economies provide different explanations, the challenge for empirical research is to disentangle both mechanisms so as to assess their importance. This will be done for the Dutch banking sector in the period 1850–1993. The data we collected provide information on the year of entry, the year of exit, the location of the head office, and the pre-entry industrial background of the entrepreneur concerning every bank that entered the industry in the Netherlands during the period 1850–1993. First, we compiled a list of banks with the assistance of Nederlandse Financiële Instellingen in de Twintigste Eeuw: Balansreeksen en Naamlijst van Handelsbanken published by the Dutch Central Bank. This list also contains firms that entered the industry before 1850 and diversified after this date. This source has been used for the most part to compile a list of every bank that existed in the period 1850–1993. This source claims to have listed about every bank in the Netherlands over the span of this period, the years they were in business, their location and any reorganizations

The banking cluster in Amsterdam, 1850–1993

197

or ownership changes the banks underwent. To complete the list on time of entry and year of exit, two other sources were particularly useful: the Nationale Vereniging van Banken, and the Nederlandsch EconomischHistorisch Archief. To determine the pre-entry industrial background of the entrepreneurs, a multitude of sources has been used. The most important were Geschiedenis van de Algemene Banken in Nederland 1860–1914 (Kymmell, 1992, 1996) and Geschiedenis van de Nederlandsche Bank (De Jong, 1967). Further information on pre-entry background was acquired from trade journals, regional and city archives, incorporation records and various other historical sources.1 The online databank on Dutch entrepreneurs of the Internationaal Instituut voor Sociale Geschiedenis turned out to be very useful for finding relevant information on the backgrounds of the entrepreneurs. Eventually all these sources have led to a database that contains up to 906 banks, of which both time of entry and year of exit are known. The location where the banks have been founded is known for all of them. We could trace the pre-entry economic background of the entrepreneur for 756 banks, which is an extremely high number for this kind of studies. This enables us to analyse 756 banks for which information is available concerning their time of entry, year of exit, the location of their head-office, and their entrepreneurial background over the period 1850–1993. Since our data sources contain data until 1993, we take that as the last year in our analysis. Our starting year is 1850. Contrary to manufacturing sectors, it is often quite complicated to demarcate the date of birth of a service industry (Fein, 1998). This also applies to the banking sector which, of course, already existed before 1850. What is interesting, though, is that just after 1850 the Dutch banking sector underwent a total rebirth. Before 1860, a real banking sector in the Netherlands did not exist, although there was a money and stock market. Following the example of the Credit Mobilier in France, a structural change in the Dutch banking sector occurred between the years 1861 and 1865: the first banks in the Netherlands emerged with a juridical structure of a limited liability company. These banks had as their main activity the supplying of credit for their own account and taking money in deposit and current accounts. They had large sums of starting capital, which was new to the Dutch banking system.2 This new type of banks became the dominant design in that industry in the following decades. This was encouraged by two developments. In the first place, there was a rapidly increasing need for capital to invest in the Dutch colonies and new infrastructure like the railways and canals. In the second place, a new banking law came into effect in 1863 which

198

Emerging clusters

obliged the Dutch Central Bank, beside its head office in Amsterdam, to start up a network of offices across the country (Kymmell, 1992). The purpose was to make the credit facilities of the Dutch Central Bank more directly available at a regional scale. The establishment of the office network in 1865 by opening an office in Rotterdam and 12 agencies in the provinces can be seen as the beginning of the banking industry in the Netherlands. We have data on the year of entry for 906 banks that entered the Dutch banking sector in the period 1850–1993.3 For 112 banks the year of entry is unknown. Of these 906 banks, 779 banks had to exit the banking sector for some reason in the period 1850–1993, 119 banks survived that period, and for eight banks the year of exit is unknown.4 Exits in the banking sector deserve special attention. This may be caused by two main reasons. Banks may close their doors for reasons like bankruptcy, closure, diversification into other activities than banking, and so on.5 Out of 779 exits, 394 exits (51 per cent) could be assigned to that category. Banks may also exit due to merger and acquisition activity. We counted a number of 385 exits (49 per cent). On the one hand, these concern cases in which we were sure who took over whom. The approximately 320 banks that were taken over by other banks have been considered exits. On the other hand, these concern merger and acquisition activity, in which we could not identify who acquired whom. In that case, we simply counted each bank that was involved in the merger activity as exit. These concerned about 70 exits.6 In Figure 8.1, we have depicted the number of entrants and exits in the Dutch banking sector for the period 1850–1993. As the figure shows, entry levels remained relatively low until the 1890s. One of the reasons was that it was considered a sign of weakness to lend money from a bank. In the 1890s, this resistance for credits from banks started to disappear and relations with a bank were considered normal among a growing number of entrepreneurs (Nierop, 1972). As Figure 8.1 shows, since the 1890s, there was a sharp and steady increase in the number of entrants (with a peak of 25 in 1924), which remained high until 1930. In October 1929, the Great Depression caused entry to drop sharply. After 1934, there were more than five entrants for only two years.7 Since the 1970s, entrants have been mainly foreign banks that established an office in the Netherlands. As shown in Figure 8.1, the number of exits was extremely low in the period 1850–1900. Before 1900, most of the large banks never had deficits, despite two crises in 1880 and 1884. Only 17 banks that had entered in the period 1850–1900 closed their doors in that same period. After the turn of the century, the number of exits started to increase, especially in the 1920s and early 1930s. Starting in 1911, the Rotterdamsche Bank employed

199

1985

1976

1967

1958

1949

1940

1931

1922

1913

1904

1895

1886

1877

1868

1859

45 40 35 30 25 20 15 10 5 0 1850

Number of entries and exits

The banking cluster in Amsterdam, 1850–1993

Time (years) Number entries

Figure 8.1

Number exits

The number of entrants and exits in the Dutch banking sector, 1850–1993

an aggressive acquisition strategy. Other banks soon followed. The Algemeene Nederlandsche Centrale Middenstands-Creditbank took over most of the tradesmen’s banks in the country (Bosman, 1989). This merger and acquisition activity reached its peak in the years 1916–19, and led to the rise of five large banks that would dominate the industry for the next decades: the Nederlandsche-Handelmaatschappij (1824), the Twentsche Bank (1861), the Rotterdamsche Bank (1863), the Amsterdamsche Bank (1871) and the Incasso Bank (1891) (Kymmell, 1996). It heralded the beginning of the concentration in the Dutch banking industry. Major process innovations (like the development of the telephone line network after 1900, the Giro transfer system in 1914, or the possibility of telegraphic transfers in 1890) occurred in the banking sector in the early twentieth century, and these required a minimal scale of operation. Smaller banks and late entrants had higher average costs, and therefore often became easy targets for takeovers (Kymmell, 1996). In the first half of the 1920s, the economic tide changed again: numerous small, mostly regional banks could not overcome the bank crisis on their own and were acquired by one of the larger banks. At the end of the 1920s, the merger and acquisition activity reached a new peak. 1930 was the first year in which the total number of banks started to decline. The underlying reason for this turnaround was the Great Depression (Van der Lugt, 1999). After 1931, the merger and acquisition activity slowed down for two decades, and most exiting banks were liquidated or stopped offering services. In the 1950s and 1960s, the level of exits went down again. In 1948, the

200

Emerging clusters

600 Number of firms

500 400 300 200 100

18 50 18 60 18 70 18 80 18 90 19 00 19 10 19 20 19 30 19 40 19 50 19 60 19 70 19 80 19 90

0

Time (years)

Figure 8.2

The number of firms in the Dutch banking sector, 1850–1993

Incasso-bank was acquired by the Amsterdamsche Bank. Another large bank, Twentsche Bank, started to take over some of the larger banks in the provinces. In 1958, Pierson & Co’s and Heldring & Pierson merged into the new bank Pierson, Heldring & Pierson. In 1964, only two large banks were left after the merger of the Rotterdamsche Bank with Amsterdamsche Bank into the Amsterdam-Rotterdam Bank (AMRO Bank), and the merger of the Nederlandsche Handel-Maatschappij with Twentsche Bank into the Algemene Bank Nederland (ABN). A lot of smaller banks joined a large bank. Two new banks emerged as major players through their merger activities. The Rabobank was the result of the merger of the cooperative rural central banks, the Raiffeisen-Bank and the Eindhovense Boerenleenbank in 1972. In the 1970s there was some increase, but only for a short period, after which the number of exits stabilized at a very low level. In 1986 the Postbank was the result of the merger of the Rijkspostspaarbank and the Postcheque- en Girodienst, and in 1989 the Postbank merged with the Nederlandsche Middenstandsbank into the NMB Postbank, which became later ING Postbank. Finally, in 1991 the ABN and the AMRO Bank merged into the ABN-AMRO Bank. This meant that the big five that existed in the early twentieth century had come together in one corporation (Van der Lugt, 1999). In Figure 8.2, we show the evolution of the number of banks over the period 1850–1993, which is derived from the number of entries and exits. The evolution of the Dutch banking sector shows a very clear pattern. Besides a small dip caused by the First World War,8 the total number of banks kept increasing until 1929 when it reached the maximum of 478

201

1985

1976

1967

1958

1949

1940

1931

1922

1913

1904

1895

1886

1877

1868

1859

180 160 140 120 100 80 60 40 20 0 1850

Number of banks

The banking cluster in Amsterdam, 1850–1993

Time (years)

Figure 8.3

The number of banks in the Amsterdam region in the period, 1850–1993

banks. From the turn of the century onwards, the industry was dominated by five large banks, of which four were based in Amsterdam.9 In 1900, this group of five banks had a market share of 35 per cent. The following years, the market share began to rise: in 1918 it was already 48 per cent. In the 1920s, the market share became slightly less, because of a lot of new banks entering the industry.10 In 1930, the number of exits overtook the number of entrants and the shakeout of the industry started. In 1940, the market share of the big five was up to 52 per cent (Kymmell, 1996). The declining trend in the number of firms decelerated in the 1970s. In 1991 there were only 119 banks left, still a considerable number. By that time, the Dutch banking sector had evolved into an oligopoly dominated by three large banks (ABN-AMRO, ING Group and Rabobank), which had a national market share of roughly 80 per cent (Bos, 2004). How about geography, and how about the spatial clustering of the banking sector in the Amsterdam region? For almost all banks over the period 1850–1993 we have been capable of determining their location (municipality), and we have allocated these to the 40 labour markets (COROP regions) of the Netherlands. Our analysis concentrates on headquarters of banks, being a knowledge-intensive service activity. Our data could not account for the development of branch offices that took off in the early twentieth century (Bosman, 1989). In Figures 8.3 and 8.4, we have depicted the number of banks in the Amsterdam region and its share in the national total in the period 1850–1993. The Amsterdam region has been defined as COROP Groot-Amsterdam. This region consists of

Emerging clusters

60,0 50,0 40,0 30,0 20,0 10,0 1985

1976

1967

1958

1949

1940

1931

1922

1913

1904

1895

1886

1877

1868

1859

0,0 1850

Share in number of banks (%)

202

Time (years)

Figure 8.4

The share of the Amsterdam region, as a percentage of the total number of banks in the Netherlands in the period 1850–1993

more than the city of Amsterdam alone: it also contains municipalities like Aalsmeer, Amstelveen, Diemen, Edam-Volendam, Haarlemmermeer, Purmerend and Uithoorn. One should remember that Figures 8.3 and 8.4 only include those banks that entered after 1850. As shown in Figure 8.3, the number of firms in the Amsterdam region has steadily increased since the 1850s. Amsterdam experienced a steep increase in mainly the 1920s, reaching its peak in 1930. After that, a decline set in until the late 1950s, after which the number of banks stabilized at a level of about 60–80 banks until 1993. In relative terms, the share of the Amsterdam region in the total number of banks increased sharply in the 1850s, up to 38 per cent in the early 1860s. Then, it started to drop until 1915 when it was a mere 19.5 per cent. This was not so much caused by exits of banks located in Amsterdam, but due to a relative increase of the shares of the Rotterdam region and the Hague region. This changed after 1915 when more and more banks located in Amsterdam. In only 15 years, 116 banks entered the sector in the Amsterdam region. As a result, the percentage of Amsterdam banks rose again, to almost 35 per cent in 1930. This share stayed more or less the same for almost 40 years, until foreign banks started to enter the Netherlands. In combination with exits that occurred mainly outside the Amsterdam region, Amsterdam increased its share even more: in 1991, around 53 per cent of all banks in the Netherlands were located in the Amsterdam region. In terms of market share, the concentration of bank activity around Amsterdam was much higher than that (Sluyterman et al., 1998).

The banking cluster in Amsterdam, 1850–1993

4

203

SURVIVAL ANALYSIS

Following others, we will employ a hazard model in order to determine which factors can explain the spatial evolution of the Dutch banking sector (Klein and Moeschberg, 1997). More in particular, we will estimate Cox regressions to assess the effects of location (being located in the Amsterdam cluster), time of entry, and pre-entry background of the entrepreneur (spinoffs, experienced and inexperienced firms) on the survival rates of banks. Doing so, we will assess whether being located in the Amsterdam region actually benefits the performance of banks, as assumed by large parts of the cluster literature. The dependent variable is the age of each bank, as is common in survival analysis. Of course, we would have preferred better economic indicators, like sales, turnover or market shares, but these data are not available over such a long period. We measure the age of each bank by counting the number of years between the first (entry) and last year of commercial activity (exit). As explained previously, we know the year of entry for 906 banks: we left out 112 banks for which the year of entry is unknown. For eight banks out of these 906, the year of exit is unknown. In the case of mergers and acquisitions, we treat banks as exits when they are taken over by another bank. In about 70 cases, we could not identify who was taken over by whom. In that case, all banks concerned were considered exits, and the newly formed bank was treated as a new entrant. Banks that continued to exist after 1993 were treated as censored cases, as is common in Cox regressions. The first set of independent variables concerns the location of banks. For all banks, we have information on their location (municipality). In the very exceptional case that a bank had moved from one region to another, we assigned the bank to the region where it had been active for most of its time. In order to test whether the location in the Amsterdam region affects the survival of banks positively, we constructed a dummy variable. We explained earlier how we defined the Amsterdam region. The second set of independent variables concerns the pre-entry background of entrepreneurs. Above, we set out that an entrepreneurial background of entrants is essential for survival, because they refer to the capabilities and routines of the parent banks that are taken to the new bank. For the 906 entrants of which the year of entry is known, we could find information on the pre-entry background of the founder for 736 banks. Unfortunately, for 170 banks, no information on the entrepreneurial background could be found. This group of entrants with an unknown entrepreneurial background is relatively small, in comparison to other survival studies. This group of entrants has been excluded from the analyses.11 Following Klepper, we distinguish between three types of banks.

204

Emerging clusters

Spinoffs have been defined as entrants that are founded by former employees working in the same banking sector. Employees of existing banks are assumed to learn from their experience, which they can exploit when they start up their own bank. New entrants were classified as spinoffs if at least one of the founders had worked for and/or had founded a bank. When a founder had worked for several firms in the past, the last firm worked for was considered the parent of the spinoff. Some spinoffs had multiple founders that had worked for different firms. If this was the case, than the parent of the spinoff was determined based on the founder that was described as the most influential in the new spinoff company. Of all 906 entrants that entered the banking sector in the period 1850–1993, 191 firms were defined as spinoff entrants. We have defined entrants as experienced firms when they had prior experience in related activities before entering the banking industry. As explained before, we expect that entrants with backgrounds in one of these related occupations to have better capabilities than entrants that lack such experience. At the beginning of the banking industry, these related activities concerned cashiers, bankers and stock-brokers (Kymmell, 1992, pp. 73–95). Cash could be obtained from a cashier, banker or stock-broker through the selling of not-due claims of bills of exchange and/or promissory notes or through making a loan with securities or personal properties. None of these three occupations had as their main activity the providing of credit. Cashiers were mainly occupied with collecting, keeping and paying money to their clients and with the brokerage in bills of exchange, thus with payments. Also they bought bills of exchange as temporary interestpaying investments and they gave loans. Stockbrokers traded in shares in principle on behalf of their clients and sometimes they gave credit for the purchasing of stock against the security of stock. Bankers did the same, but besides that they were mainly active in the issuing of stock and the international bill-broking (Kymmell, 1992, pp. 14–15). In the nineteenth century, these occupations were practised by one-man firms or sometimes together with one or more partners, who were most of the time family members. We counted a total of 316 experienced firms that entered the banking sector in the period 1850–1993. On the one hand, there were entrants that remained active in those related activities while diversifying into the banking sector.12 There were a total number of so-called 75 diversifiers. On the other hand, there were 241 entrants that set up de novo banks founded by heads of firms in related activities (Klepper, 2007). Banks were classified into this category when at least one of their founders was identified as the head of a firm that was active or had recently been sold.13 The third type of entrants contained 229 inexperienced entrepreneurs. These entrants had no prior experience in the banking sector and related industries.

The banking cluster in Amsterdam, 1850–1993

Table 8.1

205

Estimation results of the Cox regressions (standard errors in parentheses) Model 1

Amsterdam

−0.159* (0.087)

Cohort 1 Cohort 2

Model 2 −0.132 (0.088) −0.900*** (0.153) 0.154 (0.135)

Spinoffs Experienced Chi-square -2LogLikelihood

3.303 7125.736

124.185 7000.788

Model 3 −0.009 (0.094) −1.091*** (0.156) −0.036 (0.138) −0.651*** (0.116) −0.895*** (0.095) 217.899 6915.407

Notes: * significant at the 0.10 level, ** significant at the 0.05 level, *** significant at the 0.01 level. n = 733.

The third set of independent variables involves time of entry. According to Klepper (2007), early entrants in a new industry will outperform late entrants because of weaker selection and lower entry barriers. Consequently, we expect early entrants to have a lower hazard rate at every age.14 All entrants for which we have information on the pre-entry entrepreneurial background (736 banks) and their years of entry and exit (733 banks) have been assigned to three cohorts. The first cohort concerns the period 1850-1900, a period in which small size was a key feature, to which 220 firms have been assigned. The second cohort concerns 377 entrants that emerged in the expansion period 1901–1929 during which scale economies grew in importance and entry barriers increased. The third cohort includes 136 firms that entered the Dutch banking sector in the period 1930–1993. We have estimated Cox regressions for a total of 733 banks on which we had information on their year of entry and exit, the pre-entry working experience of the entrepreneur, and their location. The estimates are based on maximum likelihood, adding more variables in each model. The main findings are presented in Table 8.1 In the first model, we estimate the effect of the Amsterdam region on the hazard rates of banks. This has been done in its most simple manner, including a dummy variable, measuring all possible effects of the Amsterdam region. A major finding of model 1 is that being located in the

206

Emerging clusters

Amsterdam region does indeed matter. In line with the cluster literature, the location of the Amsterdam region lowers the hazard rate of banks, and thus increases their survival rate.15 However, if we include other variables in the model, this cluster effect tends to disappear. In model 2, we have included the time of entry variables. We have made two dummy variables, one for cohort 1 (including all entrants in the period 1850–1900), and one for cohort 2 (entrants in period 1901–1929), with cohort 3 as the omitted reference group. As model 2 shows, the coefficient estimates of cohort 1 is negative and significant. As expected, earlier entrants show indeed a lower hazard rate. This is, however, not true for cohort 2. This crucial effect of early entry has been confirmed by other studies (Cantner et al., 2006; Klepper, 2007). In model 3, we have added the possible effects of the pre-entry background of the entrepreneur. We defined two dummies that are equal to 1 for spinoffs and experienced firms. The group of inexperienced firms has been treated as the reference group. As expected, being a spinoff company and being an experienced firm have very strong negative impacts on the hazard rate of banks. This is in line with the evolutionary argument that new firms inherit routines from their parents. If the founder has acquired experience in the same or related industries, it will increase the performance of the new entrants: the more close this pre-entry working experience is to the sector, the better the new firms will perform. The banking sector is no exception to that rule. Thus, we observe clustering of banking in the Amsterdam region (housing more than half of the Dutch banks), but the location of Amsterdam itself did not appear to have a significant impact on the survival of the banking firms. Rather, early entry and the industrial background of the entrepreneurs mattered. These preliminary findings suggest that Amsterdam was just lucky to have many start-ups with pre-entry experience in the banking and related sectors, and that is why the Amsterdam region did well. In fact, the Amsterdam region had a disproportionate number of spinoffs in the banking sector: about 54 per cent of the spinoff entrants in the Netherlands located in the Amsterdam region, which is much higher than its share of 32 per cent in the total number of entrants.16 However, when we investigate in more detail the effect of the Amsterdam cluster on the survival of different types of banks, location does seems to matter. In Figure 8.5, we have drawn the survival curves of spinoffs in the Amsterdam region and spinoffs outside that region. Survival curves indicate the percentage of firms that survive at each age. The vertical axis shows this percentage plotted on a logarithmic scale. This figure shows that up to about the age of 35, the survival curves of spinoffs in the Amsterdam region and outside are almost identical. After that, spinoffs

207

119

112

105

98

91

84

77

70

63

56

49

42

35

28

21

14

7

5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0 0

Survival rate (LN%)

The banking cluster in Amsterdam, 1850–1993

Age (years) spinoffs Amsterdam region

Figure 8.5

spinoffs outside Amsterdam region

Survival curves by spinoffs inside and outside the Amsterdam region

in the Amsterdam region outperform the spinoffs outside the Amsterdam region, suggesting a positive impact of the Amsterdam cluster, in addition to the effect of being a spinoff. Apparently, spinoff companies do well in their first years of existence irrespective of their location, but they benefit from being part of the Amsterdam cluster only at later ages. In Figure 8.6, we see a similar pattern for experienced firms, although the difference between experienced firms inside and outside the Amsterdam region is less pronounced than for spinoff companies. The only group that was underrepresented in the Amsterdam cluster was the inexperienced entrepreneurs. This latter group also showed lower survival rates at higher ages in Amsterdam than inexperienced firms located elsewhere. This outcome suggests that no positive spillover effects occurred from successful to inexperienced banks in the Amsterdam region. On the contrary, inexperienced firms suffered disproportionally from being located in the Amsterdam region, probably due to higher relative location costs.

5

CONCLUSIONS

Our study suggests it is mainly through spinoff dynamics that Amsterdam became the leading banking cluster in the Netherlands. Broadly speaking, although the banking sector concentrated in the Amsterdam region, and Amsterdam was home to the Central Bank and the national stock

Emerging clusters

128

120

112

104

96

88

80

72

64

56

48

40

32

24

16

8

5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0 0

Survival rate (LN %)

208

Age (years) experienced firms Amsterdam region experienced firms outside Amsterdam

Figure 8.6

Survival curves by experienced firms in and outside Amsterdam region

exchange, being located in the Amsterdam region did not increase the survival of banks in the Amsterdam region in general. This suggests the windows of locational opportunity were quite open in the mid-nineteenth century, although our findings also suggest this was most likely restricted to the main urban field in the Netherlands. After some time, the windows of locational opportunity closed, not because of local externalities in the Amsterdam region, but because the spinoff process made Amsterdam the leading banking region. This result questions the cluster literature that almost takes for granted the positive effects of spatial clustering. Our study shows how important it is to control for other firm-specific features when assessing the economic effects of clusters. However, having said that, spinoff firms and experienced firms did benefit from being located in the Amsterdam cluster, but only at later ages. Overall, these preliminary results concerning a knowledge-intensive service sector are pretty much in line with earlier findings in studies on manufacturing industries. There are, of course, many questions for future research to be taken up. We briefly mention some of these. In order to give a full account of the location effects in these studies, it is absolutely necessary to include other location variables than dummy variables. One could think of a range of externalities indicators as in Boschma and Wenting (2007), such as Jacobs’ externalities, localization economies and related variety externalities, but one should also account for regional growth in general. It goes without

The banking cluster in Amsterdam, 1850–1993

209

saying that this requires high-quality longitudinal data at the regional level. This would also enable analysing these effects along the various stages of the industry life cycle. Another challenge is to investigate in more depth the effects of exits due to mergers and acquisitions. In the Dutch banking sector, about 50 per cent of all exits were caused by mergers and acquisitions over the whole period, which is extremely high in comparison to other industries, like the automobile sector for instance. This non-trivial feature of the history of the banking sector calls for more in-depth research. One issue is whether exits caused by mergers and acquisitions can be directly associated with failure (Fein, 1998). An even more crucial issue is that merger and acquisition activity might provide an alternative explanation for the spatial clustering of an industry. It cannot be excluded that some banks in the Amsterdam region did so well because they were very active in taking over banks in other regions. If so, this would lead to a higher survival rate of the Amsterdam banks, because the acquiring banks in Amsterdam continued to exist, while the acquired banks in other regions were treated as exits. But even more important, it is plausible that Amsterdam banks that were very active in merger and acquisition activity became much smarter over time, which might have increased their survival even more. This is not only because these Amsterdam banks acquired crucial experience in taking over other banks and how to handle that complex process, but also because they got access to the knowledge of the acquired banks. This is a field of research on spatial clustering of industries that still needs to be exploited.

NOTES 1. 2.

3.

4.

5.

Quite a lot of banks have published on their own history. Some of those publications have been used (see reference list). There was only one large bank before 1863 that had given credit as their main activity and that was the Dutch Central Bank (De Nederlandsche Bank), which had been founded in 1814 as a limited liability company by King Willem I. The only shareholder was the Dutch state (de Jong, 1967). We have also included in our database De Nederlandsche Bank and Algemeene Nederlandsche Maatschappij ter Bevordering van den Volksvlijt due to their prominent role in the Dutch banking sector, despite the fact that they entered the Dutch market before 1850 (1814 and 1822 respectively). We did not include 56 banks that had entered before 1850, despite the fact that all of these 56 banks (except for four banks) exited the sector in the period 1850–1993. We also did not include in our analysis 88 banks for which no entry data and no exit data are available, and another 24 banks for which data are available with respect to their year of exit, but their year of entry is unknown. In the database, these concerned banks that were considered exits due to ‘opheffing, failissement, liquidatie, en niet meer uitvoeren van bankactiviteiten’ (‘closure, bankruptcy, and not employing bank activities any more’).

210 6.

7. 8. 9. 10. 11. 12.

13.

14. 15. 16.

Emerging clusters In the database, these concerned mergers that were referred to as ‘gefuseerd met, opgegaan in, samengegaan met, en voortzetting van’ (‘merged with, and continuation of’). For these exits, we could not identify who was the acquiring firm, and which firm was acquired. So, we could not determine which of the firms involved should be considered exits (the acquired firm), and which one the acquiring firm (the firm that survived). One of those years was 1941. In this year, the Nazis influenced the entry total by starting three financial institutions (Van der Lugt, 1999). The other year was 1974 when six foreign banks entered the industry in the Netherlands. In 1917, the highest number of exits of all years could be observed: 41 banks, of which 38 were acquired by other banks. ‘Nederlandsche Handel-Maatschappij’ (Amsterdam) ‘Rotterdamsche bank’ (Rotterdam) ‘Amsterdamsche bank’ (Amsterdam) ‘Twentsche bankvereeniging’ (Amsterdam) ‘Incasso-bank’ (Amsterdam) The market share of the five large banks reached its low in 1928 with 38 per cent A lot of entrants that are part of this group had a shorter life span than the entrants with a known background. This makes sense, because for banks that existed for only a few years, some of them even one year, less information is available. In 1861 B.W. Blijdenstein Jr. founded the Twentsche Bankvereeniging in Amsterdam. After finishing Law school, Blijdenstein started a notary’s office in his home town Enschede. Because of his appointment as a curator in the bankruptcy of a small local cashier in 1841, he had insight for the first time into the cashier business. In 1844, two thirds of the profits came from the cashier business. At the end of the 1850s, almost every textile manufacturer in the region around Enschede was a client of Blijdenstein’s cashier’s office. When the Twentsche Bankvereeniging was established in 1861, the cashier’s office was continued. Therefore, we consider this entrance as a diversifier into the banking sector (de Graaf et al., 1996). In the first decades, a few prominent bankers were involved with the founding of several banks. These prominent bankers in most cases had a lot to say, or even made important decisions about how business was done (Kymmell, 1996). Therefore, these banks were seen as having experienced entrepreneurs. As explained above, we had data on the pre-entry entrepreneurial background for 736 banks. For three of these 736 banks, data on the year of exit were missing. We also found that a location in the major urban area of the Netherlands (defined as the regions of Amsterdam, Utrecht, Rotterdam and the Hague) did positively impact on the survival of banks. This suggests an urbanization economies effect. The share of Amsterdam in experienced firms is just above that average (35 per cent).

REFERENCES Arthur, W.B. (1994), Increasing Returns and Path Dependence in the Economy, Ann Arbor, MI: The University of Michigan Press. Audretsch, D.B. and M.P. Feldman (1996), Innovative clusters and the industry life cycle, Review of Industrial Organization, 11, 253–73. Bos, J.W.B. (2004), Does market power affect performance in the Dutch banking market? A comparison of reduced form market structure models, De Economist, 152 (4), 491–512. Boschma, R.A. (1997), New industries and windows of locational opportunity. A long-term analysis of Belgium, Erdkunde, 51, 12–22. Boschma, R.A. and K. Frenken (2003), Evolutionary economics and industry location, International Review for Regional Research, 23, 183–200.

The banking cluster in Amsterdam, 1850–1993

211

Boschma, R.A. and J.G. Lambooy (1999), Evolutionary economics and economic geography, Journal of Evolutionary Economics, 9, 411–29. Boschma, R.A. and R. Wenting (2007), The spatial evolution of the British automobile industry: Does location matter? Industrial and Corporate Change, 16 (2), 213–38. Bosman, H.W.J. (1989), Het Nederlandse Bankwezen. Serie bank- en effectenbedrijf nr. 1. Amsterdam: Nederlands Instituut voor het Bank- en Effectenbedrijf. Brenner, T. (2004), Local Industrial Clusters. Existence, Emergence and Evolution, London and New York: Routledge. Buenstorf, G. and S. Klepper (2009), Heritage and agglomeration. The Akron tyre cluster revisited, The Economic Journal, 119 (537), 705–33. Cantner, U., K. Dressler and J.J. Krueger (2006), Firm survival in the German automobile industry, Empirica, 33, 49–60. Carree, M.A. (2003), A hazard rate analysis of Russian commercial banks in the period 1994–1997, Economic Systems, 27, 255–69. Dahl, M.S., C.R. Pedersen and B. Dalum (2003), Entry by spinoff in a high-tech cluster, DRUID Working Paper, 3–11. de Graaf, T., A.J. Mensema and J.J. Mobron (1996), Van Katoentjes tot Lening: Geschiedenis van de Twentsche Bank 1861–1964, Amsterdam: ABN AMRO Historisch Archief. de Jong, A.M. (1967), Geschiedenis van de Nederlandsche Bank: van 1814 tot 1964. Parts I–V, Haarlem: Joh. Enschede and Zonen. Fein, A.J. (1998), Understanding evolutionary processes in non-manufacturing industries: Empirical insights from the shake-out in pharmaceutical wholesaling, Journal of Evolutionary Economics, 8 (3), 231–70. Feldman, M.P. and J. Francis (2004), Homegrown solutions. Fostering cluster formation, Economic Development Quarterly, 18, 127–37. Feldman, M.P. and Y. Schreuder (1996), Initial advantage. The origins of the geographic concentration of the pharmaceutical industry in the Mid-Atlantic region, Industrial and Corporate Change, 5, 839–62. Feldman, M.P., J. Francis and J. Bercovitz (2005), Creating a cluster while building a firm. Entrepreneurs and the formation of industrial clusters, Regional Studies, 39, 129–41. Grote, M.H. (2007), Foreign banks’ attraction to the financial centre Frankfurt. A ‘U’-shaped relationship, Working Paper, no. 177, Frankfurt am Main: Goethe Universitat. Hall, P.G. and P. Preston (1988), The Carrier Wave. New Information Technology and the Geography of Innovation 1846–2003, London: Unwin Hyman. Helfat, C.E. and M.B. Lieberman (2002), The birth of capabilities. Market entry and the importance of pre-history, Industrial and Corporate Change, 11 (4), 725–60. Iammarino, S. and P. McCann (2005), The structure and evolution of industrial clusters. Transactions, technology and knowledge spillovers, Paper presented at ERSA, Amsterdam, August. Klein, J.P. and M.L. Moeschberg (1997), Survival Analysis: Techniques for Censored and Truncated Data, New York: Springer-Verlag. Klepper, S. (1997), Industry life-cycles, Industrial and Corporate Change, 6 (1), 145–82. Klepper, S. (2007), Disagreements, spinoffs and the evolution of Detroit as the capital of the U.S. automobile industry, Management Science, 53 (4), 616–31.

212

Emerging clusters

Klepper, S. and K.L. Simon (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. Koster, S. (2006), Whose child? How existing firms foster new firm formation: individual start-ups, spin-outs and spin-offs, dissertation, University of Groningen, Faculty of Spatial Sciences: Groningen. Kymmell, J. (1992), Geschiedenis van de Algemene Banken in Nederland 1860–1914. Part I. Amsterdam: Nederlands Instituut voor het Bank- en Effectenbedrijf Kymmell, J. (1996), Geschiedenis van de Algemene Banken in Nederland 1860–1914. Part II. Amsterdam: Nederlands Instituut voor het Bank- en Effectenbedrijf. Maggioni, M.A. (2002), Clustering Dynamics and the Location of High-Tech-Firms, Heidelberg: Springer Verlag. Markusen, A. (1985), Profit Cycles, Oligopoly and Regional Development, Cambridge, MA: MIT Press. Marshall, M. (1987), Long Waves of Regional Development, London: Macmillan. Martin, R. (1999), The new ‘geographical turn’ in economics: some critical reflections. Cambridge Journal of Economics, 23 (1), 65–91. Menzel, M.-P. and D. Fornahl (2007), Cluster life cycles: Dimensions and rationales of cluster development, Jena Economic Research Papers 2007-076. Myrdal, G. (1957), Economic Theory and Underdeveloped Regions, London: Duckworth. Nelson, R.R. and S.G. Winter (1982), An Evolutionary Theory of Economic Change, Cambridge MA: Harvard University Press. Nierop, H.A. (1972), Schets van het Bankwezen, Haarlem: De Erven F. Bohn N.V. Norton, R.D. (1979), City Life-Cycles and American Urban Policy, New York: Academic Press. Otto, A. and S. Kohler (2008), The contribution of new and young firms to the economic development of clusters in Germany. Comparative analysis of a growing, a mature and a declining cluster, in U. Blien and G. Maier (eds), The Economics of Regional Clusters. Networks, Technology and Policy, Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 171–89. Scott, A.J. (1988), New Industrial Spaces. Flexible Production Organization and Regional Development in North America and Western Europe, London: Pion. Scott, A.J. and M. Storper (1987), High technology industry and regional development. A theoretical critique and reconstruction, International Social Science Journal, 112, 215–32. Sluyterman, K., J. Danker, J. Van der Linden and J. Luiten van Zanden (1998), Het Coöperatieve Alternatief: Honderd Jaar Rabobank 1889–1998, Den Haag: Sdu Uitgever. Staber, U. (2001), Spatial proximity and firm survival in a declining industrial district. The case of the knitwear firms in Baden-Wurttemberg, Regional Studies, 35, 329–41. Storper, M. and R. Walker (1989), The Capitalist Imperative. Territory, Technology and Industrial Growth, New York: Basil Blackwell. Ter Wal, A.L.J. and R.A. Boschma (2009), DOI: 10.1080/00343400802662658. Ter Wal, A.L.J. and R.A. Boschma (2009), Co-evolution of firms, industries and networks in space, Regional Studies, in press.

The banking cluster in Amsterdam, 1850–1993

213

Van der Lugt, J.A. (1999), Het Commerciële Bankwezen in Nederland in de Twintigste Eeuw, Een Historiografisch Overzicht, NEHA Jaarboek. Visser, E.J. and R.A. Boschma (2004), Learning in districts. Novelty and lock-in in a regional context, European Planning Studies, 12 (6), 793–808. Wenting, R. (2008), Spinoff dynamics and the spatial formation of the fashion design industry, 1858–2005, Journal of Economic Geography, 8, 593–614.

9.

The role of the university in the genesis and evolution of researchbased clusters Donald Patton and Martin Kenney*

The modern research university, by its commitment to research and the advancement of science and technology, continuously produces inventions, as well as the occasional technological breakthrough, that provide the type of opportunities that allow entrepreneurs to create new firms. The discovery of these opportunities, and assembling the resources to exploit them, has been described as the entrepreneurial event (Feldman 2001) or as the act of entrepreneurial discovery (Kirzner 1997), and the promotion of these entrepreneurial opportunities is now seen by some observers as a major responsibility of the university. Histories of cluster development reveal that cluster emergence (Braunerhjelm and Feldman 2006) is an evolutionary process that requires an initial seeding or triggering event, followed by entrepreneurial activity that builds upon this event. This initial triggering event is frequently the result of serendipity, and can in no way be planned or anticipated.1 A triggering event can also be part of a planned government initiative to stimulate cluster development. In the cases discussed in this chapter these triggering events are the discoveries that emerge from research conducted by the university. In general the first stage of cluster evolution requires a seeding event which produces an economic opportunity, the presence of entrepreneurs who have the knowledge to discover this opportunity and are in a position to act upon it, and the existence of resources for new firm formation in the cluster that are available to the entrepreneur. Because universities continuously produce potential entrepreneurial opportunities for new firm formation, it is instructive to observe their success in promoting cluster development. In particular, studying university-based clusters over time allows for a comparative examination of the first stage of cluster development. Although university-based clusters exist in other countries, these clusters were first noticed in the United States in the postwar period and are most common in American settings.2 As Scott Shane (2004) observed in 214

The role of the university in research-based clusters

215

his extensive review of university start-ups, World War II transformed American research universities, particularly with respect to federal government funding of research. Throughout the last half of the twentieth century and into the twenty-first, real university R&D expenditures increased significantly both absolutely and as a percentage of total US R&D. The Bayh–Dole Act of 1980 giving universities (and other Federal contractors) the exclusive property rights to inventions certified and generalized a process of commercialization that had already been underway by that time (Mowery et al. 2004).3 This increase in entrepreneurial activity at US universities has been mirrored by an increased academic interest in the topic. A recent literature search of this topic (Rothaermel et al. 2007) indicated that 173 academic articles have been written on university-related entrepreneurship between 1981 and 2005, and that almost 75 per cent of these were published since 2000.4 What one finds in reviewing this literature is a very large number of articles investigating the relationship between the number and type of firms spun-off from the university, and the attributes of the university from which these firms arose, including attributes of the university’s technology transfer office. Relative to the level of interest, surprisingly little research has been done on the founding and performance of these firms with respect to the locality in which they find themselves. In this chapter the characteristics and attributes of university-based clusters in the United States are described, and the history and development of two such clusters, the University of Wisconsin-Madison (UW-M) and the University of Illinois at Urbana-Champaign (UIUC), is compared.

1

UNIVERSITIES AND KNOWLEDGE-BASED CLUSTERS

Research universities are producers and disseminators of knowledge. This knowledge is transmitted to society through multiple channels. The most common channel is through the university’s role as educator. Research universities also produce knowledge which is transmitted through a variety of other channels including publication in professional and academic journals, hosting conferences, professorial consulting, and the mobility of university graduates (Stephan 2007). More recently, patents have become another form of transmitting this knowledge of society. A number of these transmission channels ensure that university knowledge is widely distributed geographically, that is, the knowledge is not necessarily confined to the region in which the university is located. The academic literature suggests that even though much of this

216

Emerging clusters

knowledge is transmitted widely, certain, more tacit, knowledge may be most easily captured locally. This local capture occurs when students join local firms and professors are involved in consulting with nearby firms. It can also occur when knowledge becomes the basis of a new, locally-based start-up enterprise. Universities are recognized as an essential institution in many of the most celebrated innovative regions of the world (Etzkowitz 2004; Storper and Salais 1997). Indeed, the entire evolution and development of Silicon Valley has been profoundly influenced by the role of Stanford University and the University of California, Berkeley (Kenney 2000; Saxenian 1994).5 These two universities are the crucial educational institutions within Silicon Valley that observers have termed an ‘ecosystem’ (Bahrami and Evans 2000), a ‘social structure of innovation’ (Florida and Kenney 1990), or an ‘incubator region’ (Eisenhardt and Schoonhoven 1990). Since the 1980s there has been a growing appreciation of the university’s role in transferring the knowledge developed through public research to the larger society, particularly the private sector in the form of commercial ventures. Scholarly interest in this role has tended to focus on the university’s direct role in promoting new firm formation on the basis of university inventions. In particular, this interest has centered on university spin-offs in high technology, science-based fields, such as biotechnology (Kenney 1986; Zucker et al. 2002) and, more recently, nanotechnology. The existence of a permanent, externally supported research university committed to the promotion of entrepreneurship is the distinguishing feature of what we refer to as the university research-centric cluster. Because this type of cluster is based on the research of various disciplines within the university it is quite different from other clusters in one fundamental way. Industry-specific clusters react to the demands of the market as products and innovations are produced explicitly in response to the market. University research-centric-based clusters, on the other hand, are characterized by the technology push that comes from research in a wide variety of university academic disciplines. As a result the new firms that spin out of the university will be as varied in their product and technology as the research areas pursued by the university. It is not the case that firms entering on the basis of university inventions are not concerned with market demand. Clearly they must be. Rather it is the research conducted by the university, which produces opportunities for entry, that is not subject to the demands of the market. The distinction between the university research-centric cluster described here, and industry-specific clusters characterized by specialization in a particular industry, is based on the source of entrepreneurial opportunities for new firm formation. In the university research-centric cluster, the university generates ‘seeds’ for high technology firm formation, but unless the region

The role of the university in research-based clusters

217

in which these seeds are planted is a rich entrepreneurial environment a successful cluster manifesting external economies may not emerge. The entrepreneurial environment surrounding a university is not a prerequisite of cluster formation, but rather something that grows with new firms and the institutions that emerge to support them (Feldman and Francis 2004). Studies of the genesis of clusters have consistently shown that the attributes of successful, mature clusters maintain entrepreneurial support networks, such as venture capital, that were not in place when the cluster first emerged (Bresnahan et al. 2001). As will be shown by a comparison of two university research-centric clusters below, whether a university produces a cluster characterized by external economies and exhibiting new firm formation through spinoffs, or simply an agglomeration of local firms, depends on the evolution of an environment supportive of university entrepreneurship. The unique attributes of the university research-centric cluster explain its initial formation, how it is maintained, and how learning occurs within it. Because this type of cluster is initially formed by start-ups based on university research, these firms will be established in close proximity to the university, usually in the university town, for two reasons. First, many of these firms will be founded by university faculty who will want to retain their position with the university. In addition, several studies have shown that there is a strong motivation for entrepreneurs to establish their startups locally to be near familiar surroundings, family, and friends (Stam 2007: 37; Dahl and Sorenson 2008). This tendency is particularly strong in the earliest stage of the start-up and is also based on consideration of access to capital and professional networks. Second, the tacit, or contextual, knowledge upon which the start-up is based will exert a strong centripetal force keeping the start-up close to the university as well. There is a large body of literature on the role of proximity in the transmission of tacit information, particularly in a university setting (see Audretsch and Stephan 1996, and Zucker et al. 1998). Peter Maskill (2001), in his knowledge-based explanation of geographical clusters, argues that the advantages of multiple co-located firms pursuing the same activities arise from the knowledge obtained from running parallel projects. ‘Co-localized firms undertaking similar activities find themselves in a situation where every difference in the solutions chosen, however small, can be observed and compared’ (Maskell: 928–9). It is not just that the costs of input–output transactions among firms can be greatly reduced within a cluster, which Storper (1995: 201) refers to as the ‘traded interdependencies’ of a cluster. In addition there is a great advantage in having firms engaged in similar activities competing, and occasionally cooperating, with each other in the same location. This ability to observe other firms pursuing the same activity in close

218

Emerging clusters

proximity is the basis of Brown and Duguid’s observation that there is both a high level of knowledge in firms in Silicon Valley, and a high level of knowledge about firms (Brown and Duguid 2000: 20–23). This knowledge about firms is a function of proximity and shared practice, and explains Marshall’s ‘mysteries in the air’ that is to be found in places like Silicon Valley. Shared practice is knowledge that is embedded in a social setting, a knowledge that comes from learning by being in the place where the knowledge is being used and having the opportunity to use it in that setting. Such a setting, where people working together produce a body of actionable, community-based knowledge, is a community of practice. The concept of communities of practice can assist us in understanding the role of universities in initiating new firms and sustaining existing firms in the university research-centric district. In its early stages the start-up is basically an extrusion from the university laboratory, and the founders and original staff will most likely be drawn from this community. Initially a given start-up, and the cluster itself, is maintained by this interaction between the start-up and its university-based community of practice. Later in the development of the cluster entrepreneurs, in conjunction with the university, they develop institutions and networks that advance their ventures and foster new firm formation. The role of the university in sustaining common codes of communication and networks among actors has been noted by several authors. Miner et al. (2001: 144–5) argue that universities can play a key role in industry formation because they provide a neutral territory in which scientists can form relationships outside of the world of competition. Paniccia (2006) observes that universities, together with alumni associations and others, act as social as well as professional institutions within university-based clusters, where they act as centers of socialization and as arenas for the exchange of ideas and reputation building.

2

TWO EXAMPLES OF UNIVERSITY-BASED CLUSTERS

The dynamics of these clusters can be understood by examining case studies of two elite universities, the University of Wisconsin-Madison (UW-M), and the University of Illinois at Urbana-Champaign (UIUC). Both of these universities have experienced spin-offs and have conscious policies of encouraging cluster formation. Both UW-M and UIUC are large, comprehensive, and highly rated universities. UIUC has top-tier computer science and engineering departments, and technologies that can be traced to UIUC are the basis of Lotus

The role of the university in research-based clusters

219

Notes, the email program Eudora, and web browsers. Inventors coming directly from UIUC include those that founded firms such as Netscape and Paypal. However, these two very important start-ups were not founded in Champaign, but rather were founded in Silicon Valley some 2000 miles away. The importance of regional considerations in the startup process is captured in the following statement by Marc Andreessen, the founder of Netscape, on why he did not consider Champaign as the location for Netscape in the mid 1990s: ‘there’s no infrastructure at all in Illinois for a start-up company. It’s not there. No one does it. They just don’t know how to react to it.’6 In comparison Bill Linton, the founder of biotechnology firm Promega of Madison, said of the environment around UW-M in 1976: ‘A tradition of educational excellence has contributed to an environment of intellectual curiosity, exploring spirit, and intuitive visions – together they create a rich business development environment.’7 While these are only anecdotal observations, our research into the sources of firm foundings in these two university towns indicates that there has been and continues to be a difference between these two regions. UW-M has a much longer history of promoting new firm formation, going back to 1925 with the establishment of a private non-profit entity, the Wisconsin Alumni Research Foundation (WARF) to patent inventions and license technologies emerging from UW-M research. Several significant biotechnology firms have emerged in this cluster including Promega, PanVera, and Tomo Therapy, all of which were founded in Madison and not outside the region. The most significant firms that have been founded on the basis of UIUC research have located in Silicon Valley. In this study all high technology firms that were founded in Madison and Champaign-Urbana, or founded by university personnel, including secondary spinoffs, were recorded going back to 1957 and 1958. In our examination of firm founders we found in both Madison and Champaign that the university was by far the largest source of entrepreneurship among these firms. But unlike Champaign, Madison has developed a biotechnology cluster that supports spinoffs from existing firms. That is, the firms themselves seed new firms, so that new generations of firms emerge that are not directly related to the university. This pattern of firm formation was not observed in Champaign.

3

GENESIS AND EVOLUTION OF TECHNOLOGYBASED CLUSTERS

While studies of clusters have increased in number in recent years, few have examined clusters from a dynamic perspective that appreciates that cluster

220

Emerging clusters

formation is a co-evolutionary process which reflects the joint development of institutions, technologies, and firms. This shortcoming has been recognized, and two recent works on biotechnology clusters (Owen-Smith and Powell 2006; Romanelli and Feldman 2006) have provided insight into the evolution of these clusters over time.8 Owen-Smith and Powell’s study compared the development of the San Francisco Bay Area biotechnology cluster with the development of the Cambridge/Boston cluster from 1988 through 1999, with regard to network evolution among different cluster actors, particularly universities, venture capital firms, and pharmaceutical corporations. They observe that studies that rely on comparative statics may conclude that these clusters evolved in a similar manner. But in fact a consideration of their history reveals that their patterns of development diverge significantly, even though they share many attributes in their mature phase. Romanelli and Feldman’s study focused upon an entire industry, human biotherapeutics, rather than comparing regions. They recorded the history of the founders of 688 biotherapeutic firms from 1976 through 2003 to track cluster formation across all regions. Based on the sources of entrepreneurs of these firms and the emergence of biotherapeutic clusters over this time period, they came to the following conclusions. First, the majority of firms were founded by entrepreneurs in the regions in which they resided, and most regions generated new firms by entrepreneurs out of local universities and research institutes at a fairly steady rate. Second, the largest clusters – San Diego, Boston, and San Francisco – exhibited growth by entrepreneurs leaving local, established firms to create local start-ups. Only regions in which this secondary firm formation occurred grew relative to the other clusters. Third, a significant number of entrepreneurs relocated from one region to another to found firms. This tertiary growth by immigration of entrepreneurs was pronounced in the largest clusters, and occurred late in their development (Romanelli and Feldman 2006: 108–10). These observations imply a three-stage pattern of growth that can be examined by the examples of Madison, Wisconsin and ChampaignUrbana, Illinois. The first stage of cluster development occurs as a result of firm formation by local entrepreneurs based on perceived economic opportunities. A second stage of cluster development may occur as a result of secondary growth through the spinning off of start-ups from established firms in the cluster. Feldman and Francis (2004) observe that it is in this stage that entrepreneurs, by interacting with their environment, emerge as social actors within the cluster, establishing networks and institutions to support their ventures and address their concerns. Finally a third stage of cluster development is reached when a cluster becomes sufficiently well

The role of the university in research-based clusters

221

established to attract entrepreneurs from other regions. These latter two stages, though, do not inevitably emerge from the first. The triggering event in cluster development is only the necessary first stage. The critical moment in cluster development occurs after this initial seeding event. In university research-centric clusters the potential initial triggering event occurs continuously through the research a university conducts.

4

DATA FOR THE UNIVERSITY RESEARCHCENTRIC CLUSTERS OF UW-M AND UIUC

The data for this study was collected over 2006 and 2007, and is comprised of a census of all high technology firms founded in Madison, Wisconsin and Champaign-Urbana, Illinois from 1950 through 2006.9 Firms that were established by university personnel outside of the cluster were also included in this study. The data collection effort was based primarily on Internet sources, but was also based on direct contacts with university officials10 and an e-mail survey of Madison firms.11 The Internet sources used in building this census were numerous, and included local development agencies, venture capital data, the local press, business associations, company websites, and of course Internet searches. Once firms were identified and screened to meet criteria for inclusion the names of the firm founders were established and their biographies obtained. Only de novo, high technology firms were included in this census.12 To be considered de novo a firm had to be founded locally, not be a spin-off from an existing firm, or be a subsidiary or branch operation. Very small firms of just one or two employees providing only services were also excluded, as were all exclusively retail establishments. In determining which firms were high technology, and what type of technology category most accurately described them, the authors relied on consensus in classification by other sources whenever possible. These sources included the firm website, a description in the local or business press, a description by the university technology transfer office, or business association. The initial guidance for the technology classification used in this study was provided by the MG&E high technology directory of Madison, Wisconsin (Madison Gas & Electric Co. 2004). All start-ups are assigned to one of five general technology categories: information technology, engineering, physical sciences, biological sciences, and medical sciences. Information technology includes all Internet and software firms as well as firms dedicated to computer systems and IT services. Engineering includes companies involved in the manufacture of

222

Emerging clusters

computers, scientific instruments, and machinery, while physical sciences includes firms involved in the manufacture of electronics and telecommunications. Biological sciences includes firms involved in the fields of biotechnology, veterinary science, and agriculture. Medical sciences includes all firms that are directly involved in medical instruments, equipment, and services, with the exception of biotechnology and pharmaceutical companies.

5

THE UNIVERSITY AS PLANTER OF SEEDS

Because the university is the primary source of knowledge within the university research-centric cluster, and because this knowledge is the basis for many of the new firms founded within the district, one would hypothesize that the characteristics of new firms would reflect the relative disciplinary excellence of the university. In Tables 9.1A and 9.1B all start-ups founded by faculty, staff, and students at UIUC and UW-M respectively are tallied and grouped into five technological categories. The research and development expenditures by the universities for these categories are for the single year of 2004, while the rankings of university academic programs that fall into these categories are given for the years 2006 and 1995.13 UIUC and UW-M differ significantly in the types of firms that have spun-off from the university. At UIUC, information technology and engineering start-ups account for almost two thirds of the total. Over the years, only seven start-ups based on the life sciences had UIUC founders. At UW-M, on the other hand, the life sciences account for over half of the total number of start-ups. Without reference to either the academic reputation of these universities, or the R&D expenditures by various programs, the differences in start-up technologies in these clusters would be difficult to explain. In the case of UIUC we can see that there is an exact ordinal ranking match between the number of start-ups in each technology category and the rank of comparable university programs by R&D expenditures. UIUC’s R&D expenditures on computer science (information technology) were the highest in the US in 2004, and this was the most important category of start-ups. Its second highest program in R&D expenditure rank was engineering, and this corresponds to engineering being its second most important category of start-ups. This ordinal match proceeds through the other categories. Further, this roughly parallels the academic ranking of these programs as well. UW-M start-up technologies match up ordinally with R&D ranks

The role of the university in research-based clusters

Table 9.1A

University of Illinois Urbana-Champaign-Founded Start-ups

Information Technology Engineering Physical Sciences Biological Sciences Medical Sciences UIUC Total

Table 9.1B

Start-ups 1958–2006

R&D 2004 in $1000s (US rank)

2006 Program (US rank)

1995 Faculty rank

22 36.7% 17 28.3% 14 23.3% 6 10.0% 1 1.7% 60

113 320 (1) 120 032 (10) 50 152 (17) 61 911 (45) 11 331 n.a.

5

8

4

5.5 17.8

24

31.3

no medical school

University of Wisconsin Madison-Founded Start-ups Start-ups R&D 2004 in 1957–2006 $1000s (US rank)

Information Technology Engineering Physical Sciences Biological Sciences Medical Sciences UW-M Total

223

23 20.0% 12 10.4% 15 13.0% 44 38.3% 21 18.3% 115

13 457 (23) 94 860 (14) 51 853 (14) 155 682 (6) 272 640 (11)

2006 Program US rank

1995 Faculty rank

10

10

15

15.6 16.5

12

10.0

26

Notes: The R&D expenditures, as well as academic and faculty rankings, of computer science are used for the category of information technology. The faculty ranks for biological sciences, physical sciences, and engineering are based on averages of fields within these categories. See note 13. Sources: University start-ups: Martin Kenney and Donald Patton. Data furnished on request by the authors. R&D data: National Science Foundation (2006). 2006 program ranks: US News and World Report (2006). 1995 faculty ranks: National Research Council (1995).

224

Emerging clusters

and program ranks with the exception of information technology which appears to have too low an R&D ranking, and engineering which appears to have too high a program ranking, relative to the number of start-ups by UW-M faculty and staff. Basically, though, the rankings are congruent for both universities suggesting that the types of start-ups within each cluster mirror the relative strengths of the universities at their center. These results agree with empirical work on the characteristics of universities and their propensity to produce spin-offs (Di Gregorio and Shane 2003; O’Shea et al. 2005). In most studies the number of spin-offs per year are regressed on a variety of university level attributes. It is then found that the prestige of the university, measured by either the quality of faculty in science and engineering (O’Shea et al. 2005), or by overall graduate school ranking (Di Gregorio and Shane 2003), is positively and significantly related with the number of spin-offs per year based on university licensed technology. Because the spin-off data of these studies was based on the Association of University Technology Managers (AUTM) surveys, the individual identities of the start-ups was suppressed, thereby restricting the analysis to the university rather than department level. The results, though, clearly show that university prestige, and therefore the quality of the ideas emerging from them, is directly related to the number of firms founded upon those ideas.

6

THE INSTITUTIONAL ROLE OF THE UNIVERSITY IN NEW FIRM FORMATION AND ENTREPRENEURSHIP

The extent to which universities extrude their knowledge into the larger economy through start-ups depends not only on the quality of the technology and ideas of their departments. It is also shaped by the offices of the university that attempt to promote entrepreneurship, and the institutions and social relations in which faculty are embedded. Kenney and Goe (2004), in their comparison of the electronic engineering and computer science (EE&CS) departments of UC Berkeley and Stanford found that Stanford faculty were significantly more involved in entrepreneurship than UC Berkeley faculty, and that the primary explanation of this difference lies in the historical legacies and cultures that developed at these two universities. Stanford had a history of encouraging entrepreneurship, while UC Berkeley did not. This explains why two departments of equal prestige, and roughly similar proximity to Silicon Valley, produce a different number of spinoffs. The fact that Stanford produces many more EE&CS spinoffs than UC

The role of the university in research-based clusters

225

Berkeley lies in the institutional differences in the universities. It was assumed that since the disciplines were held constant, the institutional context was the differentiating factor. Feldman and Desrochers (2004) argue that similar institutional factors are at work in explaining why Johns Hopkins University produces fewer start-ups than its reputation and size would predict. Since the passage of the Bayh–Dole Act of 1980 which gave universities the property rights to federally funded inventions developed at their campuses, most universities have established technology licensing offices (TLOs) to expedite the licensing of university ideas and promote the founding of firms based on university technology. In addition, many universities have supplemented these efforts with university sponsored research parks and start-up incubators, as well as sponsored venture capital firms and other types of organizations aimed at providing support to university entrepreneurs. For the University of Wisconsin, the Wisconsin Alumni Research Foundation (WARF) plays a unique and critical role in the UW-M ecosystem as an intermediary in the commercialization of university research. Established in 1925 as a non-profit patent organization funded initially by UW alumni and managed by a Board of Trustees composed of alumni, its independence allows it to operate in an entirely business-like fashion, separate from university politics and academic administration. WARF’s primary purpose is to manage patents based on UW-M research, and since 1928 it has provided more than $915 million to the university to support further research (WARF 2008). WARF was established in 1925 as a vehicle to administer the discovery of UW-M Professor Harry Steenbock to prevent the bone disease rickets. Prof. Steenbock had developed a method of increasing the Vitamin D content of food products by ultraviolet irradiation, but since Steenbock could not secure the cooperation of the university Board of Regents he and other alumni founded WARF, which granted its first license using Steenbock’s discovery to Quaker Oats in 1927 (Sobocinski 1999: 310–11). This patent and other later discoveries on the use of Vitamin D continue to provide between 60 to 70 percent of WARF’s total income (Gulbrandsen 2003). In addition to WARF, the university’s Office of Corporate Relations (OCR), established in 1963, is a critical link from the university to small businesses and the larger economy. The role of the OCR is to act as a broker and counselor. Madison has also experienced a proliferation of small business incubators and business parks, the most important of which is the UW-M-sponsored University Research Park, established in 1984 (Sobocinski 1999: 306).

226

Emerging clusters

The University of Illinois has similar institutions to UW-M, but their experience is much more limited having been established only recently. The university’s agent for technology transfer, the UIUC Office of Technology Management, was established in 1995, a full 70 years after the formation of WARF in Madison. Like UW-M, UIUC has established organizations to support entrepreneurs, but these too are of recent origin. The universitysponsored University Research Park only began construction in 2000, and the university venture capital fund, Illinois Ventures, was proposed as an entity in the same year. Another institution intended to provide mentoring to university entrepreneurs is the Technology Entrepreneur Center. Currently housed within University Research Park, it too was only recently established in 1999 (Technology Entrepreneur Center 2008). It is clear that UIUC is attempting to quickly develop university institutions to support technology transfer from laboratories through new firm formation.

7

THE ENTREPRENEURIAL ENVIRONMENT WITHIN THE CLUSTER

Although the university provides the initial seeding of firms within a potential cluster, this is just the primary stage of growth. For a cluster to thrive new firm formation must be based, at least in part, on the existing firms in the cluster. This secondary, or second generation, growth is the hallmark of vibrant clusters (Romanelli and Feldman 2006; Klepper 2001). In Tables 9.2A and 9.2B all high technology start-ups associated with UIUC and UW-M are presented. The university-founded start-ups are combined with other start-ups founded within Champaign and Madison respectively. The first column presents the count of all start-ups founded within the university town by technology category, including all those founded outside the region by university faculty, staff, and students. The second column indicates the number of firms founded by university personnel, the third column indicates the number of forms founded by one or more individuals from other local high technology start-ups, and the fourth column gives the number of firms founded by university personnel outside the region. These columns are not mutually exclusive. Firms founded by individuals from other local firms may also have been founded by university faculty, and the fourth column is simply a subset of the second column. Several observations can be made from this data. First, in both cases approximately half of all start-ups were founded by university personnel. This holds across technology categories with information technology having a somewhat smaller proportion of university start-ups for both

The role of the university in research-based clusters

Table 9.2A

Champaign-founded and UIUC-founded combined All start-ups 1958–2006

Information Technology Engineering Physical Sciences Biological Sciences Medical Sciences Other Total

Table 9.2B

46 39% 28 24% 23 20% 11 9% 5 4% 4 3% 117

Founded Other local by UIUC high tech faculty/staff founder 22 0.48 17 0.61 14 0.61 6 0.55 1 0.20 1 0.25 61 0.52

2 0.04 0

0 0

0

0 0

2 0.02

Engineering Physical Sciences Biological Sciences Medical Sciences

Source:

13 0.11

Madison-founded and UW-M-founded combined

Information Technology

Total

UIUC founded outside of region 5 0.11 1 0.04 5 0.22 2 0.18 0

All start-ups Founded Other local 1957–2006 by UW-M high tech faculty/staff founder

Other

227

54 27% 22 11% 27 14% 65 33% 26 13% 6 3% 200

23 0.43 12 0.55 15 0.56 44 0.68 21 0.81 2 0.33 117 0.59

4 0.07 2 0.09 0 12 0.18 0 0 18 0.09

UW-M founded outside of region 1 0.02 0 0 0 2 0.08 1 0.17 4 0.02

Martin Kenney and Donald Patton. Data furnished on request by the authors.

228

Emerging clusters

schools, while biological sciences and medical sciences at UW-M, and engineering and physical sciences at UIUC, have a greater proportion of university founders. Second, the mix of technology of all start-ups parallels that of university start-ups. This is what one would expect if one believes that universityaffiliated founders are only a portion of the channels of innovation within a locality. Third, the role of secondary foundings of firms is much greater in Madison than it is in Champaign. Just two start-ups in Champaign had founders who came from other local high technology start-ups. This indicates that in Champaign new firms have been established, but have thus far failed to produce secondary offspring by becoming a source of entrepreneurs in their own right. The situation is quite different in Madison where approximately one in ten of all start-ups had founders from other, locally founded high technology firms. This proportion rises to almost one in five in the biological sciences, the area in which Madison excels. What this indicates is that in biotechnology Madison start-ups encourage the formation of secondary offspring. Although the university is the driver in Madison biotechnology, the resulting firms generated yet other firms, implying that they operate in a fertile entrepreneurial environment. The generation of new firms, as measured by secondary spin-off activity, is an important measure of the vitality of the entrepreneurial environment within a cluster. Earlier we indicated that one of the best known spin-offs from UIUC, Netscape, was founded outside the region due in part to the shortcomings of the Champaign environment. This was not an isolated case. Slightly over one in five (13 out of 61) of all start-ups founded by UIUC personnel were founded outside the Champaign area. Silicon Valley has been a strong attractor of individuals from UIUC seeking to form startups, particularly in the field of information technology as seen in Table 9.2A, but this does not explain all of the firms founded outside of the Champaign region by UIUC personnel. Moreover, just five of these firms were founded in Silicon Valley. Another five were founded in Illinois and one each in Virginia, North Carolina, and Masachusetts. Because UIUC entrepreneurs have founded firms in other areas besides Silicon Valley, and in other technologies besides information technology, it seems clear that Champaign is not retaining all of its entrepreneurs. In Madison all but four start-ups were founded in the Madison area. The question of whether this is a characteristic of the life sciences or the university is not entirely clear, but it should be noted that Madison retained start-ups in the physical sciences and information technology as well as the biological sciences.14

The role of the university in research-based clusters 16

229

Medical Sciences

14

Biological Sciences

12

Physical Sciences

10

Engineering Information Technology

8 6 4

0

pre 1960 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

2

Source:

Martin Kenney and Donald Patton. Data furnished on request by the authors.

Figure 9.1A

Champaign-Urbana and UIUC Start-ups

16

Medical Sciences Biological Sciences Physical Sciences Engineering Information Technology

14 12 10 8 6 4 2

pre 1960 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

0

Source:

Martin Kenney and Donald Patton. Data furnished on request by the authors.

Figure 9.1B

Madison and UW-M Start-ups

The observations that have been made to this point are based on the start-ups in these clusters around UW-M and UIUC over their entire history from the late 1950s through 2006. The dynamic nature of these clusters’ development can be shown by time series data on new firm formation over time, as shown in Figure 9.1A and Figure 9.1B. In these figures two features immediately stand out. First, the cluster around UW-M started much earlier in its development than did UIUC, and second, the cluster in Madison is much larger that the one in Champaign. Another consideration in comparing these two clusters is whether the

230

Emerging clusters

characteristics noted earlier remain constant if we examine their development over time. In particular, does the observed tendency of UIUC entrepreneurs to leave Champaign remain constant, or has this tendency declined as the cluster has developed? The data indicates that entrepreneurs from UW-M founded start-ups outside of Madison in the 1980s only. Since that time all university entrepreneurs have found that Madison has a suitable business environment for start-ups. The situation in Champaign is quite different. Although Netscape was the most famous of the start-ups established outside of Champaign, it was just one of the first. The data indicates that 10 of the 13 firms that were founded outside the region emerged from the university during the current decade, the other three being founded before the year 2000. It would seem that efforts to encourage local entrepreneurship are failing to keep many of the new start-ups in town, even as the number of new firms being founded has remained quite high through the first decade of this century as shown in Figure 9.1A. Without looking more closely into the motivations of individual entrepreneurs it is difficult to conclude much more from the data on start-ups within these two university research-centric clusters. To remedy this situation, two histories of innovation at these universities are discussed below. The first history considers the series of discoveries on the uses of Vitamin D at UW-M over many years, primarily by Professor Hector DeLuca and his associates. The second history is of the development of the first graphical Internet browser, Mosaic, at the National Center for Supercomputing Applications (NCSA) at UIUC in the early 1990s. Both of these innovations are chosen for their high profile and insight they provide into the operation of these clusters. Hector DeLuca came to UW-M in 1951 and was the last graduate student to work under Harry Steenbock, the professor whose discoveries in Vitamin D research was directly responsible for the founding of WARF in 1925. DeLuca’s subsequent work has resulted in both academic achievements and patent royalties of close to $100 million for UW-M through WARF administration, as well as three university start-ups founded directly from his work; Lunar Corporation, Bone Care International, and Tetrionics (Sobocinski 1999: 294). Professor DeLuca and his research team synthesized calcitriol in 1971, a substance that increases calcium absorption and regulates blood calcium levels. Calcitriol has been used to successfully prevent osteoporosis, and it has been modified into vitamin D analogs. These vitamin D analogs are used to treat renal osteodystrophy, vitamin D resistant rickets, and parathyroid gland failure (Sobocinski 1999: 190). The production of these vitamin D analogs is the basis of the Madison start-up Bone Care

The role of the university in research-based clusters

231

International, founded in 1984. Synthetic vitamin D pharmaceuticals were also the basis of the local start-up Tetrionics, founded in 1990. The work of DeLuca and his colleagues has resulted in over 150 US patents files by WARF, many of which have been licensed to US and foreign corporations. Over the years DeLuca has established and maintained extensive networks of academics and entrepreneurs, effectively linking his laboratory to a set of firms within Madison and outside the region, as well as other university biochemistry departments. Such university entrepreneurs are not uncommon in Madison, and the richness of the networks and culture that have grown up are evident in the number of organizations that have been established precisely for the purpose of networking in the area, such as the Wisconsin Technology Network, the Wisconsin Angel Network, the Wisconsin Technology Council, and the Wisconsin Entrepreneurs’ Network. These networks are embedded in both the business environment and the community of Madison, and are characteristic of the second stage of cluster development as described by Feldman and Francis (2004: 129). The development of the Mosaic browser at UIUC is an important part of the history of the development of the Internet. Prior to the release of Mosaic in 1993, access to the Internet on the World Wide Web was limited to browsers that were based solely on text, ran on the Unix operating system, and were oriented towards academics and professional engineers rather than mainstream users (Naughton 1999: 236–7; Reid 1997: 4–5). By early 1993 there were around 50 Web servers in the world, one of which was located at the National Center for Supercomputing Application (NCSA) at UIUC (Berners-Lee 1999: 67–8). Marc Andreessen, an undergraduate intern at the NCSA, together with Eric Bina, a full time employee at NCSA, developed a Unix version of a graphical browser called Mosaic that could be run on personal computers. Mosaic was released in January 1993 and soon became the standard browser in the rapidly emerging world of the Web, but its limitations required customer support which the development team at NCSA was unable to handle effectively. In addition, with Mosaic’s great success, issues of ownership and control over the innovation began to arise at the NCSA. Marc Andreessen decided to leave the area after it became obvious that he would not be the head of the Mosaic project at the NCSA after he graduated from UIUC. Andreessen initially took a position in Palo Alto, California with Enterprise Integration Technologies (EIT), a developer of Web security products. Although the firm was involved with the Web, Andreessen was not hired to extend the development of Mosaic. The fact that he received a good offer from EIT and was attracted to the region were his motivations in taking this job (Stark 1995). It was not until February

232

Emerging clusters

1994, several months after he started working for EIT, that Andreessen resumed his work on browser software, having received an offer from Jim Clark, the founder of Silicon Graphics, to build a firm based on the Web. What this history suggests is that Marc Andreessen was not so much drawn to Silicon Valley to pursue his innovation of Mosaic, as he was pushed from the source of this innovation in Champaign. Andreessen’s and Clark’s collaboration resulted in the founding of Mosaic Communications Corporation, later known as Netscape, in 1994. To obtain the talent needed to advance their venture Andreessen and Clark flew to Champaign in early 1994 and basically hired away the Mosaic development team from the NCSA. Would there have been any conditions where Andreessen would have founded a firm based on Mosaic in Champaign, Illinois? It was certainly the case that Andreessen had at least thought of such an idea before he left for California (Reid 1997: 21). The main deterrence was the absence of an entrepreneurial infrastructure for new firm formation in Champaign. Jim Clark’s start-up experience and financial means, coupled with his wide array of contacts into Silicon Valley entrepreneurial support networks, provided this infrastructure. Yet coming to Silicon Valley to found a company was not the motivation for Andreessen’s move. Clearly chance was a factor in all of these decisions, yet the establishment of Netscape in Mountain View, California was due less to the pull of Silicon Valley than it was to the failure of institutions and networks in Champaign to retain the entrepreneurs that established the firm. Consider this situation with that facing the university entrepreneur scientist Hector DeLuca at UW-M. When DeLuca came to UW-M in 1951 he worked as a graduate student under Harry Steenbock. By this time the model of the scientist researcher, whose work is directed to commercial applications, was well established. Indeed, it was Prof. Steenbock whose discoveries led directly to the establishment of WARF, and established the pattern, and the legitimacy, of professors taking the results of research from the university laboratory and extruding them into the larger economy through licenses issued by WARF and new start-ups. Not only was engaging in research that had direct commercial application not discouraged, but at UW-M it was seen as acceptable and a means to advance one’s scientific reputation.

8

CONCLUSION

At the heart of a university research-centric cluster is a research university, a permanent, externally supported institution that is mandated to teach

The role of the university in research-based clusters

233

and conduct research. Because such a cluster is based on the research of various disciplines within the university, it may be distinguished from other clusters in a number of ways. Other clusters achieve external economies by their market specialization. They are characterized by numerous competitors producing similar goods, resulting in deep horizontal and vertical relations. The management and technical expertise is steeped in industry knowledge and lore so that a new start-up can quickly attract key personnel. The founding of Netscape is a perfect illustration of the vitality of such clusters. In Silicon Valley, Netscape quickly attracted seasoned managers and was able to draw upon the industry knowledge of the region. In Champaign these resources were simply not available, and in fact, there are few other places in the world where such a depth of information technology-related talent can be found. University research-centric clusters, in comparison with industryspecific clusters, are characterized by technological innovations that come from a wide range of disciplines. As a result, the new firms that spin out of the university will be as varied in their product markets as the research areas pursued by the university. In the university research-centric cluster, the university generates ‘seeds’ for firm formation, but unless the region in which these seeds are planted is a rich entrepreneurial environment, a well functioning cluster may not emerge. Indeed, one of the primary differences between the two examples of university research-centric clusters discussed in this chapter is the vitality of the entrepreneurial settings in which the universities of Wisconsin and Illinois are located. The cluster of firms found around UIUC has only the structural core of a university research-centric cluster, namely the presence of a large, highly ranked research university. UIUC satisfies the role of an institution which plants seeds for new firm formation. Yet a number of these seeds, such as Netscape and Paypal, took root in Silicon Valley rather than Champaign. The cluster around UW-M, on the other hand, has achieved a level of success where existing firms are the basis of second generation spinoffs. This secondary growth is a hallmark of vibrant clusters, and an indication that a cluster has entered a second stage of development from an initial seeding event (Feldman and Francis 2004: 129–30). UW-M has several features that UIUC lacks. First, the university is deeply involved in the governance of the cluster through such longstanding organizations as WARF and the Office of Corporate Relations, among others. The efforts made by UIUC in promoting entrepreneurship are much more recent, going back only to the 1990s, and are much more modest in size and scope. Second, the ties of networks among entrepreneurs in the community, and their counterparts in university laboratories, are deep and long-standing

234

Emerging clusters

in Madison. These ties are promoted by university organizations, and by private organizations that emerged from the entrepreneurial community in Madison. Such private organizations are almost completely absent in Champaign. One finding from this investigation is that new firms in university-based clusters closely reflect the university strengths at its core, at least in the case of UIUC and UW-M. Although this is a modest effort based on comparing two universities, it does suggest that academic excellence contributes directly to entrepreneurial opportunities in a locality in an unambiguous way. Given the level of interest in university entrepreneurship among researchers in economic geography and economic development, it seems clear that further studies of university-based clusters of this kind would yield valuable insights.

NOTES *

1.

2. 3. 4.

5.

6. 7. 8. 9. 10.

The authors would like to thank the participants of the Workshop on Emerging Clusters: Theoretical, Empirical and Political Aspects of the First Stage of Cluster Evolution, held at the Max Planck Institute of Economics in Jena, Germany, June 26–28, 2008, for their comments. In particular we want to thank two anonymous referees for their very valuable comments on an earlier version of this chapter. An important triggering event in the formation of Silicon Valley was the decision of William Shockley to locate his transistor firm in Palo Alto to be near his mother. Had he chosen the Boston area it seems likely that the history of Silicon Valley would have been quite different. Yet other high technology firms such as Varian and HewlettPackard were already in the area suggesting that it would have emerged as an important electronics cluster in any event (Sturgeon 2000). The clusters of Oxford University and Cambridge University are notable examples of such clusters in the UK that have received considerable attention (Lawton Smith and Ho 2006; Proudfoot 2004; Garnsey and Heffernan 2005). For a critique of Bayh–Dole as an encouragement of entrepreneurship, see Kenney and Patton (2008). Rothaermel et al. 2007 is a literature analysis of articles explicitly focused on university entrepreneurship. These 173 articles cover four major research streams: (i) entrepreneurial research university, (ii) productivity of technology transfer offices, (iii) new firm creation, and (iv) environmental context including networks of innovation. The influence of Stanford University through the role of Frederick Terman, department chair of the electrical engineering department and university provost, would be hard to exaggerate given his encouragement of William Hewlett, David Packard, and the Varian brothers to establish firms in the area. As quoted in Scott Shane’s study of university spin-offs (2004: 99). As quoted in Randall Willis’ report on Madison’s biotechnology cluster (2004: 41). Both of these articles appeared in Cluster Genesis (Braunerhjelm and Feldman 2006), a collection of studies directed towards examining the emergence of clusters as an evolutionary process. In this study the regions around Madison and Champaign encompass all locations within one hours’ drive by automobile, or around a 50 mile radius. At UW-M the Office of Corporate Relations provided very valuable information on university start-ups, as did the Initiative for Studies in Technology Entrepreneurship

The role of the university in research-based clusters

11.

12. 13.

14.

235

at the Wisconsin School of Business. At UIUC the Office of Technology Management provided useful information on university start-ups. In addition we relied heavily on the High-Tech Directory published by MG&E (Madison Gas & Electric Co. 2004) and the text by Philip Sobocinski (1999) for data on Madison. An e-mail survey of 103 Madison start-ups was conducted in February 2007. The following information was requested: the year and location of the firm founding, the names of the founders, what these founders did prior to the start-up, and the education of the founders. Out of 103 firms surveyed 47 responses were received. About seven of these responses contained no information, so 40 out of 103 were actually informative. High technology is quite broadly defined. For example, a firm that produces equipment for biotechnology laboratories would be included as a high technology firm, while a firm that raises mice for laboratories would not. US News and World Report publishes an annual ranking of US graduate programs. They do not provide a rank for Physical Sciences as a graduate program. The 1995 faculty quality rankings are drawn from the National Research Council’s 1995 review of US doctoral programs. The ranking of faculty quality for biological sciences is the average faculty rank of four biological fields: biochemistry and molecular biology; cell and development biology; ecology, evolution and behavior; and molecular and genetic studies. The ranking for engineering is based on four engineering fields: chemical, civil, electrical, and mechanical. The ranking of physical sciences is the average of four fields: chemistry, geosciences, mathematics, and physics. The NRC did not consider medical schools, so no faculty rank of medical sciences is provided. See Kenney and Patton (2005: 223–5) for a discussion of the impact of the university on the geography of different high technology start-ups.

REFERENCES Audretsch, D.B. and P.E. Stephan (1996), ‘Company-scientist locational links: The case of biotechnology’, The American Economic Review, 86 (3), 641–52. Bahrami, H. and S. Evans (2000), ‘Flexible recycling and high-technology entrepreneurship’, in M. Kenney (ed), Understanding Silicon Valley, Stanford, CA: Stanford University Press, pp. 165–89. Berners-Lee, T. (1999), Weaving the Web: The Original Design and Ultimate Destiny of the World Wide Web by its Inventor, New York: Harper Collins Publishers. Braunerhjelm, P. and M. Feldman (eds) (2006), Cluster Genesis, Oxford: Oxford University Press. Bresnahan, T., A. Gambardella, and A. Saxenian (2001), ‘Old economy inputs for new economy outcomes, cluster formation in the new Silicon Valleys’, Industrial and Corporate Change, 10 (4), 835–60. Brown, J.S. and P. Duguid (2000), ‘Mysteries of the region: Knowledge dynamics in Silicon Valley’, in C.M. Lee, W. Miller, M. Hancock, and H. Rowen (eds), The Silicon Valley Edge: A Habitat for Innovation and Entrepreneurship, Stanford, CA: Stanford University Press, pp. 16–39. Dahl, M.S. and O. Sorenson (2008), ‘The social attachment to place’, electronic copy available at: http://ssrn.com/abstract=1292224, accessed 13 March 2009. Di Gregorio, D. and S. Shane (2003), ‘Why do some universities generate more start-ups than others?’, Research Policy, 32, 209–27. Eisenhardt, K. and C.B. Schoonhoven (1990), ‘Organizational growth: Linking

236

Emerging clusters

founding team strategy, environment, and growth among U.S. semiconductor ventures, 1978–1988’, Administrative Science Quarterly, 35 (3), 504–29. Etzkowitz, H. (2004), ‘The evolution of the entrepreneurial university’, International Journal of Technology and Globalisation, 1 (1), 64–77. Feldman, M. (2001), ‘The entrepreneurial event revisited: Firm formation in a regional context’, Industrial and Corporate Change, 10 (4), 861–91. Feldman, M. and P. Desrochers (2004), ‘Truth for its own sake: Academic culture and technology transfer at the Johns Hopkins University’, Minerva, 24, 105–26. Feldman, M.P.and J.L. Francis (2004), ‘Homegrown solutions: Fostering cluster formation’, Economic Development Quarterly, 18 (2), 127–37. Florida, R. and M. Kenney (1990), The Breakthrough Illusion: Corporate America’s Failure to Move from Innovation to Mass Production, New York: Basic Books. Garnsey, E. and P. Heffernan (2005), ‘Clustering as multi-levelled activity; The Cambridge case’, presented at the 4th European Meeting on Applied Evolutionary Economics, 19–21 May, De Uithof, Utrecht, The Netherlands. Gulbrandsen, C. (2003), ‘WTN interview: Carl Gulbrandsen WARF – A unique and successful technology transfer organization’, December 9, by M. Klein, Wisconsin Technology Network, http://wistechnology.com/article. php?id=416. Kenney, M. (1986), Biotechnology: The University–Industrial Complex, New Haven, CT: Yale University Press. Kenney, M. (ed.) (2000), Understanding Silicon Valley, Stanford, CA: Stanford University Press. Kenney, M. and W.R. Goe (2004), ‘The role of social embeddedness in professorial entrepreneurship: A comparison of electrical engineering and computer science at UC Berkeley and Stanford’, Research Policy, 33, 691–707. Kenney, M. and D. Patton (2005), ‘Entrepreneurial geographies: Support networks in three high-technology industries’, Economic Geography, 81 (2), 201–28. Kenney, M. and D. Patton (2008), ‘Reconsidering the Bayh–Dole Act and the current university technology licensing regime’, Berkeley Roundtable on the International Economy Working Paper (draft). Kirzner, I.M. (1997), ‘Entrepreneurial discovery and the competitive market process: An Austrian approach’, Journal of Economic Literature, 35 (1), 60–85. Klepper, S. (2001), ‘Employee start-ups in high-tech industries’, Industrial and Corporate Change, 10 (3), 639–74. Lawton Smith, H. and K. Ho (2006), ‘Measuring the performance of Oxford University, Oxford Brooks University and the government laboratories’ spin-off companies’, Research Policy, 35, 1554–68. Madison Gas and Electric Company (2004), 2004 Greater Madison Wisconsin Area Directory of High-Tech Companies, Madison, WI: Madison Gas and Electric Company. Maskell, P. (2001), ‘Towards a knowledge-based theory of the geographical cluster’, Industrial and Corporate Change, 10 (4), 921–43. Miner, A.S., D.T. Eesley, M. Devaughn, and T. Rura-Polley (2001), ‘The magic beanstalk vision’, in C.B. Schoonhoven and E. Romanelli (eds), The Entrepreneurial Dynamic, Stanford, CA: Stanford University Press. Mowery, D.C., R.R. Nelson, B.N. Sampat, and A.A. Ziedonis (2004), Ivory Tower and Industrial Innovation, Stanford, CA: Stanford University Press.

The role of the university in research-based clusters

237

National Research Council (1995), Research-Doctorate Programs in the United States: Continuity and Change, Washington DC: The National Academies Press. National Science Foundation, Division of Science Resources Statistics (2006), Academic Research and Development Expenditures: Fiscal Years 2004, Arlington, VA: National Science Foundation. Naughton, J. (1999), A Brief History of the Future: The Origins of the Internet, London: Weidenfeld & Nicolson. O’Shea, R.P., T.J. Allen, A. Chevalier, and F. Roche (2005), ‘Entrepreneurial orientation, technology transfer and spinoff performance of U.S. universities’, Research Policy, 34, 994–1009. Owen-Smith, J. and W.W. Powell (2006), ‘Accounting for emergence and novelty in Boston and Bay Area biotechnology’, in P. Braunerhjelm and M. Feldman (eds), Cluster Genesis, Oxford: Oxford University Press, pp. 61–83. Paniccia, I. (2006), ‘Cutting through the chaos: Towards a new typology of industrial districts and clusters’, in B. Asheim, P. Cooke, and R. Martin (eds), Clusters and Regional Development: Critical Reflections and Explorations, London: Routledge, pp. 90–114. Proudfoot, N. (2004), ‘The biopharmaceutical cluster in Oxford’, in C. Crouch, P. Le Gales, C. Trigilia, and H. Voelzkow (eds), Changing Governance of Local Economies: Responses of European Local Production Systems, Oxford: Oxford University Press, pp. 237–60. Reid, R.H. (1997), Architects of the Web, New York: John Wiley & Sons, Inc. Romanelli, E. and M. Feldman (2006), ‘Anatomy of cluster development: Emergence and convergence in the US human biotherapeutics, 1976–2003’, in P. Braunerhjelm and M. Feldman (eds), Cluster Genesis, Oxford: Oxford University Press, pp. 87–112. Rothaermel, F.T., S.D. Agung, and L. Jiang (2007), ‘University entrepreneurship: A taxonomy of the literature’, Industrial and Corporate Change, 16 (4), 691–791. Saxenian, A. (1994), Regional Advantage: Culture and Competition in Silicon Valley and Route 128, Cambridge: Harvard University Press. Shane, S. (2004), Academic Entrepreneurship: University Spinoffs and Wealth Creation, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Sobocinski, P.Z. (1999), Creating High-Tech Business Growth in Wisconsin, Madison, WI: University of Wisconsin System Board of Regents. Stam, E. (2007), ‘Why butterflies don’t leave: Locational behavior of entrepreneurial firms’, Economic Geography, 83 (1), 27–50. Stark, T. (1995), ‘The Marc Andreessen interview page’, May, http://www. starkrealities.com/marca.html. Stephan, P. (2007), ‘Wrapping it up in a package: The location decision of new PhDs going to industry’, in A. Jaffe, J. Lerner, and S. Stern (eds), Innovation Policy and the Economy, Volume 7, Cambridge, MA: MIT Press. Storper, M. (1995), ‘The resurgence of regional economies, ten years later’, European Urban and Regional Studies, 2 (3), 191–221. Storper, M. and R. Salais (1997), Worlds of Production: Frameworks of Action in the Economy, Cambridge, MA: Harvard University Press. Sturgeon, T. (2000), ‘How Silicon Valley came to be’, in M. Kenney (ed.), Understanding Silicon Valley, Stanford, CA: Stanford University Press, pp. 15–47.

238

Emerging clusters

Technology Entrepreneur Center (2008), ‘About TEC’, http://www.tec.uiuc.edu/ tec/about_tec.php, accessed 1 October 2008. US News and World Report (2006), America’s Best Graduate Schools, 2006 Edition. Washington DC: US News & World Report Inc. Willis, R.C. (2004), ‘America’s biotech heartland’, Modern Drug Discovery (June), 41–6. Wisconsin Alumni Research Foundation (WARF), 2008, ‘Our history’, http:// www.warf.ws/about/index.jsp?cid=26, accessed 12 October 2008. Zucker, L. G., M. R. Darby, and J. S. Armstrong (2002), ‘Commercializing knowledge: University science, knowledge capture, and firm performance in biotechnology’, Management Science, 48 (1), 138–53. Zucker, L.G., M.R. Darby, and M.B. Brewer (1998), ‘Intellectual human capital and the birth of U.S. biotechnology enterprises’, The American Economic Review, 88 (1), 290–306.

10.

Sources of ‘second generation growth’: spin-off processes in the emerging biochip industries in Jena and Berlin Max-Peter Menzel

1

INTRODUCTION

Clusters are important elements in regional development (Porter 2000; Martin and Sunley 2003; Malmberg and Maskell 2002). Firms in clusters exhibit stronger growth and knowledge between them diffuses faster compared to non-clustered firms (Swann and Prevezer 1996; Audretsch and Feldman 1996; Baptista 2000). Yet, the many failed attempts to copy the features of Silicon Valley to other places led to the insight that the way clusters work has little to do with the way they emerge (Bresnahan et al. 2001; Menzel and Fornahl 2007). There are mainly two insights about cluster emergence that are commonly accepted. The first is that a wave of new firms emerges in many regions at the beginning of an industry life cycle, but these firms will become the source for a cluster only in few regions (Romanelli and Feldman 2006; Klepper 2007b; Boschma and Wenting 2007). Several approaches explain this pattern. The window of locational opportunity (Storper and Walker 1989), the core–periphery model (Krugman 1991) and the stochastic approaches that build on the models of Arthur (1994) explain cluster emergence as a result of ‘trivial historical accidents’ (Krugman 1991; 35) that manifest themselves in firm formations and subsequent agglomeration economies that start to work in the region with the most firms at the beginning.1 In these approaches, the cluster arises where economies of agglomeration start to work at first. But these approaches do not explain the second important insight on cluster emergence, namely that only a particular form of formation seems to be responsible for it, namely spin-offs (Klepper 2007a, b; Romanelli and Feldman 2006). Klepper (2007b) shows that clusters emerge by spin-off processes that

239

240

Emerging clusters

originate in only some ancestor firms. Romanelli and Feldman (2006), using different concentrations of biotherapeutics firms in the USA, describe that clusters emerged in those places where the firm formations of the first wave created spin-offs. They termed this process ‘second generation growth’. In contrast, other kinds of formations like start-ups or branches of other firms had a negligible effect on cluster emergence. This pattern led Klepper (2007a, b) to assume that clusters arise without the necessity of agglomeration economies. Instead, the inheritance of routines from incumbent firms to their spin-offs is the crucial process in cluster emergence. In this line of argumentation, better firms grow stronger than other firms. Due to their growth, they will sooner be able to generate spin-offs that again grow above average rate. As spin-offs tend to locate near their incubator, large concentrations come into being without economies of agglomeration, with the largest concentration located in the region with the best firm. Klepper supported his argument for example amongst others with the growth of the automobile cluster in Detroit (Klepper 2007a) and the tyre cluster in Akron (Bünstorf and Klepper 2009). This strand of work strongly emphasises that it is not simply firm formation, but the growth pattern that decides if a cluster emerges or not. However, the argument that only firm-specific characteristics play a role in cluster emergence contradicts the common wisdom of those research areas that scrutinise the regional embeddedness of firm formation processes (Saxenian 1994; Martin and Sunley 2006). The chapter tries to shed light on sources of second generation growth by focussing on the event that precedes the spin-off, namely on the knowledge creation that forms the basis for the actual spin-off (Menzel 2008b). In doing so, it scrutinises how this knowledge actually is created, either within the firm or by interactive learning in a regional context. Two case studies on the emerging biochip industries in Jena and Berlin serve to analyse the dynamics behind second generation growth in these young industries. This chapter proceeds as follows: after some theoretical considerations in the second section, it illustrates the background to the biochip industry and the development of its spatio-temporal pattern in Germany in the third and fourth sections. The fifth section describes the methods. The sixth and seventh sections analyse the emergence of the industry in Jena and Berlin with special consideration of second generation growth. The eighth section discusses the results, and the ninth section concludes.

Spin-off processes in biochip industries in Jena and Berlin

2

241

FIRM FORMATION AND VARIETY CREATION IN YOUNG INDUSTRIES

Before the start of theoretical and empirical analysis, it may be helpful to begin with the question of what clusters actually are. There are different definitions of a cluster (see Martin and Sunley 2003). Probably the most applied definition is from Porter (2000, 16): A cluster is a geographically proximate group of interconnected companies and associated institutions in a particular field, linked by commonalities and complementarities.

Besides spatially concentrated firms in a certain area, this definition contains both a supporting infrastructure and interactions between the enterprises. Connections and mutual dependencies between firms are considered to be the cause of the positive effects within clusters. As emerging clusters are neither marked by a large concentration of firms nor by extensive agglomeration effects, the definition of a cluster therefore does not apply for their phase of emergence. Emerging clusters differ in two points from normal regions with normal economic activity. The first difference lies in the enterprises themselves. One or several enterprises exist that offer a sustainable vision for a new local technology path. The second is in the local context. Certain conditions are available that give an arising cluster the potential to reach a critical mass, for example a strong scientific base or political support (Menzel and Fornahl 2007). Therefore, the term ‘emerging cluster’ only describes a certain potential to form a growing and functioning cluster. Spin-offs contribute to realising this potential. Spin-offs are generally considered as a source of variety. However, they actually represent a certain manifestation of variety that already exists within incumbent organisations (Frenken and Boschma 2007). Therefore, new firms are the result of variety created before and the two processes of firm formation and variety creation have to be considered separately (Menzel 2008b). The scenarios under which spin-off processes occur, like discontented employees (Klepper 2007a), a crisis of the firm (Longhi 1999; Feldman 2001; Bünstorf and Fornahl 2006) or a particular milieu (Saxenian 1994), are already described in the literature. Therefore, the focus of this chapter is on the knowledge creation that precedes a spin-off. New knowledge is created by recombination of existing knowledge (Jacobs 1969, Nooteboom 1999). This recombination can take place within firms, for example by organisational learning (Kogut and Zander 1992). Klepper (2007a, b) argues that firms with better routines grow

242

Emerging clusters

stronger, create more variety and hence also more spin-offs. In this vein, recombination of knowledge and knowledge generation depends on the firm’s routines and thus on firm-internal capabilities. However, studies point out that especially the characteristics of new industries necessitate localised learning processes and thus knowledge generation in an interactive way. In new industries, a technological trajectory has not been established fully (Dosi 1988). The insecurity about the future trajectory, the mostly tacit nature of the new technology and the volatility of relations result in geographically proximate interactions especially during the early stages of the industry life cycle (Audretsch and Feldman 1996; Brenner 2005). In these cases, firms produce new knowledge by recombination of knowledge from different sources in a regional context. It is reasonable to assume that these recombinations would become the basis for spin-offs. To elaborate on the sources that are responsible for second generation growth, this chapter contrasts two hypotheses. The first is based on the work of Klepper and states that firm routines are responsible for the variety upon which spin-offs are based. A proof of this hypothesis would require that these routines allowing spin-offs to emerge are generated from within incumbent firms and that spin-off processes are based on few ancestries. The second states that regional learning processes create the knowledge that results in spin-offs. A proof of this hypothesis would require tracing the competencies on which spin-offs are based back to interactions with other organisations in the same region. These contrasting hypotheses are tested on the emerging biochip industries in Jena and Berlin.

3

TECHNOLOGICAL AND INSTITUTIONAL BACKGROUND OF THE BIOCHIP INDUSTRY IN JENA AND BERLIN

Biochip technology is a bio-analytical method to detect genes or proteins in an assay. The application field of biochips ranges from the detection of genetically modified food to cancer. The biochip itself is normally a small slide of silicon, glass or plastic with a special surface to bind biomarkers. These biomarkers detect particular genes or proteins and are the actual tests. Like a computer chip, a biochip is only a part of a larger system and requires complementary equipment: spotting devices, which work similarly to ink-jet printers and fix the biomarkers to the chip; micro fluidic technologies which in some cases are necessary to mix, move and distribute liquids on the chip; laser scanners or cameras

Spin-off processes in biochip industries in Jena and Berlin

243

to read out the signals of the chip, and special software to analyse the data gained by the signals. The advantages of biochips are the parallelisation and miniaturisation of bio-analytical tests. The industry started first with the formation of Affymetrix in Santa Clara in 1992, which was the first firm with the aim of developing a commercially available biochip. The scientific base of the biochip industry in Germany is mostly formed by three organisations: the Max-Planck-Institute for Molecular Genetics (MPIMG) in Berlin, the German Cancer Research Center (DKFZ) in Heidelberg, and a branch of the Fraunhofer Institute for Biomedical Engineering (IBMT) in Potsdam, near Berlin. Beside the research institutions, several public technology development programmes implicitly or explicitly supported the diffusion of biochip technology. In the mid 1990s, the German Federal Research Ministry started a new funding concept of a region-oriented technology policy (Dohse 2007). The intention was to support the regional collaboration between firms and the regional science base to support knowledge diffusion. Three programmes with mandatory regional collaborations were developed with this intention and had an impact on the development of the biochip industry in Jena and Berlin: BioRegio, InnoRegio and InnoProfile. In 1995, the BioRegio competition initiated a catching up process in the German biotech industry. The programme would stimulate the formation of new and the growth of existing biotechnological enterprises in a selected number of regions (Dohse 2007). In 1996, the regions Munich, Rhineland and Rhine-Neckar triangle were selected as winners of the competition. The Jena region got a special vote, due to the reorientation of its scientific programmes and economy to biotechnology. The promotion of projects started in 1997. From the 20 projects that were supported in Jena, two were directly focussed on biochips. A similar programme was InnoRegio. The competition started in 2000 with the intention of supporting the catching up process of the East German economy in general. Unlike BioRegio, InnoRegio had no technological orientation. This programme supported the BioHytec network in Berlin, which has a strong focus on biochips and related industries in Berlin and Brandenburg. InnoProfile started in 2005 and was also limited to East Germany. The intention of the programme was to support technologyoriented collaboration in regions between regional SMEs and research organisations. This programme supports biochip initiatives like the Jena Biochip Initiative which intends to develop and commercialise a new biochip-based platform technology and iPOC (Integrated Protein Chips for Point of Care Diagnostics) in Berlin.

244

4

Emerging clusters

SPATIO-TEMPORAL PATTERN OF THE DEVELOPMENT OF THE BIOCHIP INDUSTRY IN GERMANY

A complete survey was conducted to describe the spatio-temporal development of the industry. The niche character hinders the delimitation of the industry to standard industry classification. Therefore, data about the firms in this industry were mainly collected from two databases. The first is www.biochipnet.com,2 which lists firms in the biochip field worldwide. From this database, German-based firms were selected. The second source is www.biotechnologie.de3 which contains firms in Germany that deal with biotechnology. From this database, firms were selected that corresponded to the keywords ‘biochip’ or ‘microarray’. This data was supplemented with information from market surveys or word-of mouth information. Which firms of the biochip field are form formations and which firms are diversifiers was ascertained by telephone interviews. The survey revealed that 104 firms were in the biochip field in 2007, either as their main business field or as a part of it, of which 45 had diversified into the biochip field and 59 already had competencies at the time of their formation. Figure 10.1 gives the spatial pattern of the industry for the years 1997, 2002 and 2007. In 1997, the industry in Germany was still in its infancy. The 17 firms at that time were spatially scattered and an industry was not recognisable. But with the DKFZ in Heidelberg, the MPIMG in Berlin and the BioRegio-Initiative in Jena, there already had been a basis for the emergence of that industry. One interviewee described the early phase as follows: There have been a couple of large enterprises that conducted base studies. There were also small enterprises like (XY), small and fast growing companies, which have taken up the new possibilities and started a biochip programme. Therefore, at that time there have also been firms in Germany. But surely, the attention of the world has focused on Silicon Valley, and the whole VC also went there. The small companies here were relatively weakly equipped with it.

This picture changed in 2002. A boom in the industry led to an increase to 75 firms and created the first concentrations in Munich, Jena and Berlin. The boom was certainly supported by the publicly funded programmes. Beside BioRegio in Jena, the ‘Cluster of Competence’ in BadenWürttemberg and BioHytec in Berlin/Brandenburg started in 1998 and 2000 respectively. Additionally, with the Fraunhofer IBMT in Potsdam (Brandenburg), a third major research facility with strong competencies in biochips was established in the year 2000. But the main triggers of the

245

Figure 10.1

München

Jena

Berlin

Formations

Diversifiers

Heidelberg

Dortmund

München

Jena

2002

Spatial development of the biochip industry in Germany

Research Institutes

Heidelberg

Dortmund

1997

Regional Programmes

Berlin

München

Jena

Berlin

Regional Concentrations

Heidelberg

Dortmund

2007

246

Emerging clusters

boom were the prospects of the technology at the time. A project manager of an SME described them as follows: The imagination . . . was that every doctor would have such a machine. When the patient is coming, you take a drop of blood, put it in the machine, and receive a complete profile of all parameters you want to know.

In 2007, the industry grew to 104 firms. The concentrations remained the same. Jena had eight firms, Munich remained with nine. Additionally, the Dortmund area had seven firms. Berlin was still the largest concentration with fourteen firms; with an additional six in the surrounding area. When the geographical focus is enlarged, North-West Baden-Württemberg with Heidelberg as its major centre may be counted as a concentration, too. The development pattern shows that firms were dispersed at the beginning. During the boom in the industry, a spatial pattern with five concentrations emerged. With North Baden-Württemberg, Jena and Berlin, the largest concentrations are in those areas where research facilities and/or targeted public support programmes exist. The spatial correlation between research institutes and public support programmes is not surprising, as the initiative to apply for public support often comes from the research institutes.

5

METHODS

The evolution of two concentrations, Jena and Berlin, shall be examined in more detail. These cases were selected as they share similar developments regarding development of knowledge base, support programmes and start of the industry. Additionally, both cases have a GDR background and both are not among the centres of biotechnology in Germany. The surrounding areas like Potsdam for Berlin are also assigned to the regions. The chapter contrasts between two mechanisms that lead to spin-offs: variety creation due to firm routines and inheritance as well as variety creation due to regional learning. Both mechanisms are analysed within a single framework. The analysis of the inheritance mechanism requires the year of entry and exit as well as the origin of the firm. The origin is assigned by the previous positions of the founders. Firms that stem from the same source are grouped together. The group is named after the common ancestor. When the firm has several founders and at least one founder stems from an identifiable group of firms, its origin is assigned to this group. Otherwise, the firm is counted as an isolated formation.4

Spin-off processes in biochip industries in Jena and Berlin

Table 10.1

247

Number of firms and interviews

Firms Formations Total Interviewed Diversifiers Total Interviewed Organisational Environment

Jena

Berlin

5 3

9 6

5 5 6

11 9 3

The analysis of the learning mechanism first of all needs the identification of the variety creation that precedes a spin-off. The start of variety creation is marked as the point in time where the firm began to develop the respective competencies. Variety can be created internally, for example through learning by doing or by hiring of new skilled employees. Other forms are by interactive learning from other actors. When the knowledge is delivered by another regional organisation, the interaction is termed as regional. In conclusion, proof of the Klepper hypothesis would require a growth pattern where spin-offs can be traced back to few sources where competencies for spin-offs were generated solely within the incubator. A proof of the regional learning hypothesis would require that the variety upon which spin-offs build is developed during interaction with organisations in the same region. The necessary information was collected by qualitative, semi-structured interviews. The semi-structured interviews were conducted between 2007 and 2009 and complemented with interviews from an older study of Jena from 2002 (Menzel 2005). Interviewees were mostly representatives of the firms that are close to the technological development: in small firms the owner, in larger firms the project manager. All in all, 27 interviews with firms were done with information obtained about 23 firms. Table 10.1 shows the respective numbers of total firms in Jena and Berlin and numbers of interviews, classified in formations and diversifiers. In cases where diversifiers spun-out a firm, the interviews were assigned to the diversifying firm, as the variety of the spin-off base was created in the diversifying firm. The qualitative interviews served to retrace development of competencies. The study was complemented by nine interviews with experts from institutionalised networks or research institutes.

248

6

Emerging clusters

THE DEVELOPMENT OF THE BIOCHIP INDUSTRY IN JENA

Jena is one of the success stories of restructuring of the East German economy due to its technology orientated firm formations (Hassink and Wood 1998). In recent years, several biotech firms formed, with the first biochip firm in 1998. The basic sample of biochip firms in Jena contains ten firms: five diversifiers and five firms that already had respective competencies at their formation. In two cases, diversifiers spun out their biochiprelated divisions. There were two exits. Thus, eight firms had competencies in biochips in the year 2007. The firms are, or have been, active in different technological areas. One firm applies biochips for clinical diagnostics, two firms develop biochip platforms, two firms deal with microstructures and thus with the chip itself, two firms deal with the detection of the signals stemming from the chip. Three firms offer (or intend to offer) special inputs or services for biochips. Thus, the focus is on the hardware, rather than on the application of biochips. Most biochip firms can be traced back to two sources. The first source is the former Zeiss combine that was restructured during the reunification of Germany in 1990. Three firms have this background. The origin of the Zeiss combine dates back to 1846, when Carl Zeiss founded a workshop for optics and precision mechanics in Jena. Until World War II, the firm grew to be a major producer of optical precision instruments. In the German Democratic Republic, Carl Zeiss Jena became the headquarters of the combine for optics and the centre of the engineering of scientific instruments within the COMECON (Council for Mutual Economic Assistance) with a total of 70 000 employees, 27 000 of them located in Jena at the time of the reunification (Hessinger et al. 2000). After the fall of the Berlin Wall, the reprivatisation and restructuring of the combine and later of its successor, Jenoptik, resulted in about 100 spin-offs (Plattner 1997). Three other firms have their background in the research institutes of the scientific Beutenberg Campus. Its origin can be traced back to the establishment of the Schott-Zeiss-Institut in the year 1944, which had a focus on research and production of penicillin. In 1950, the research facilities of the Schott-Zeiss-Institut were relabelled as the Institute for Microbiology and Experimental Therapy (IMET). It later attained the status of a Central Institute (ZIMET) and became one of the largest scientific institutes for biological research in the GDR. Next to the ZIMET, physical research institutions developed on the campus and were subsumed in the Physical-Technical-Institute (PTI) in the 1980s (Renno 1993). After the reunification, the ZIMET was also divided into smaller units, with

Spin-off processes in biochip industries in Jena and Berlin

1. Phase

2. Phase

249

3. Phase

A B C D E F G H I J 1998 1999

2000

Entry and Exit: formation diversifier exit

Figure 10.2

Spin-off Spin-out

2001

2002

2003 2004

2005

2006

Business field

Origin:

full biochip

Zeiss

part biochip

Beutenberg

2007

other

Development of the biochip industry in Jena

the Hans-Knöll-Institute for Research of Natural Products (HKI) and the Institute of Molecular Biotechnology (IMB, today the Leibniz Institute for Age Research) as the largest parts. Additionally, the PTI became the Institute of Physical High Technologies (IPHT, today the Institute of Photonic Technology). As the different institutes mostly have the same ancestry and some formations were founded by researchers from several institutes, these institutes are taken together as having a single origin (Menzel 2005). Additionally, the institutes of the Beutenberg Campus were the main drivers behind the BioRegio initiative. One of the institutes, the IPHT, was the leader behind the second publicly financed support programme, the Jena Biochip Initiative. Four firms have different origins. The origin of one firm could not be detected, one was an old firm located outside Jena, one firm was a spin-off from the local university, and one firm was a branch of a foreign company. Figure 10.2 describes the genealogy of the biochip industry in Jena. The figure contains the year of entry and exit, how the firm entered the industry, if biochips are the main or a sub-field of the business and the origin of the firm. The development of the biochip industry in Jena is divided into three phases that each differ in their characteristics of entries.

250

Emerging clusters

First Phase: Formation of the Biochip Industry in Jena The biochip industry in Jena started in 1998. The beginning was marked by three firms A, B and C. Firm A was set up as a manufacturer of optical equipment in the year 1991. Its origin was a former division of the Zeiss combine. At about 1995, firm A intended to orientate itself towards new fields based on its competence in opto-electronics. Biotechnology was identified as one possibility. Almost in parallel, the BioRegio initiative started in the year 1995. The initiative facilitated new contacts to the local scientific institutes and the idea for a related product emerged. In 1998, the first concrete project was started together with a research group from a scientific institute of the Beutenberg Campus. This project was intended to acquire basic knowledge in the new field and to find the starting point for the development of its own product. During the start of the project, the research group together with other researchers from institutes at the Beutenberg Campus spun-off to form firm B. A later collaboration between A and B led to the first product of A and firm B was among the first customers. Firm C was connected to these developments. This firm was formed in 1999, also as a spin-off from an institute of the Beutenberg Campus and developed tools for physics laboratories. Not much later, the firm started the development and production of biochip-related products. Firms B and C are connected by personnel relations due to their common ancestry in the research institutes of the Beutenberg Campus. Within these relations, the idea emerged for firm C to enter the biochip field with its own products. Firm C acquired the required competencies by both learning from firm B and new employees, and became a supplier of B. Subsequently, the two diversifiers A and C also had common projects. At the end of the first phase in the year 2000, three firms were active in the biochip industry. Two of the firms were actually pulled into the field by company B. While they also had collaborations on the national or even global level, the main impetus to develop the new product lines came from regional collaborations. Second Phase: New Firms The second phase was marked by the heyday of the biochip industry. It started in 2000 and ended in 2002. During this phase, five new firms entered the biochip field in Jena. For four of the five firms, biochips were part of the business model from the beginning. One firm was formed by a founding team from the Beutenberg institutes and the university clinic. This is the only firm in the Jena sample that developed clinical diagnostic

Spin-off processes in biochip industries in Jena and Berlin

251

tests. One formation had its origin in the Zeiss network. Two formations stemmed from other sources. One of them was a branch of a multinational company. The reasons to locate the branch in Jena were the highly skilled employees that were available in Jena due to the expiration of the German Human Genome Project. The origin of the last firm could not be detected. Beside the formations, firm E diversified and acquired biochip-related knowledge. For this firm, biochips were only a new application of existing products and few learning processes were necessary to adapt the product. At the time of the diversification, there had already been some firms in Jena. Yet, a personal contact between the founder and a West German firm with more experience of West German biotech firms compared to the young firms in Jena led to the first contacts with West German firms and later firms in other European countries. Later, however, firm D became a supplier for several Jena firms. While the first phase was marked by regional learning which pulled new firms into the field, the second phase was characterised by firm formations. There may be two reasons for this. Biochip technology diffused fast and the boom of that time facilitated funding. The reason for the only diversification during this phase was relations with firms outside Jena. Third Phase: Restructurings and (Attenuated) Second Generation Growth The third phase started in 2002 with the crisis and consolidation in biotech in general and biochips in particular. One firm went bankrupt and one left Jena. Both exits were not part of the two groups that were based on Zeiss and the Beutenberg. Nevertheless, firms also acquired new competencies during this phase, such as Firm H with bioinformatics. Reasons for this were ongoing product development, the need for this competence and difficulties in outsourcing this task. The firm acquired the new knowledge through new employees. With the start of the Jena Biochip Initiative in 2005, two additional firms moved into biochips. Firm I used the initiative to start its first learning processes. Firm J had a Zeiss background and was a long-term collaboration partner of the IPHT as a driver behind the Jena Biochip Initiative. It participated in the network to get general knowledge about biochips, but also acquired a biochip firm located elsewhere in Germany. Additionally, several firms had organisational changes during this phase. Four firms were acquired. Three of these were exactly those firms that started the field. The diversifier A was acquired by another firm of the Zeiss network. As biotechnology was not part of the core competencies of the buyer, it subsequently spun out the biochip division of the acquired firm. Firm B was acquired by a large US-based diagnostics firm. The

252

Emerging clusters

division of firm C that was responsible for the development and production of biochip-related products was acquired by a German-based global player and became an independent subsidiary of this company. Firm D as the fourth firm was acquired by a Jena-based firm. The acquisitions in all cases contributed to strengthening the biochip-related competencies and to enlarging the production and research capacities of the firms. The subsequent spin-outs of the biochip-related businesses of the diversifiers as dedicated biochip firms led to a further enhancement of the biochip industry in Jena. In conclusion, the third phase was characterised by two distinct developments. The first were the organisational changes by acquisitions. All three firms that started the biochip industry in Jena were affected by these developments. These acquisitions led to two spin-outs. Therefore, the generation of variety by regional learning in the first phase resulted in second generation growth in the third phase, which yet was attenuated as the incubators left the field. The second development during this phase was the start of new regional learning processes, which were initiated by the Jena Biochip Initiative and which tracked two firms into the biochip industry. This process resembles the processes that took place in the first phase and led to spin-outs in the third phase. Yet, as the initiative comprised mostly firms outside Jena, these processes took place on a larger geographical scale with a looser connection to the firms and research institutes in Jena.

7

THE DEVELOPMENT OF THE BIOCHIP INDUSTRY IN BERLIN AND BRANDENBURG

The biochip industry in Berlin is considerably larger than in Jena. At the time of investigation, 20 firms had competencies in biochips in Berlin and the surrounding Brandenburg. Ten of them were diversifiers and ten already had competencies in biochips at the time of their formation. Like in Jena, most firms stem from two sources. The first sources are the biological and medical institutes of the GDR in Berlin-Buch, in the north of Berlin. Around 1900, several hospitals and clinics were located there to provide medical support for the fast-growing city. They were complemented by the Kaiser-Wilhelm-Institut5 für Gehirnforschung (institute for brain research) in 1930. In the GDR, Buch became a centre for biological and medical research with several institutes, among them the Central Institute for Molecular Biology (ZIM) as the largest (Bielka 1992, 90). After the reunification, the different institutes were merged to the MaxDelbrück-Centre for Molecular Medicine (MDC) and a research campus in medicine and biology was developed in Berlin-Buch. This research

Spin-off processes in biochip industries in Jena and Berlin

253

campus also comprises facilities of the Charité, one of the largest university clinics in Europe, which is located in Berlin-Mitte in the centre of the city. Four firms have this respective background. One firm stems from another GDR Institute, but due to tight connections to the Buch institutes was assigned as the fifth firm to this group. The second source is the Max-Planck Institute for Molecular Genetics (MPIMG), which was formed in 1964. This institute became a major source for the formation of the biochip industry in Berlin with the assignment of a new director and the reorientation from bacterial towards human genetics and automation in genomic research in 1994. Seven firms in the sample have this origin. An additional eight firms stem from diverse sources like other firms, other research institutes or local universities. The technological orientation of the firms can broadly be distinguished by their origin. The Buch firms mostly deal with techniques to detect biomaterials in a probe, mostly by ELISA detection. ELISA is a kind of predecessor of biochip detection, yet without the high degree of miniaturisation. The MPIMG firms mostly have a focus on high density biochips and automation processes. This focus refers to different stages of the production chain like bio-informatics, sequencing, spotting and diagnostic applications. Firms from origins other than Buch or the MPIMG deal with diverse topics like lab automation, assay development or development of diagnostic tests. Like in Jena, the biochip industry in Berlin was also supported by public support programmes. The first respective regional initiative was BioHytec, which was supported by the InnoRegio programme. The central actor behind this initiative was the Fraunhofer IBMT (Institute for Bio-Medical Research). It started in 1998 and has a technological focus on biochips, especially on upscaling of production processes and component integration. The researcher who built up the institute previously had been at the MDC in Berlin-Buch. This relation between Fraunhofer IBMT and MDC is also apparent within BioHytec, where both institutions are important supporters. A subsequent initiative named iPOC was supported by InnoProfile. This initiative aims to develop an integrated protein chip for point-of-care diagnostics. All participants in this network were already involved in BioHytec projects. Figure 10.3 shows the location of the firms in Berlin. Biochip firms are mainly located in three places. Four firms are in the biotech park in Berlin-Buch. Two of them had their origin at the medical and biological institutes there. Four firms are at prestigious locations in the centre of Berlin in Mitte, but also the Charité as important collaboration partner is located there. Only dedicated biochip firms are located in Mitte and three of them stem from the MPIMG. The third small concentration

254

Emerging clusters

Buch

Mitte

Entry by Formation Diversification Origin Buch MPIMG other

Adlershof

Brandenburg

Figure 10.3

Location of biochip firms in Berlin

is Adlershof in the south eastern part of the city with four firms likewise. Adlershof was a traditional location of research institutes. After the German reunification, Adlershof was planned as a ‘science city’ with a focus on physics and media and is the largest technology park in Germany today (Kulke 2008). Four additional firms are scattered in different places in the city. The remaining four firms are located in the surrounding Brandenburg area. First Phase: Formation of the Biochip Industry in Berlin The temporal development of the biochip industry in Berlin is illustrated in Figure 10.4. As in Jena, it can largely be divided into three phases. The first activities in Berlin in biochips were market by a singular event. Firm A moved into the biochip field quite early, namely around 1995. At that time, even the biotech industry in Berlin was scarcely in existence, let alone biochip firms. Firm A produced specialised inputs for a spotting device, which was developed by a firm in Western Germany for a larger pharmaceutical firm. Firm A applied this step for a more general move into producing instruments and machines for biotech firms, but did not develop its activities in biochips for several years. The first phase of considerable activity started with the formation of two firms in 1998. Firm B started developing spotting devices for high

Spin-off processes in biochip industries in Jena and Berlin

255

A B C D E F G H I J K L M N

O P Q

previously 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Entry and Exit: formation diversifier exit

Figure 10.4

Spin-off Spin-out

Business field

Origin:

full biochip

Zeiss

part biochip

Beutenberg other

Development of the biochip industry in Berlin

density chips and Firm C biomarkers for a diagnostic application. Later, firm D formed with the intention to develop software to analyse the data generated on biochips. Firm E exemplifies the only diversification into biochips during this phase. It was formed by the MPIMG and another research institution in Southern Germany to offer sequencing services for firms and research institutes. By establishing competencies in high density micro-arrays, the firm followed the technological frontier of its trajectory. The firm was organisationally connected to the MPIMG and therefore got considerable knowledge from this source as well as from another research institution outside of Berlin. In contrast to Jena, variety generation by regional learning could not be detected during this phase. Instead, the exploitation of the possibilities of a new technology marks this phase. Knowledge flows were strongly

256

Emerging clusters

connected to the MPIMG and used to define and exploit the emerging technological trajectory. Second Phase: BioHytec The second phase started in 2001. It was marked by seven firms that entered the field with projects that were supported by the BioHytec initiative. Most of these firms had their background in the research institutes of Berlin-Buch. All five firms in the sample that had this background entered the biochip field in this phase. The remaining two diversifiers had other origins. The only formation during this phase did research on biochips for cancer diagnostics. Beside these developments, firm A again started activities in the biochip field. The five Buch firms started their move into biochips either with a partner stemming back to the time of the GDR or with a contact established by a common third party, also known since the time of the GDR. In all these cases, there had been no formal collaboration before. Also the remaining two diversifiers with other backgrounds already knew their collaboration partners, but had no formal collaboration with them before. Therefore, the BioHytec initiative heavily influenced the exploitation of already existing ties. The diversifiers acquired new competencies from other regional partners. As part of BioHytec, the Fraunhofer IBMT was an important source of respective knowledge. Most learning processes were driven by the intention to diversify the product portfolio and to increase the application fields of their core product. However, some processes of variety generation contributed to generating a localised value chain around a biochip platform. Firm H had a central position in this respect. The competencies of the firm were development of antibodies and ELISA tests. In 1998, the idea emerged within the firm to apply the competencies of the firm on a much more miniaturised scale, namely micro-arrays. In the year 2001, with the start of BioHytec, the firm began the development of a microarray platform for different applications. This development was started together with firm G, which was a producer of diagnostic instruments. In this firm as well, the idea to develop a micro-array device for diagnostic application emerged several years earlier. Firms G and H had complementary competencies, but finally were connected by a common third party: a scientist that had both been at the research institutes in Berlin-Buch during the GDR and also was affiliated to the Fraunhofer IBMT. The fact that both firms are located in the same place facilitated daily interaction. The social division of labour between these two firms was deepened as parts that previously were outsourced to a firm outside the Berlin area were

Spin-off processes in biochip industries in Jena and Berlin

257

internalised by firm G. This task did not belong to the core competencies of firm G before. Later, firm N was also pulled into the field by firm H. Firm N also had a Buch background and the founders of firm M and H knew each other. Firm H tested a product of firm N which is required for molecule detection and had not been applied on biochips before, yet only with mediocre success. Both firms worked together to adjust the product for its application on biochips. Afterwards, firm N became a supplier for firm H and also to other producers of biochips. In conclusion, there has been a large degree of variety creation by regional learning that was induced by the BioHytec initiative. In three cases, which all have a Buch background, the generation of new products was integrated into regional production processes. Third Phase: Second Generation Growth The third phase started in 2003. It cannot be delimitated clearly from the second phase, as several processes overlap. The first process refers to the ongoing BioHytec initiative. The initiative supported projects until 2004 and pulled two additional firms into the field in its last year. In contrast to the previous phase, none of these two firms stemmed from the Buch institutes. This indicates that the network that started the initiative and was strongly connected to the Buch firms increasingly opened up for firms of other origins during its development. The main quality of the third phase is the spin-off process as a sign of second generation growth, which started in 2003. With one exception, these spin-offs had their origin in MPIMG firms. Firm B generated the first spin-off in 2003. This spin-off was the result of a personal relation between the founder of the spin-off and the manager of another Berlin-based biotech firm. Both firms had different origins. The manager demanded a product that was not produced by firm B and not part of its core field. In 2004, firm D generated the second spin-off. This spin-off resulted from collaborations of firm D with the Fraunhofer IBMT. In this case also, the spin-off stemmed from collaboration between organisations of different origins. Finally, firm E, which diversified into biochips earlier, was divided into two independent parts. Both firms had their core field in applications on a biochip basis.6 The only spin-off that had another origin than an MPIMG firm formed from firm M. Firm M spun out the division that dealt with biochips only shortly after it entered the field. With the move into biochips the firm followed its technological trajectory and the knowledge for the spin-off basis was created within the firm. Additionally, the spin-out was planned before the move into biochips. Other developments in the industry during this phase were a change of the business field

258

Emerging clusters

of firm L, whereby micro-arrays became only a minor focus, and the entry of firm O, which intended to develop a chip for cancer detection. While the second phase was marked by a large degree of regional learning process that resulted in many diversifiers, mostly by Buch firms, the third phase was characterised by second generation growth from the MPIMG firms. These spin-offs doubled the number of firms with their main business field in biochips. This second generation growth was based both on variety created within the firms and on variety generated by regional learning processes.

8

RESULTS

The biochip industries in both Jena and Berlin are mostly based upon long established regional technological competencies. Biological competencies are important in both cases. These competencies are connected to optics and precision engineering in Jena and to medicine in Berlin. The different ‘re-bundlings’ (Bathelt and Boggs 2003) of existing competencies in Jena and Berlin result in a different technological orientation of the two industries. The biochip industry in Jena has its focus on engineering and instrumentation and most firms are specialised suppliers. In contrast, the industry in Berlin is more application oriented, which reflects both the medical competencies of the institutes in Berlin-Buch and the Charité. Additionally, both cases resemble each other in their development phases. The industries were started by only few firms respectively. Most firms entered the industry in the second phase. These firms were mostly new formations in Jena and mostly diversifiers in Berlin. In the third phase, spin-off processes started (with one exception) based upon firms that were formed in the first phase. Thus, second generation growth took place in both cases. Yet, this process was more pronounced in Berlin than in Jena, where diversifiers spun-out their respective branches and left the industry. The analyses of the genealogies reveal that most spin-offs can be traced back to few sources. From the 27 firm formation and diversifications in both samples, 15 firms had their origin either in the Zeiss combine, the Beutenberg Campus, the MPIMG, or the institutes in Berlin-Buch. Firms from three of these four groups, that is, Zeiss, Beutenberg and MPIMG, within all ten formations were responsible for six of the seven spin-offs. Therefore, only three ancestries were responsible for nearly all spin-offs, while the two exits had other origins. This result strongly points toward the Klepper hypothesis that the crucial process of regional industry formation is birth and heredity, whereby the better firm creates more and better

Spin-off processes in biochip industries in Jena and Berlin

259

spin-offs, the cluster subsequently arises in the region with the best firm and externalities do not play a significant role. However, the qualitative analysis of the processes behind the spin-off processes qualifies this assumption. For three spin-offs, which also comprise the two spin-offs that resulted from the splitting firm, a particular relation that provided the necessary knowledge could not be detected. The competencies upon which the spin-offs were based were developed within the incubators and their regular network as the firms followed the technological frontier of their particular trajectory. The actual spin-offs resulted from restructuring in the parent firms. Yet, the other four spin-offs were based upon competencies that were created during interactions with other regional firms and organisations. The competencies for one spin-off were generated during interactions within the same group of firms. The other three spin-offs emerged through recombination of knowledge from organisations of different origin. The qualitative survey revealed the impact of regional learning on the generation of spin-offs, but not why it seems to be focussed on certain groups of firms. The reason for the one spin-off that resulted from interactions between firms of the same group might be explained by the concept of incubator networks (Menzel 2005; Henn 2008). Beside their relation to their direct incubator, spin-offs are also connected to other spin-offs from the same incubator. Due to pervasive spin-off processes, complex networks form with relations that are characterised by different kinds of proximity and therefore enable efficient recombination of existing knowledge. However, this explanation does not work for the three spin-offs that emerged through knowledge that was created between firms and organisations of different origin. This may be explained by network theories. These three spin-offs formed at structural holes (Burt 1992), where networks were only weakly connected. Burt (2004) shows that actors that bridge structural holes have a higher probability of ‘good ideas’, as these actors are ‘more familiar with alternative ways of thinking and behaving, which gives them more options to select from and synthesize’ (ibid. 349f). Also Obstfeld (2005) emphasises that loose networks, where different kinds of knowledge are connected, support the generation of radical innovations. Therefore, these interactions resulted in knowledge that deviated more strongly from established paths than knowledge generated by interactions within more closed networks. Studies on industry evolution indicate that exactly this kind of knowledge fits the emergence of new industries. When a technological trajectory (Dosi 1988) or a dominant design (Abernathy and Utterback 1978) has not manifested itself yet and there are only weak indicators of the direction of innovative activity, firms ‘have to bet on the future trajectory before it manifests its potential’ (Bresnahan et al. 2001).

260

Emerging clusters

The results indicate that knowledge necessary for this strategy is generated at structural holes. While these points could explain why regional interactions between networks generate knowledge that might result in spin-offs in new industries, graph theory could explain why these interactions focus on certain groups of firms. Two different mechanisms might be responsible for it. One mechanism is preferential attachment, whereby firms try to connect to particular firms, for example to firms with superior routines (Powell et al. 2005). This explanation would support the Klepper argument that firm routines have an impact on spin-offs, yet require frictions with other routines and forms of knowledge to form the basis for spin-offs. The other mechanism is the ‘rich getting richer’ rule. It assumes that new nodes try to connect to the most connected nodes, whose connectivity hence grows stronger than that of the remaining network (Albert and Barabasi 2002). In this case, firms connect to other firms not because of their superior routines, but because these firms represent the strongest nodes. This accounts for the described groups of firms. Their existent network positions might have been responsible for further connections. But these are just speculations and further research is necessary. In all cases, the exploitation of existent relations for the building of new connections took place in a regional context. Geographical proximity contributed to connect those nodes whose interaction might produce the kind of knowledge based upon which new firms in new industries form. This supports theoretical reasoning about how geographical proximity contributes to connect diverse actors (Grabher 2002; Malmberg and Maskell 2002; Menzel 2008a).

9

CONCLUSION

In new industries, new firms form in several places. Clusters arise, however, only in some locations. The reasons why a cluster emerges in a particular region are still unclear. Yet, cluster emergence seems to be connected to a particular growth pattern, namely spin-offs that form upon incumbent firms. This growth pattern is termed ‘second generation growth’ by Romanelli and Feldman (2006). The intention of this chapter was to shed some light on the processes that lead to second generation growth. Two mechanisms were opposed. The first was the Klepper argument that instead of externalities or agglomeration effects, birth and heredity are responsible for cluster emergence. The second mechanism considered the processes upon which the competencies on which spin-offs are based are developed. It was assumed that

Spin-off processes in biochip industries in Jena and Berlin

261

regional learning processes are responsible for it and thus the generation of spin-offs depends on the regional context. The biochip industries in Jena and Berlin were surveyed to reflect these two arguments. While the pattern of spin-offs in Jena and Berlin seems to support Klepper’s hypothesis, the qualitative analysis reveals that the competencies upon which the spin-offs are based are mostly created by interactions with other regional organisations. Yet, these interactions seem to be concentrated on groups of firms that stem from the same origin. The two cases of Jena and Berlin indicate that second generation growth requires a connection of the two discussed mechanisms. Firms with better routines require a particular environment where their routines are recombined with other kinds of knowledge. It is still too early to assess if the signs of emergence in Jena and Berlin will result in functioning clusters. The probability is quite high that the described developments are comparable to those that happened in so many places where a cluster did not form in the end. However, if the intensity of second generation growth is an indicator for the future development, which is the actual starting point of this chapter, some prognosis might be derived from the results. Due to the intensity of second generation growth, there is a certain probability for the Berlin biochip industry to become a growing and functioning cluster. In Jena, the second generation growth was attenuated, as it was accompanied by several exits. Even if single firms are successful, it is more probable that a distinct cluster will not emerge and firms orient themselves towards other related regional industries. The results of two small case studies are far from being generalisable. But they might indicate the direction of further research. Klepper (2007a, b) already assumed that other processes than inheritance of better routines might lead to the observed pattern, but these processes would be difficult to analyse. The in vivo analysis of the cluster emergence in Jena and Berlin shows that qualitative method can reveal these dynamics. Unfortunately, this approach is not applicable to long-term growth patterns or deeper investigations of (also failed) emergence of clusters further in the past. But further analyses might reveal how under the contemporary socioeconomic environment second generation growth takes place.

NOTES 1. For a critique of these approaches, see among others Pinch and Henry (1999) and Martin and Sunley (2006, ch. 6). 2. See Bachmann and Bachmann (2002) for a description. 3. See BMBF (2008) for a description. 4. It should be mentioned that also isolated formations can be part of a group, but this

262

Emerging clusters

group remains unrecognised if other firms do not belong to the biochip industries in Jena and Berlin. 5. The Kaiser-Wilhelm-Institutes were the predecessors of the Max-Planck-Institutes. 6. There were two additional spin-offs from firm B and C, but to other industries.

REFERENCES Abernathy, W.J. and Utterback, J.M. (1978): Patterns of Industrial Innovation. Technology Review 80(7), 40–47. Albert, R. and Barabasi, A. L. (2002): Statistical Mechanics of Complex Networks. Reviews of Modern Physics 74(1), 47–97. Arthur, W.B. (1994): Increasing Returns and Path Dependence in the Economy. Ann Arbor: University of Michigan Press. Audretsch, D.B. and Feldman, M.P. (1996): Innovative Clusters and the Industry Life Cycle. Review of Industrial Organization 11(2), 253–73. Bachmann J. and Bachmann L. (2002): BioChipNet – an Internet Database for the Microarray Community Screening – Trends in Drug Discovery 3, 24–6. Baptista, R. (2000): Do Innovations Diffuse Faster within Geographical Clusters? International Journal of Industrial Organization 18(3), 515–35. Bathelt, H. and Boggs, J.S. (2003): Towards a Reconceptualization of Regional Development Paths: Is Leipzig’s Media Cluster a Continuation of or a Rupture with the Past? Economic Geography 79(3), 265–93. Bianchi, G. (1998): Requiem for the Third Italy? Entrepreneurship and Regional Development 10(2), 93–116. Bielka, H. (1992): Beiträge Zur Geschichte Der Medizinisch-Biologischen Institute Berlin-Buch 1920–1992. Berlin: Max-Delbrück-Centrum für Molekulare Medizin (MDC). BMBF (2008): Die Biotechnologie-Branche 2008. www.biotechnologie.de. Boschma, R. and Wenting, R. (2007): The Spatial Evolution of the British Automobile Industry: Does Location Matter? Industrial and Corporate Change 16(2), 213–38. Brenner, T. (2005): Innovation and Cooperation During the Emergence of Local Industrial Clusters: An Empirical Study in Germany. European Planning Studies 13(6), 921–38. Bresnahan, T., Gambardella, A. and Saxenian, A. (2001): ‘Old Economy’ Inputs for ‘New Economy’ Outcomes: Cluster Formation in the New Silicon Valleys. Industrial and Corporate Change 10(4), 835–60. Bünstorf, G. and Fornahl, D. (2006): B2C – Bubble to Cluster: The Dot.com Boom, Spin-off Entrepreneurship, and Regional Industry Evolution. Papers on Economics and Evolution. Max-Planck Institute of Economics – Jena. 30p. Bünstorf, G. and Klepper, S. (2009): Heritage and Agglomeration: The Akron Tire Cluster Revisited. Economic Journal 119 (April), 705–33. Burt, R.S. (1992): Structural Holes. Cambridge MA: Harvard University Press. Burt, R.S. (2004): Structural Holes and Good Ideas. American Journal of Sociology 110(2), 349–99. Dohse, D. (2007): Cluster-Based Technology Policy: The German Experience. Industry and Innovation 14(1), 69–94.

Spin-off processes in biochip industries in Jena and Berlin

263

Dosi G. (1988): Sources, Procedures and Microeconomic Effects of Innovation. Journal of Economic Literature 26(3), 1120–71. Feldman, M. (2001): The Entrepreneurial Event Revisited: Firm Formation in a Regional Context. Industrial and Corporate Change 10(4), 861–91. Frenken, K. and Boschma, R.A. (2007): A Theoretical Framework for Evolutionary Economic Geography: Industrial Dynamics and Urban Growth as a Branching Process. Journal of Economic Geography 7(5), 635–49. Grabher, G. (2002): The Project Ecology of Advertising: Tasks, Talents and Teams. Regional Studies 36(3), 245–62. Hassink, R. and Wood, M. (1998): Geographic ‘Clustering’ in the German Opto Electronics Industry – Its Impact on Randd Collaboration and Innovation. Entrepreneurship and Regional Development 10(4), 277–96. Henn, S. (2008): Entstehung und Wachstumspotenziale regionaler Technologiecluster. Das Beispiel Nanotechnologie. Zeitschrift für Wirtschaftsgeographie 52(2/3), 95–113. Hessinger, P., Eichhorn, F., Feldhoff, J. and Schmidt, G. (2000): Fokus Und Balance: Aufbau Und Wachstum Industrieller Netzwerke Am Beispiel Von Vw Zwickau, Jenopotik Jena Und Schienenfahrzeugbau Sachsen-Anhalt. Wiesbaden: Westdeutscher Verlag. Jacobs, J. (1969): The Economy of Cities. New York: Vintage Books. Klepper, S. (2007a): Disagreements, Spinoffs, and the Evolution of Detroit as the Capital of the US Automobile Industry. Management Science 54(4), 616–31. Klepper, S. (2007b): The Evolution of Geographic Structures in New Industries. In K. Frenken (ed.), Applied Evolutionary Economics and Economic Geography, Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 69–92. Kogut, B. and Zander, U. (1992): Knowledge of the Firm, Combinative Capabilities, and the Replication of Technology. Organization Science 3(3), 383–97. Krugman, P. (1991): Geography and Trade. Cambridge, MA: MIT Press. Kulke, E. (2008): The Technology Park Berlin-Adlershof as an Example of Spatial Proximity in Regional Economic Policy. Zeitschrift für Wirtschaftsgeographie 52(4), 193–208. Longhi, C. (1999): Networks, Collective Learning and Technology Development in Innovative High Technology Regions: The Case of Sophia Antipolis. Regional Studies 33(4), 333–42. Malmberg, A. and Maskell, P. (2002): The Elusive Concept of Localization Economies: Towards a Knowledge-Based Theory of Spatial Clustering. Environment and Planning A 34(3), 429–49. Martin, R. and Sunley, P. (2003): Deconstructing Clusters: Chaotic Concept or Policy Panacea? Journal of Economic Geography 3(1), 5–35. Martin, R. and Sunley, P. (2006): Path Dependence and Regional Economic Evolution. Journal of Economic Geography 6(4), 395–437. Menzel, M.-P. (2005): Networks and Technologies in an Emerging Cluster: The Case of Bioinstruments in Jena. In C. Karlsson, B. Johansson and R.R. Stough (eds), Industrial Clusters and Inter-Firm Networks. Cheltenham, UK and Northampton, MA, USA: Edward Elgar pp. 413–49. Menzel, M.-P. (2008a): Dynamic Proximities – Changing Relations by Creating and Bridging Distances. Papers in Evolutionary Economic Geography 08-16. University of Utrecht. 27p. Menzel, M.-P. (2008b): Zufälle und Agglomerationseffekte bei der Clusterentstehung. Eine vergleichende Diskussion des Core-Periphery-Modells, des

264

Emerging clusters

Window-of-Locational Opportunity-Konzepts sowie stochastischer Ansätze. Zeitschrift für Wirtschaftsgeographie 52(2–3), 114–28. Menzel, M.-P. and Fornahl, D. (2007): Cluster Life Cycles – Dimensions and Rationales of Cluster Development. Jena Economic Research Papers. 2007–076, Jena, 45p. Nooteboom, B. (1999): Innovation, Learning and Industrial Organisation. Cambridge Journal of Economics 23(2), 127–50. Obstfeld, D. (2005): Social Networks, the Tertius Lungens and Orientation Involvement in Innovation. Administrative Science Quarterly 50(1), 100–130. Pinch, S. and Henry, N. (1999): Paul Krugman’s Geographical Economics, Industrial Clustering and the British Motor Sport Industry. Regional Studies 33(9), 815–27. Plattner, M. (1997): Vom Zeiss-Kombinat Zur Jenoptik-Holding – Niedergang und Neuanfang der Industrie am Beispiel Jena. in G. Meyer (ed.), Von der Plan- zur Marktwirtschaft – Wirtschafts- und sozialgeographische Entwicklungsprozesse in den neuen Bundesländern. Mainz: Universität Mainz. Porter, M.E. (2000): Location, Competition, and Economic Development: Local Clusters in a Global Economy. Economic Development Quarterly 14(1), 15–34. Powell, W.W., White, D.R., Koput, K.W. and Owen-Smith, J. (2005): Network Dynamics and Field Evolution: The Growth of Interorganizational Collaboration in the Life Sciences. American Journal of Sociology 110(4), 1132–205. Renno, D. (1993): Hinweise Auf Forschungslücken Zur Geschichte Ausseruniversitärer Forschungseinrichtungen in Jena. Jenaer Stadtgeschichtliche Beiträge – Academica & Studentica Jenensia e. V., Jena, 121–8. Romanelli, E. and Feldman M. (2006): Anatomy of Cluster Development: Emergence and Convergence in the US Human Biotherapeutics, 1976–2003. In P. Braunerhjelm and M. Feldman (eds), Cluster Genesis: Technology-Based Industrial Development. Oxford: Oxford University Press, pp. 87–112. Saxenian, A. (1994): Regional Advantage: Culture and Competition in Silicon Valley and Route 128. Cambridge: Harvard University Press. Storper, M. and Walker, R. (1989): The Capitalist Imperative: Territory, Technology, and Industrial Growth. Cambridge, MA: Basil Blackwell. Swann, P. and Prevezer, M. (1996): A Comparison of the Dynamics of Industrial Clustering in Computing and Biotechnology. Research Policy 25(7), 1139–57.

11.

The emergence and development of the Cambridge ink jet printing industry Elizabeth Garnsey, Erik Stam and Brychan Thomas

1

INTRODUCTION

Recent cluster studies have moved from a predominantly static approach to a more dynamic analysis of the emergence and development of clusters (Braunerhjelm and Feldman 2006; Menzel and Fornahl 2007). The activities of firms that make up a high-tech cluster are distinctive and it is only by understanding the processes through which constituent firms and clusters develop and mature that we can gain understanding of collective trends. We define a cluster as a local concentration of firms that have horizontal (ecological) and/or vertical (genealogical) relations. In this chapter we focus on the ink jet printing (IJP) industry in the Cambridge (UK) area to explore the nature of maturation of a local knowledge-based (hightech) cluster. The Cambridge region is well known as a high-tech centre (Garnsey and Heffernan 2005), made up of diverse clusters of mainly small knowledge-based firms. What makes IJP distinctive in the area is that it has no direct university lineage, involves advanced product-engineering, and has achieved international market reach by anticipating and responding to global demand, resulting in several relatively large firms. The local IJP industry more than doubled in size during the 1990s, while the overall local IT hardware sector hardly expanded in this period. We organise our theme in terms of genealogical and ecological issues. That is, we examine the lineage of firms with reference to their technologies, spin-out activity and location near Cambridge. We address ecological issues with reference to production chains, competition and new relations of ownership accompanying the maturation of the international industry. We show how renewal has accompanied recognition of the generic nature of ink jet technologies as a means of deposition of valuable substances including intelligent materials. 265

266

2

Emerging clusters

CLUSTER DYNAMICS

In what follows we use a literature review to delineate evolutionary processes in local clusters. While dynamic approaches on clusters1 are still rare, prior work is useful in pointing to ecological processes of interaction and genealogical processes of replication (Baum and Singh 1994). The genealogy of organizational evolution – the structures of organizational inheritance and speciation – can be traced through new firms that spin off from other organizations (see, for example, Garnsey et al. 2008). This process is very localized, as most new firms are sited in the region in which the founder has worked and/or lived (Stam 2007). Ecological processes involve interaction with other firms, suppliers and customers in value chains. This is the vertical dimension of clusters and involves, for example, input– output relations with customers and suppliers (Maskell 2005). The cluster may also support firms carrying out similar activities as competitors in the same product market or for the same pool of labour. Such processes of interaction are not always in proximity, like in local labour markets, but very often transcend the local. For specialised high-tech firms, competition in product markets is expected from outside the region rather than within the region. Firms may also have interaction with firms in other populations that have dissimilar but complementary capabilities and activities (see, for example, Richardson 1972). These ecological and genealogical processes contribute to the competence base of the region (see, for example, Lawson 1999), through processes of collective learning (Keeble et al. 1999). The literature has recognized three phases of cluster emergence. First, there is the more or less ‘random’ location of early entrants. This chance location of successful early entrants sets in motion a self-reinforcing mechanism. The second phase of cluster emergence is shaped by a spin-off process from successful early entrants (Arthur 1994; Klepper 2007). Third, there is the attraction of outside investments. The first phase involves random location of early entrants in the sense that the founding entrepreneurs just happen to be located there. Entrant firms are likely to be founded by local entrepreneurs originating from related industries or knowledge bases. Not all regions have the same probability of being the home region of an emerging cluster. The incubator organizations of these early entrants and of the emerging cluster can be firms, but also public research organizations. Universities and research laboratories often provide the initial knowledge base (including both scientific and technological knowledge and skilled labour) for the emergence and growth of entrepreneurial clusters (Braunerhjelm and Feldman 2006). Early entrants are more likely to be successful if originating from related local industries (Audia et al. 2006; Boschma and Wenting 2007).

The Cambridge ink jet printing industry

267

The second phase involves spin-off processes from some successful early entrants (see, for example, Arthur 1994 for a stylized model of agglomeration by spin-off). These successful early entrants may become ‘anchor firms’ (Feldman 2003; Lazerson and Lorenzoni 2005; Klepper 2007). Romanelli and Feldman (2006: 105) found in their study of biotech clusters in the US that the start-up of new firms by entrepreneurs from other biotech firms is critical to the overall growth of the cluster. Only those regional clusters that exhibit second-generation growth, that is, spin-offs created from the early entrants, grow to substantial size in comparison to other potential regional clusters. The Schumpeterian (1934) logic is that clusters grow when the knowledge and other resources created by the early firms are combined and recombined by entrepreneurs who originate from the early entrants. Second-generation activity stimulates new contributions when spinoffs avoid direct competition with the company of origin, as illustrated by spin-off firms of Acorn Computers in Cambridge (Garnsey and Heffernan 2005; Garnsey et al. 2008). In the third phase, investment and talent is attracted from outside the cluster. Investments may include multinational firms investing in the region, or entrepreneurs moving to the area to set up a new independent firm. In addition, ‘magnet organizations’ (Harrison et al. 2004) attract talented people from outside the locality. These may move to other firms or start up their own firm nearby later in their career. We look for evidence to confirm or challenge these predicted trends from our case history. Although we find this to be largely congruent with the three-phase model outlined above, evidence moves beyond these accounts and expansion is followed by further phases of maturation as well as renewal. We argue that these phases require attention in knowledge-based no less than in rustbelt regions to inform policy and practice.

3

RESEARCH DESIGN AND DATA

On the basis of the literature review and initial conceptual model, our research question focuses on the following: ‘How has firm spin-off led to the emergence of the local IJP industry and how has the industry matured locally?’ This is approached through the operationalisation of constructs that depict these processes. Genealogical processes are examined on the basis of evidence on firm spin-offs in the IJP sector and the accumulation of technological knowledge by firms. Ecological processes are examined through evidence concerning horizontal and vertical relations of firms in the IJP cluster. These constructs are operationalised on the basis of quantitative

268

Emerging clusters

Table 11.1

Case study IJP companies in the Cambridge area

Company Cambridge Consultants Ltd Domino Printing Sciences Linx Printing Technologies Xaar Group

Biodot Ltd

Xennia Inca Digital Inkski Ltd

Note:

Number of Year of employees* founding

SIC code

Core activities

213

1960

7310

Technology consulting

550

1978

3002

245

1987

3002

80

1990

7310

(4)

1994

3320

30

1996

7310

100 1

2000 2004

2956 7310

Coding and marking printers, laser marking Coding and marking printers, laser marking Development of DOD printing, manufacturing of industrial printheads Non-contact nanoliter and low microliter dispensing for the development and manufacture of Rapid Diagnostic test devices, Biosensors and BioChip Arrays Contract ink development, test equipment Digital printing Development of non-impact digital printing technology

* at the Cambridgeshire sites in 2006.

and qualitative data. Quantitative data is derived from the Cambridge University Engineering Department high-tech database, which includes all establishments in high-tech industries in Cambridgeshire in the period 1988–2006. Data on patents is used from the database of the European Patent Office. For in-depth information on key firms in the Cambridgeshire IJP industry, case study analysis was undertaken. Seven in-depth case studies were undertaken with IJP companies at different stages of development (Table 11.1). The cases selected cover the leading firms in the Cambridgeshire IJP industry (CCL, Domino, Linx, Xaar, Xennia, and Inca) and some spin-offs from these IJP firms in closely related industries (Biodot, Inkski). The research strategy was based on semi-structured interviews of senior level personnel and on direct observation during student projects, next to archival evidence, press reports and company websites. The study was undertaken over an extensive period – 1995–2008 – to address the problem of retrospective bias.

The Cambridge ink jet printing industry

269

Ink jet printing is the collective name for a variety of different techniques to generate droplets of ink, which are propelled towards a surface to produce a printed mark. These include continuous (binary, multi-level, greyscale), drop-on-demand (DOD)/valve jet (shutter, array) and impulse jet (piezo activated and chevron) ink jet printing techniques (Garnsey and Minshall, 2000, pp. 18–19). Drop-on-demand ink jet printing is a complex technology that converts full pages of electronic text and images into tens of millions of signals, via individual ink jet nozzles in the print head for reproduction. Ink jet printing embodies many different skills and technologies: digital image processing, micro-machine semiconductor processing, mechanical, control, and electronic system design, computational fluid dynamics, chemistry of ink and paper, and precision manufacturing. These technologies have been applied in a wide range of industries and markets that can benefit from the key features of ink jet printing which are that it provides: non-impact/contact process for printing; infinitely variable output on demand; and high speed and high resolution (Garnsey and Minshall 2000). In the late 1980s and early 1990s ink jet printing disrupted what was then the dominant design in desktop printing, dot matrix printing. The ink jet printing industry grew rapidly in the 1990s, with a 6-year average growth rate of 73.3 per cent in the period 1990–95 and an average annual growth rate of 14.4 per cent in the second half of the 1990s (Clymer and Asaba 2008). The overall IJP industry emerged in the 1970s, entered a growth phase in the 1990s, with the maturation phase setting during the late 1990s (see Figure 11.1). The ink jet printing industry is divided between the products developed for industrial applications (marking, labelling and coding for production lines), printing applications (commercial printing) and home and office equipment applications (desktop printing; for example provided by HP and Canon). In addition, selling inks can make a substantial contribution to revenues over the lifetime of a printer. The Cambridge firms specialised in the first two applications and markets, together with inks. Further to this there has been a recent shift to laser printing. In the next section we will describe the development of the IJP industry in the Cambridge region.

4

CAMBRIDGE INK JET PRINTING INDUSTRY: EMERGENCE, GROWTH AND MATURATION

The local inkjet printing industry can be traced to one organization, the technical design consultancy Cambridge Consultants Ltd (CCL). At the start of the 1970s CCL, a spin-off from the University of Cambridge, was

op vel De

Manufacturers introduced photo quality printers

Epson Stylus Color printer

per

2003

1999

iod

th ow Gr

iod

ity

tur Ma

per

1995

HP and Lexmark settle infringement suit with new cross-licensing aggreement

Seiko Epson files critical ink-jet piezo print head patents in the US

1991

iod

nt me

Hewlett-Packard DeskJet 1200C

Canon BJ-10e & 10v

Hewlett-Packard DeskJet

Emerging clusters

Hewlett-Packard ThinkJet

Siemens PT-80i introduced in Europe

270

per

Xaar’s basic patents for piezoelectric-shear print heads are filed

1987

1983 HP and Canon negotiate ink-jet cross-licensing agreement 1979

Source:

Canon files the first thermal ink jet patents in Europe and the US

Clymer and Asaba (2008: 140).

Figure 11.1

Three phases in the global ink jet printer industry

working on various continuous ink jet printing technologies funded by the chemical multinational ICI. CCL was contracted to develop ink jet technologies for printing textiles at high speed, over wide widths and in colour. ICI withdrew from this project a few years later on the advice of external consultants when it became clear that the level of complexity required to achieve their quality and cost targets had been underestimated (reflecting the very nascent phase of the IJP industry at that time). However, the project manager at CCL (Graeme Minto) saw the commercial potential in ink jet printing. Graeme Minto obtained support from CCL to spin out the technology in a new company founded in 1978: Domino Printing Sciences. Domino was an independent start-up which took over the intellectual property in the technology from ICI and CCL. In the 1980s the cluster was dominated by CCL and Domino Printing Sciences, which made up almost 100 per cent of the IJP employment in the region (see Figure 11.2). In the 1990s – the growth phase of the IJP industry – a number of further spin-offs occurred by former employees of CCL, partly motivated by the success of Domino. These included Linx, Videojet, Xaar, and Inca (see Figures 11.2 and 11.3). These firms have achieved global expansion on the basis of a set of related technologies. The local IJP industry more than doubled from the late 1980s to the early 2000s, reaching a size of more than 1300 employees. Employment in the

The Cambridge ink jet printing industry

271

1400 Inkski 1200

Imaje Inca Digital Printers

Employees

1000

Biodot Xaar Group Plc

800

Elmjet 600

Videojet

400

Linx Printing Technologies Domino Printing Sciences

200

CCL

0 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 Year

Source:

Cambridge University Engineering Department high-tech database.

Figure 11.2

Employment in the Cambridgeshire ink jet printing industry

local industry decreased somewhat following 2000, just like the overall global IJP industry that entered the maturity phase in the late 1990s (Clymer and Asaba 2008). The early growth of the local industry was stimulated by new European legislation relating to consumer information on food, drink and pharmaceutical packaging, and factory automation gathering momentum. Regulation led to rapid growth in the demand for production line labelling and coding equipment. Interstices in the wider IJP industry, where the market for home and office printers dominated, were created by this new regulation. Incumbent firms were stretched by the home and office printer markets and left a product identification market to new entrants (see, for example, Penrose 1959). The need for flexible systems for applying variable data at speed became critical as food, drink and pharmaceutical industries increased their reach into global markets where minor variations in national legislation necessitated differing information on packaging. The legislation was serendipitous for Domino; although they had not anticipated this development they exploited the potential demand that it represented. Genealogical Processes The lineage of the IJP companies in Cambridge is shown in Figure 11.3. Two ink jet printing companies not located in the Cambridge region are

272

Emerging clusters 2005 2000 1990 Inca 1980 Domino 1970 Xaar Biodot Cambridge Consultants Limited

Elmjet Linx

Inkski Willett

Xennia Imaje (France)

Figure 11.3

Lineage of ink jet companies in Cambridgeshire

Imaje in France and Willett in Corby (UK). The companies were dominant in international markets for non-impact product identification, which is a smaller market than the larger market for desktop ink jet printing. The group of Cambridge IJP firms formed organically after spinout from CCL or its descendants. CCL was a technical design/ contract research consultancy originating in the University of Cambridge. Interviews with founders of companies showed that the most prominent factor for the original choice of location in the Cambridge area was the unwillingness of the founders to relocate (see, for example, Stam 2007). This does not necessarily support the prediction of the model that the location of the cluster is initially random, since there are causal factors at work in the attractions of residence in a university city and in family ties obstructing mobility, leading to structured rather than random incentives to co-locate spin-off companies. Spin-off companies must continually innovate and develop new technology and products. The inherited technology and expertise lose importance as this occurs. We expected linkages between incubator organisation and spinoff company to weaken as the spin-off company creates its own value chain. We found this to be the case; for example Xaar had reduced contact with CCL and Xennia had weakened links with its company of origin, Domino.

The Cambridge ink jet printing industry

273

200 180

Number of Patents

160 140 120 100 80 60 40 20 0 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 Domino Printing Sciences Linx Printing Technologies Xaar Group Inca Digital Tonejet

Source:

European Patent Office.

Figure 11.4

Cumulative patents for the Cambridgeshire ink jet printing companies

In order to achieve reliability and keep ahead of competition, the IJP companies patented their innovations (Figure 11.4). The patent data provides a proxy measure of the steady growth of technical capability in the local cluster. Among ink jet printing patents and other related patents reported for the IJP companies 55 per cent are made up of ink jet printing patents and 45 per cent of patents in other or related areas. Five (71 per cent) of the firms are specializing in ink jet printing equipment patents. Two companies had most of their patents in other applications, for example in inks. Taking patents as a proxy measure of competences, we can infer cumulative development of skills and expertise. Ecological Processes There are no close formal links between the companies, which are largely in competition with each other. Interviews did reveal examples of informal knowledge transfer. Knowledge is also transferred through the work of the consultancy Pivotal Resources and by the Ink Jet Academy, a local training course for employees in the IJP and related industries. The movement of personnel between IJP firms is extensive, as measured by manager moves for example. For instance, amongst seven senior staff at Xennia,

274

Emerging clusters

five had previously worked for ink jet printing companies. Personnel have also moved to ink jet printing companies outside the Cambridge area. Much of the informal social contact has occurred amongst those employed in the Cambridge IJP companies through social networks. Formal relationships between the Cambridge firms using IJP technology have only emerged recently through the creation of an R&D consortium, as will be discussed later on. As markets expanded, competition became increasingly intense. Barriers to entry for competitor firms were not very high. Key competitors included Videojet (USA) and Hitachi (Japan). There was also entry into the industry from other sectors by well-resourced players including Danaher (US). Domino, Videojet (including Willett) and Linx were already competing directly in product identification markets. Inca is an original equipment manufacturer (OEM) and produces machines for different applications. Xaar produces print heads and Xennia’s activity involves inks, so these two firms operate at different stages in the supply chain. Partnerships in the area between complementary firms are predictably more common than among competing firms; Xaar, in particular, has relationships with a number of other companies. Cambridge firms are affected by competition from outside the region. Competition between two leading print head manufacturers and licensers, Xaar (Cambridge) and Spectra (now FUJIFILM Dimatix, USA) is an example of this. These firms started around the same time and competition between them has been a development driver. Industrial IJP is a small part of the international IJP industry and specialist suppliers have not been drawn to the area (Micropump being an exception). Manufacturing operations of the IJP companies are largely assembly operations with parts increasingly outsourced, apart from core technology such as the print head. A large number of suppliers is used by each firm and these encompass a wide range of different types of operation. The Cambridge IJP companies use both local and international suppliers. When Domino was founded, it had a policy of working closely with small suppliers in the region and they were able to upgrade local suppliers by passing on equipment to them and treating them as an extension of their own activities. Today suppliers are chosen for quality and product price rather than location. For example, Spectra (US) print heads are used by many ink jet companies instead of Xaar heads (Spectra heads produce higher resolution prints than Xaar, which is important for certain applications). This shift reflects the rise of low-cost global manufacturing centres in the Far East and improved communications. We were told in interviews that at the time of the founding of earlier ink jet companies, parts could not be sourced internationally because of quality issues and the difficulty of transfer of design and blueprint materials. The ease with which

The Cambridge ink jet printing industry

275

computer aided design (CAD) drawings can be transferred digitally has changed the supplier–customer relationship. Local suppliers, who lacked funds to invest in design capability, have not been able to upgrade to satisfy OEMs. Thus the volume of parts sourced locally has decreased over time although local suppliers remain important to the IJP companies and are often used for non-specialist and lower value parts, such as casings, printed circuit boards (PCBs) and small components; larger assemblies and specialist parts are sourced further away. Local suppliers are involved with a range of industries and their products tend to be generic. Suppliers of generic parts, such as metalwork for example, are located near to Cambridge, whereas ink suppliers are further away. Proximity is viewed as a benefit, particularly where development relationships are concerned, although its importance is outweighed by quality of supply. Companies generally look for local suppliers before looking further afield. From both the customer and supplier perspective, proximity is helpful, especially at the development stage, but may be superseded by cost considerations. These are usually assumed to relate to labour costs, but US and Japanese competitors do not enjoy lower labour costs as such. Where production is capital-intensive, the high cost of capital in the UK acts as a major disincentive to local production (IfM, 2005). As markets for IJP internationalized, the production chain of the Cambridge IJP firms became international. Initially there was a local production chain as local suppliers were supported and came to be shared by several firms, even those in competition with each other. Domino in particular helped local printed circuit boards and precision engineering suppliers to upgrade their performance and these contractors have then been able to help other customers in the area to upgrade their products and production processes. Although IJP clients have remained an important customer for local suppliers, sub-assemblies have come to be sourced internationally as the industry has matured. Supplier relations have emerged with firms in other countries in a global production ecosystem. Ink jet printing firms source jewels from Switzerland, pumps from the United States and precision components from many other areas. Market reach was extended through further innovation, in recognition of greater market needs. The Cambridge ink jet printing businesses realized that technologies initially used to provide time-dependent product information for the consumer could provide additional value to customers through further applications. They could be used to improve efficiency of production and distribution processes. Technologies used to apply ‘best before’ dates for food packaging could also be used to improve product traceability by printing batch information. As well as the products themselves, ink jet printing technologies were applied to packaging of drink,

276

Emerging clusters

food and pharmaceuticals. An example of this was the labelling of individual soft gelatine drug capsules using edible inks. There were also markets in the distribution of newspapers and magazines where inkjet marking, coding and labelling could be used. At the end of the production process for addressing, personalizing and coding purposes, prior to distribution, commercial printers were able to add variable information. The benefit of technologies for marking products within the factory was seen by firms operating many other production processes, varying from healthcare products to electronics. They were able to apply these in various ways. With new demands for more reliable and flexible systems, ink jet technologies were developed and adapted to cope with these. Thus the market for the application of ink jet technologies has differentiated into a number of sectors as the industry has matured. Firms have made a considerable effort to renew their technologies, as shown by the high level of patenting. However, though advances in technology have created potential for new markets these have also been used to improve efficiency in existing markets. The development of wide web drop-ondemand technologies is an example of this. These have had early success in basic label printing applications although they were anticipated to revolutionize printing and publications markets. The expansion to additional locations has allowed the cluster to reach international markets. Around 80 per cent of the Cambridge ink jet printing companies’ markets are abroad and survival depends on the ability to sell products world wide. Links with international customers and distributors have been developed and since applications for ink jet printing are very specific and customized, a local customer service base has been necessary for the customer interface. As a result there is investment in the global presence with companies having international offices. Biodot has moved its headquarters to the US, while Xaar, Videojet and Willett ceased to manufacture in the area and others such as Linx and Domino manufacture at international locations in addition to the Cambridge area. There is an extensive distributor network through which ink jet printing companies sell their products but none of these distributors are shared by the companies. Given that the IJP sector is not an industrial district with a local production web, but a cluster of firms related mainly by their origins, what are the advantages to them of being in a local cluster?2 Specialist suppliers are no longer a consideration. But a shared labour market pool and the transfer of tacit knowledge can be a major benefit of proximity. The people, skills and informal knowledge base in the area are a significant benefit of being located near to other IJP firms. The Cambridge address and the prestige

The Cambridge ink jet printing industry

277

IJP Research Centre, Dept. of Engineering (2005) 2000 Inca Digital Printers (acquired by Dainippon Screen)

1990 1980 1970

Domino Printing Sciences

Xaar

Biodot Cambridge Consultants Limited

Elmjet (acquired by Videojet later Danahar – closed)

Linx (acquired by Danahar) Inkski Willett Cambridge R&D Imaje (France) Office (acquired by Danahar)

Figure 11.5

Xennia (acquired by Ten Cate)

IJP firms acquired

associated have also been a consideration in retaining firms in the area. The amplification effects of co-location are apparent in multi-generational effects. Spin-off firms became the source of further spin-offs and attraction of entrepreneurs and firms from outside the region. This latter process is characteristic of the third phase of cluster dynamics: maturation. The Cambridge IJP firms were pioneers in targeting the industrial product identification market in which larger US and Japanese firms had not shown interest initially. The further expansion of these markets resulted in new and better resourced international competitors moving into this sector on a global basis. With increased competition and consolidation of the product identification sector of IJP, there has been a rash of mergers and acquisitions of local IJP firms in the Cambridge area (see Figure 11.5). In 2001 there were seven industrial ink jet printing companies operating in the Cambridge area. Currently (2009), only two of these companies (Domino and Xaar) have not been acquired. The Danaher Corporation had created their product identification division through takeovers of Videojet (2002), Willett (2003) and Linx (2005), and Dainippon Screen Manufacturing Company bought out Inca in June 2005. The Elmjet site, acquired by Videojet, was closed after a further acquisition by Danaher in 2002, while the manufacturing function of Xaar was relocated to Sweden after a merger. In 2008 Xennia was taken over by Ten Cate from

278

Emerging clusters

the Netherlands, leaving only two substantial independent players in the Cambridge IJP industry (Domino and Xaar). There are different views on implications of acquisition. Proponents point to local benefits of capital inflow and the introduction of managerial expertise and marketing power of larger international companies. Critics see negative impacts on the supplier network, personnel mobility, and attraction of business and personnel. On the other hand, post acquisition spin-off activity can be a source of innovation and shift into emerging areas via new applications. These new applications indicate a renewal of the cluster, a phase that is often not recognized in the literature on cluster dynamics. We will discuss this phase in the next section.

5 CAMBRIDGE INK JET PRINTING INDUSTRY: RENEWAL With time, it has been seen that jet printing technologies have very wide applications, well beyond printing and product identification. New applications are to some extent a response to the maturation of now traditional ink jet printing. Continuous IJP, once the foundational technology in the Cambridgeshire IJP industry, is now a mature technology. It was first threatened by drop-on-demand printing and more recently by laser technologies. Domino Printing Sciences has purchased several laser companies in order to gain expertise in a competing technology. The markets served by continuous ink jet printing are established and sustainable and all the companies involved have mitigated threats by developing competencies in the two emerging technologies. Drop-on-demand printing is a developing technology with rapid progress still being made in terms of performance and reliability giving rise to new markets through performance improvement. But these markets have differentiated needs and it is difficult for companies to diversify across applications in the face of resource constraints and significant market differences. If cluster companies are to compete across new markets, further R&D and strategic partnerships are likely to be pivotal. One important development in this respect is the initiative to set up the IJP research centre at the University of Cambridge. We saw that the stage model of cluster emergence and growth discussed in the first part of this chapter does not address the kind of renewal of local industry through a move from specialist to generic technological applications, on which we have found empirical evidence from Cambridge ink jet printing activities. Despite the loss of independence of several IJP firms and the move of many manufacturing operations away from the area, it is likely that

The Cambridge ink jet printing industry

279

Cambridge will continue as a centre for R&D. One reason for this is the localized accumulation of knowledge on ink jet technologies and related display technologies. Another is the initiative to set up the Inkjet Research Centre at the University of Cambridge in 2005. This centre has been funded by the UK government through the Engineering and Physical Sciences Research Council (EPSRC) and through industrial partners. Other universities in the UK are in the consortium and companies (including Sun Chemical, Sericol, Xaar, Fujifilm, Domino, Inca, Linx and CDT) involved originate mostly from the Cambridge area. The Inkjet Research Centre intends to develop understanding of the fundamental behaviour of liquids in environments presented in ink jet printing. The aim of the project is to reduce duplicated research in local companies and to spread the financial burden on the one hand, and to deepen the knowledge of ink jet performance and broadening knowledge about potential applications of the technology on the other hand. This represents the reconnection of the IJP cluster to the Engineering Department at the University of Cambridge from which it originated several decades earlier. This should help the cluster to be sustained into the future and to add cohesion and build a reputation. The development of the consortium has raised many challenges. For example, difficulties arose in the formation of the consortium around the acquisition of Linx by Danaher with Domino expressing concern that the takeover would mean that the development of the Research Centre would migrate to the US through Linx and the cluster would not realise the rewards. In order to address this issue contracts include that companies involved need to maintain or increase their level of R&D investment and deployment in the area – failure to do so leading to removal from the group. With the expansion of potential markets for ink jet products, speed and precision requirements have moved beyond the current state of the art. The need for new applications to open up new markets and recognition that ink jet printing has much wider potential than had yet been realised has resulted in re-involvement with the university after several decades when the IJP firms operated quite autonomously in the business sphere. Renewal has been achieved through recognition of the generic nature of IJP as a deposition technology for valuable materials including intelligent materials. Development of technology in this direction requires a return to basic research. Although ink jet has been a printing technology for more than 50 years, the processes involved are still not completely understood. With modern ink jet printing, droplets are generated at high speeds with fluids containing significant levels of particulates including metals leaving complex and inadequately explained behaviour. The shift from

280

Emerging clusters

manufacturing to a knowledge production focus is in line with developments in the Cambridgeshire high-tech cluster as a whole (Garnsey and Heffernan 2005). Ink jet technology can be used to develop markets for low cost electronic goods as instanced by disposable radio frequency identification (RFID) chips. This has required complementary research at the University’s Auto ID Centre.3 The Xennia case study provides an example of a new printer technique with applications in RFID. The development has involved Xennia with Carcio, a British company which has pioneered a way to print conductive inks with a digital inkjet printer. When Carcio was working on a way to customize cell phones, by printing personal images on the plastic bodies, it commissioned Xennia to find a way to print metallic inks with an ink jet printer. Xennia developed a novel approach which could also print on copper. Carcio and Xennia formed a joint venture called CIT (Conductive Inkjet Technology) to hold the patents for the new technology. The new printing technique could have an important impact on the RFID industry since it could replace the etching process used for making copper antennas which creates toxic waste and is expensive. Other new species of technology have emerged from new technology combinations, with firms spinning off from the Physics Cavendish Laboratories using jet printing of polymers for display markets, building on and related to the local knowledge, skills and competences in ink jet technologies. Cambridge Display Technology is the leading company in this area. The ink jet printing technologies are now being applied to ever more diverse areas. One application is in the production of printed circuit boards where the very precise delivery of conducting material onto an insulating substrate material is required (display technologies). Plastic Logic is a leading firm in this new sector, which has retained R&D activities in Cambridge although attracted to Dresden, Saxony by government subsidies.

6

CONCLUSIONS

The history of the Cambridge IJP industry reveals some of the mechanisms underlying the dynamic nature of local clusters. The first phase of the cluster started with the Cambridge University spin-off firm CCL getting involved in the IJP business in the early 1970s as one of its diversification efforts. However, this did not really take off, and it was only after one of its employees started a focussed IJP firm – Domino Printing Sciences – in 1978 that the local industry started to grow considerably. CCL remained an import incubator of new firms in the IJP industry, together with Domino.

The Cambridge ink jet printing industry

281

The second – growth – phase of the cluster took place in the 1990s, when a second and third generation of spin-offs emerged, with several substantial firms. A third phase was ushered in with a wave of acquisitions of local firms by foreign firms. We have seen that a specialist activity for which there is global demand cannot be immune to the forces of globalization. The very success of the sector has attracted acquirers looking to extend their own innovative portfolios with easy credit available to acquirers during the boom period. The case reveals the way genealogy can be the basis for a cluster through common origins and a shared labour market pool even where there are minimal local production relations. The local ecology evolved gradually, eventually to be dominated by the labour market pool that emerged in the area. This represents knowledgeable supply, but it only remains local because of career opportunities provided in the area by a number of firms, which offer promotion and skill extension possibilities. Although certain IJP firms have moved their production operations away from Cambridge to other countries in Europe and America, they have kept R&D operations at Cambridge. This has brought the cluster back into interaction with its original base, the university. Renewal is taking place through co-operative efforts between academics and IJP companies. This phase of renewal goes beyond the logic of current models of industry life cycles and cluster dynamics, which emphasize inevitable decline in the maturity phase of the industry and cluster respectively. In line with recent attempts to explain the long-term performance of regions (Bathelt, 2001; Glaeser 2005; Martin and Sunley 2006), we recognize that regions can ‘reinvent’ themselves and escape the inevitable decline of maturing industries by building new industries on the knowledge accumulated in older industries. Firms in the emerging sectors are drawing employees from IJP firms in the area. When IJP technologies were adopted by new entrants who developed advanced materials such as light emitting polymers (Cambridge Display Technology and Plastic Logic) they were able to hire professional staff with experience in the local IJP industry, demonstrating the role of job mobility in the diffusion of competence in the area. Recognition of the generic nature of jet-based technologies for purposes of deposition provides for a shift to a new technological alternative as old technologies wind down.

ACKNOWLEDGEMENTS This research was funded by the EPSRC Innovation and Productivity Grand Challenge EP/C534239/1. We would like to thank Max-Peter Menzel and two anonymous referees for their constructive comments.

282

Emerging clusters

NOTES 1. There is a related literature on the life cycles of industries (see Klepper 1997), but this has no explicit spatial dimension (implicitly it is perhaps the nation that is taken as the context of analysis: see Vernon 1966). 2. Willett, a spin out from CCL, was located in Corby, with only the R&D unit in Cambridge, suggesting that not all firms saw benefits in co-location. 3. See www.ifm.eng.cam.ac.uk/automation.

REFERENCES Arthur, W.B. (1994) Increasing Returns and Path Dependency in the Economy. Michigan: Michigan University Press. Audia, P.G., Freeman, J.H. and Reynolds, P. (2006) Organizational foundings in community context: instruments manufacturers and their inter-relationship with other organizations, Administrative Science Quarterly 51: 381–419. Bathelt, H. (2001) Regional competence and economic recovery: divergent growth paths in Boston’s high technology economy, Entrepreneurship and Regional Development 13(4): 287–314. Baum, J.A.C. and Singh, J.V. (1994) Evolutionary Dynamics of Organizations. New York: Oxford University Press. Boschma, R. and Wenting, R. (2007) Spatial evolution of the British automobile industry: Does location matter? Industrial and Corporate Change 16(2): 213–38. Braunerhjelm, P. and Feldman, M. (2006) Cluster Genesis. Technology-Based Industrial Development. Oxford: Oxford University Press. Clymer, N. and Asaba, S. (2008) A new approach for understanding dominant design: The case of the ink-jet printer, Journal of Engineering Technology Management 25: 137–56. Feldman, M. (2003) The locational dynamics of the US biotech industry: Knowledge externalities and the anchor hypothesis, Industry and Innovation 10(3): 311–29. Feng, V.Y. (2008), Micro Funding and the Development of Technology Startups: The Inski Ltd. Case, Report no. 08/12/14, Institute for Manufacturing, Department of Engineering, University of Cambridge. Garnsey, E. (2002) Case Study – BioDot/Polaris, Centre for Technology Management Institute for Manufacturing, University of Cambridge. Garnsey, E. and Heffernan, P. (2005) High tech clustering through spin out and attraction: The Cambridge case, Regional Studies 39(8): 1127–44. Garnsey, E. and Minshall, T. (2000) Ink Jet Printing and Domino Printing Sciences, Case Profile, Centre for Technology Management, Institute for Manufacturing, University of Cambridge. Garnsey E., Ferriani S. and Lorenzoni G. (2008) Speciation through entrepreneurial spin-off: The Acorn–ARM story, Research Policy 37: 210–24. Glaeser, E.L. (2005) Reinventing Boston: 1630–2003, Journal of Economic Geography 5: 119–53. Harrison, R.T., Cooper, S.Y. and Mason, C.M. (2004) Entrepreneurial activity and the dynamics of technology-based cluster development: The case of Ottawa, Urban Studies 41: 1045–70.

The Cambridge ink jet printing industry

283

IDCH (1993) International Directory of Company Histories, Vol. 7, Farmington Hills, MI: St. James Press. Institute for Manufacturing (IfM) (2005) Growth through HighValue Manufacturing, Engineering Department, University of Cambridge, Cambridge. Keeble, D., Lawson, C., Moore, B. and Wilkinson, F. (1999) Collective learning processes, networking and institutional thickness in the Cambridge region, Regional Studies 34(4): 319–32. Klepper, S. (1997) Industry life cycles, Industrial and Corporate Change 6(1): 145–81. Klepper, S. (2007) Disagreements, spinoffs, and the evolution of Detroit as the capital of the U.S. automobile industry, Management Science 53: 616–31. Lawson, C. (1999) Towards a competence theory of the region, Cambridge Journal of Economics, 23: 151–66. Lazerson, M.H. and Lorenzoni, G. (2005) The Firms that Feed Industrial Districts: A Return to the Italian Source. In S. Breschi and F. Malerba (eds), Clusters, Networks, and Innovation. Oxford: Oxford University Press. Martin, R. and Sunley, P. (2006) Path dependence and regional economic evolution. Journal of Economic Geography 6: 395–437. Maskell, P. (2005) Towards a Knowledge-based Theory of the Geographical Cluster. In S. Breschi and F. Malerba (eds), Clusters, Networks, and Innovation. Oxford: Oxford University Press, pp. 411–32. Menzel, M.-P. and Fornahl, D. (2007) Cluster Life Cycles – Dimensions and Rationales of Cluster Development. Jena Economic Research Paper No. 2007–076. Penrose, E.T. (1959) The Theory of the Growth of the Firm (3rd edition 1995). Oxford: Oxford University Press. Richardson, G.B. (1972) The organisation of industry, Economic Journal 82(327): 883–96. Romanelli, E. and Feldman, M.P. (2006) Anatomy of Cluster Development: The Case of U.S. Biopharmaceuticals, 1976–2002. In: P. Braunerhjelm and M. Feldman (eds) Cluster Genesis. Technology-Based Industrial Development. Oxford: Oxford University Press. Schumpeter, J.A. (1934) The Theory of Economic Development. Cambridge, MA: Harvard University Press. Stam, E. (2007) Why butterflies don’t leave. Locational behaviour of entrepreneurial firms, Economic Geography 83(1): 27–50. Vernon, R. (1966) International investment and international trade in the product cycle, Quarterly Journal of Economics 80: 190–207.

284

Emerging clusters

APPENDIX 11A Case Studies Case study findings for seven companies in the Cambridge ink jet printing industry are summarized below. Domino Printing Sciences1 Domino was built around products utilizing single jet ink jet technology, with CCL continuing to develop multi-jet versions. In its infancy Domino was supported and nurtured by CCL and development work continued to be undertaken for them following spin-off. A licence agreement allowed the company non-exclusive access to CCL know-how and patents enabling it to manufacture and sell ink jet systems. In return CCL received royalties on sales of all Domino products and CCL was entitled to grant licences to other companies if sales fell below a certain threshold (Domino was obliged to offer CCL ‘first refusal’ on development programmes for further ink jet products). The company went public in 1985 with an initial valuation of £26m rising to £40m and in the same year the company received the Queen’s Award for innovation. Domino’s machines consist of a collection of electronics which guide the ink nozzles and, because they are operated by electro-magnetic impulses and not by compressed air, the machine can be installed in a small metal or plastic cabinet. The essential elements of the machine are the microprocessors and their development and Domino has spent most of its time developing this part of the business rather than construction of machines. Continued leverage of connections with CCL was attempted and in 1987 a new company called Elmjet spun-off from CCL exploiting further technologies developed originally as part of the ICI project. This new spin-off aimed to design and manufacture wide web full colour printings and Domino’s chairman and founder, Graeme Minto, also acted as company chairman. Through Domino being an investor in Elmjet and the latter being contracted to develop new printing devices to complement and extend the range of Domino products, the two businesses were linked not only by their personnel. Domino now employs over 1000 people worldwide and continues to develop, sell and support industrial ink jet and laser printing systems for international packaging and printing markets and remains a major player in the industrial ink jet printing industry. Most of Domino’s activities are in product development and marketing (the company’s operations are concerned with product development and subsidiaries (specifically Domino Amjet) focus on marketing the product in Europe). Domino’s

The Cambridge ink jet printing industry

285

marketing orientation led it into a joint distribution deal with American Technologies Incorporate (ATI) and exploitation of A.B. Dicks’ European weaknesses of poor after-sales services. Domino has stressed quality and flexibility, instituted a rigorous programme of product development, concentrated on developing its research and development (R&D) capability, encouraged strong co-ordination among functions in R&D, product development and marketing, and nurtured strong co-operation from its distribution channels, especially in the US with ATI, and in Europe after the formation of Domino Amjet. There has been a recent shift to laser printing with Domino buying US and German laser firms. BioDot Ltd2 One of the smallest ink jet printing companies in the area is BioDot, which was founded as a spin-off from Domino in 1994. The company was formed following Philip Shaw, an employee of Domino Printing Sciences, taking redundancy, which concluded an agreement with Domino granting him access to IP relating to enzyme printing. It was agreed that the company would not produce ink jet printers and Domino would supply components and not produce enzyme printers. The company supplies non-contact nanoliter and low microliter dispensing equipment for the development and manufacture of BioChip Arrays, Biosensors and Rapid Diagnostic test devices. The core technology has descended from ink jet printing. The company was presented with many challenges when starting and there was a need to learn how to build an ink jet machine but advice fortunately came from former colleagues at Domino. One of the early orders came from Domino concerning manufacture of a special application machine allowing rapid changeover of inks. This helped the firm’s early cash flow situation. The operating costs were covered by revenues within the first years despite the challenges. In March 1994 Biodot commenced trading and included Selwyn Image a colleague at Domino who took a 5 per cent stake in the business, but left soon afterwards moving to Willet, another Cambridge based ink jet firm, since he found himself more suited to working in a large business. It was estimated by Philip Shaw that he needed £45 000 to start up the firm – he had £20 000 in redundancy compensation from Domino and it was found difficult to raise more money. This arose due to venture capitalists not being interested in investing small amounts; few venture capitalists were active at this time and there was reluctance by banks to invest in technology-based ventures. There was also the need for premises. Since rents were high in Cambridge the company found premises in Diddington, near St Neots. Although the contract did not include right of renewal the company stayed there without additional capital through having low-cost

286

Emerging clusters

premises and dealing with company matters in-house. Biodot moved its European headquarters from Huntingdon to a larger site in Chichester (West Sussex) in 2005. Its global headquarters is now located in Irvine (California, US). Linx Printing Technologies plc3 Linx Printing Technologies plc was founded in 1987 by two former members of the Cambridge Consultants ink jet team at Willett, to exploit legislation driven marking/coding opportunities in the UK and the European market. The company is involved particularly in the manufacture and marketing of ink jet and laser coders to a range of global industry sectors including food, beverage, pharmaceutical and industrial customers for ‘on-line’ variable information marking/coding. Linx Printing Technologies has been a developer of industrial coding and marking equipment, based on ink jet and laser technologies, used to print variable information such as serial numbers and ‘sell-by’ dates on products and product packaging at manufacturing line speeds. Following the company being founded (venture capital backed), main market flotation took place in 1992 and a FTSE fledgling stock in 2004. In 1999 the company acquired a Chinese distributor and in 2000 acquired Xymark, the laser company, from GSI-Lumonics. The company was acquired by the Danaher Corporation (USA)4 for £85m ($171 million) in 2005. It has about 718 employees worldwide and it had estimated revenues of £52.1m in 2004. The company has five locations with two sites in the UK (St Ives and Hull), one in France, one in the USA and two in China. By operating through direct subsidiaries, representing 50 per cent of total revenues, and a worldwide network of specialist distributors, Linx has served a global customer base in a wide range of manufacturing industries. For the Linx product range major overseas markets have included China, France, Germany, Italy, Japan, Spain and the USA. There are 350 to 400 employees in a Chinese factory and R&D is in Cambridge with manufacturing. The company spends 7 per cent of sales on in-house R&D. Xaar Group plc5 The Xaar Group plc spun-off from CCL in 1990 and although the initial business plan was for licensing only this has now been transformed to manufacturing since licensing was not sustainable. The main sector of work is printing, ink jet, wide format graphics (posters), involving the design and manufacture of ink jet print heads. It is a world leader in ink jet print heads design and manufacture. For the Xaar Group plc some 20 per cent of turnover goes into R&D covering technology which is even higher

The Cambridge ink jet printing industry

287

than the figure given by Inca Digital Printings Ltd. Following foundation in 1990 with £1m venture capital (VC) money, the first licence was sold in 1991, followed by a second round of VC in 1992 (approximately £1.5m), and private placement in 1996 which raised £12m. There was flotation in 1997 on the London Stock Exchange (LSE), which raised £10m, and in 1999 the company bought MIT (an ex-IBM licensee) and established manufacturing in Sweden. Whereas there were four employees in 1990 by mid 2004 there were about 250 employees with revenues of £30m in 2004. There are two locations with prototyping in Cambridge and volume manufacturing in Sweden, and four sales offices with two in China and one each in Japan and the USA. Competition is mainly from own licensees and the main challenge is to expand into new markets. The main market is wide format graphics, with Chinese machinery makers dominant, and coding and marking was becoming significant in 2004. Over the whole period the intellectual property rights (IPR) portfolio was continuously developed. Original strategic intentions were to produce a dominant digital printing technology and this remained the same in 2005. For the expansion of the market for digital printing active business development and the promotion of new initiatives through joint ventures has been started. Xaar has found that bootstrapping from an R&D company to a volume manufacturer is not easy and conservatism in the market place has been an impediment, which Xaar has attempted to address. Xennia Technology Ltd6 Xennia was founded in 1996 by Alan Hudd, ex- Domino ink and R&D group leader, who saw an opportunity in industrial ink jet from the drop-on-demand (DOD) techniques that were being developed. The company was founded to provide ink formulation for DOD, although the background of the founder was continuous ink jet. The company is in the industrial ink jet, chemistry layered integrator sector and its activities include new solutions for manufacturing companies, starting from fluids to provide solutions in hardware and software for specific applications. It is a world leader in drop-on-demand industrial ink jet technology, and provides its customers a one-stop shop for customized solutions for inks, hardware and software as a total package. The firm has a breadth and knowledge of all print head types. Xennia is independent and has specific expertise around ‘difficult’ materials in ink jet printing. These include dense or large inorganic materials like metals, phosphors, pigments, biomedical fluids, structural scaffold materials, conductive inks and materials for displays. The company is active in electronics, resistive, conductive displays, biomedical reagents,

288

Emerging clusters

enzymes, DNA materials for forensics, diagnostics for pregnancy tests, product decoration for mobile phones, packaging and coatings (optical or protective). An interesting development involving Xennia has been with Carcio, a British company which has pioneered a way to print conductive inks with a digital ink jet printer. When Carcio was working on a way to customize cell phones, by printing personal images on the plastic bodies, it commissioned Xennia to find a way to print metallic inks with an inkjet printer. Xennia developed a novel approach which could also print on copper. Carcio and Xennia formed a joint venture called CIT (Conductive Inkjet Technology) to hold the patents for the new technology. The new printing technique could have an important impact on the Radio Frequency Identification (RFID) industry since it could replace the etching process used for making copper antennas which creates toxic waste and is expensive. The firm employed 30 staff members and had an estimated value of £3.5m in 2004. Xennia has grown organically without capital inputs from investors. The firm had one site at Royston. Although in 2005 production was not important, by 2006/7 it was considered to be significant with relocation to Stevenage to facilitate manufacturing and accommodate the growth in the number of employees. The company is not interested in products for markets but in delivering customized solutions for specific customers. One of the major obstacles has been in recruiting high-skilled foreign workers, due to government regulations. Competition is in the USA and there are companies that are customers, partners and competitors at the same time. Therefore relationships are complex where Xennia competes and co-operates simultaneously. All Xennia’s activities involve R&D and are paid for by clients and the company. Most of the personnel are involved in R&D, which is mainly specific development work for clients and contract research. Inca Digital Printers Ltd7 Inca Digital Printers Ltd was founded in 2000 by Will Eve and Bill Baxter from Cambridge Consultants Ltd. The business idea was to sell high end assembled printers through ink distributors, while retaining an excellent set of engineers to build machines. The founders believed the ultimate selling point was in the ‘art’ of assembly of super fast, efficient wide format machines. They regarded it as an art, since empirical methods were still used, rather than fluid mechanical mathematical models, and since engineering and assembly are reliant upon the jetting of inks, they too are an ‘art’ in this case. The company is in the industrial ink jet printing sector, in particular wide format, and flatbed machines and it is a world leader in flatbed printing for the signage market. R&D roots have continued

The Cambridge ink jet printing industry

289

to play an important role in the business. The successful combination of R&D with commercial awareness explains the success that the company has already had. Its strength is in the core technology for industrial ink jet printers and partners in its markets help it define what customers need to take the product to market. It uses existing equipment (for handling the product into and out of its printer) so that it can supply core print engines to its OEM partners. When the founders were at CCL, customers enquired if it was possible to print packaging at the end of production lines. Following this a sample printer was made and it was exhibited at Ipex 1998, when it became obvious that there was a clear opportunity to develop machines for the display and signage markets. Inca Digital Printers began trading in 2000 and it progressed through the normal rounds of private venture capital finance to 2004. It has around 100 employees and it had estimated sales of £18m in 2004. The company has one main site located in Cambridge. The original strategic intentions were to access the market though a distributor while retaining excellent engineering staff. The distributor was an ink formulation and sales company since consumables companies have good access to customers. Inca does not rely on IPR to protect and build market share since it takes out patents where useful but it always underplays them. Slightly less than 14 per cent of sales are spent on R&D, with about 30 per cent of Inca Digital’s staff working in R&D. Inkski Ltd8 Inkski Ltd was founded in March 2004 by Dr Daniel Hall who, following his PhD degree in Computing Science at the University of Cambridge, had the idea of designing an innovative digital print head which was initiated by his observation on ink drop ejection. In early 2004 Daniel Hall observed that ink drops can be transported in, and then ejected from, an immiscible carrier liquid, with the carrier liquid imparting all the necessary momentum and direction to the transported ink drop. From this simple observation, the ideas behind Inkski’s technology evolved, and with help from contacts in the University of Cambridge Cavendish Laboratory, and initial funding from Providence Investment, Inkski was set up to start the formal development and exploration of the technology. The company has received venture capital and R&D grants in multiple rounds. When the company was founded Daniel Hall held a 75 per cent equity stake and venture funding of £25 000 from Providence Investment Company representing a 25 per cent stake of the business. It then experienced another three rounds of venture funding (by institutional investors, corporate investors (Xaar) and Cambridge business angel investors), bringing the total institutional investment to £635 000, until the most recent funding in

290

Emerging clusters

2007. The largest external investors in the company have been Providence Investment Company and Xaar plc, with 26 per cent and 9 per cent of the business respectively. These investments enabled further development of Inkski’s unique Light Initiated Liquid Output (LILO) technology and protecting this intellectual property with patents, without resources generated from production. By early 2005, a lab/workshop space had been established in a light industrial unit and with a laser module installed. By late 2006 Inkski started testing its system with a pico-second laser. By late 2007 the company demonstrated the controlled ejection of conventional black pigmented ink onto a paper substrate. Towards the end of 2007 the company had four patents covering its technology and intends to apply for more as a result of further research and development. The patent plan had delayed the pace to scale up as well as the progress of prototypes. The company contacted a German university with a technology platform to help accelerate the production of prototypes. Inkski received Department of Trade and Industry (DTI) funded R&D grants in 2005 totalling £60 000 through EEDA (East of England Development Agency) over a 9-month period. Inkski Ltd’s technology has attracted the interest of a number of players in the ink jet printing industry which has helped the company to build a collaborative partnership and commercial contacts with companies such as FUJIFILM and ManRoland. Since then the company has faced challenges in its technology development and target market, both of which have restricted its attractiveness to micro funds investors and potential customers. In relation to the company’s evolution and analysis of its outlook, key breakthrough and demonstration of technology is considered to be the most important driver of future funding and long-term success of the business.

NOTES 1. 2. 3. 4.

This case is largely based on Garnsey and Minshall (2000). This case is largely based on Garnsey (2002). This case is largely based on IfM (2005). Danaher (www.danaher.com) has a carefully considered strategy of acquisition centred around the purchase of companies that have a ‘high performance potential’. They also acquire companies with well-known trademarked brands, high market shares, a reputation for innovative technology, and extensive distribution channels on which to build (IDCH 1993). The three main qualities they seek in acquisition targets are strong brands, market leadership and proprietary technology. Revenues increased from $300 million to $1 billion within a decade and by 2004 the company was approaching $7 billion and averaging dozens of acquisitions a year. The company has 35 000 employees (17 000 in the US) and international sales from acquisitions (a total of 47 companies had been acquired for $3.4 billion).

The Cambridge ink jet printing industry 5. 6. 7. 8.

This case is largely based on IfM (2005). This case is largely based on IfM (2005). This case is largely based on IfM (2005). This case is largely based on Feng (2008).

291

PART IV

Cluster emergence and emergence of cluster politics

12.

Neither planned nor by chance: how knowledge-intensive clusters emerge Rolf Sternberg

1

INTRODUCTION1

‘Silicon Valley is probably the only place on earth not trying to copy Silicon Valley’ (Robert Metcalfe 1998, inventor of Ethernet and founder of 3Com). Many policymakers and even some academic scholars try to make the public believe that the success of Silicon Valley can be replicated if certain policy ingredients and instruments are given. According to this chapter such a ‘recipe’ (Bresnahan and Gambardella 2004b) perspective is based on a serious (and costly) misinterpretation of the reasons for the genesis and later growth not only of Silicon Valley, but of other knowledge-intensive regional clusters in the US and elsewhere, too. It seems necessary to interpret the high-tech cluster phenomenon in the light of these reasons specific to the local and national conditions given. Too many policymakers who try to create regional–sectoral clusters are blinded by the recent performance of well-known international clusters like the ‘Silicon Valley’ in California. They ignore the concrete and very location-specific processes and framework conditions in these regions before they reached their take-off phase that took place several decades ago. The determinants relevant during the emergence of a cluster may differ significantly from the determinants that influence the recent growth of these regions (Sternberg 1998; Bresnahan and Gambardella 2004b). Policymakers who want to create clusters now have much more to learn from the origins of current clusters than from its actual performance. Given the fact that most of the cluster initiatives in modern economies are based upon (and have to be based upon) knowledge-intensive industries, products, services and firms, it seems to be obvious that the learning effects of these policymakers are larger the more knowledge-intensive the older clusters are they want to learn from (see Hospers and Beugelsdijk 2002). Thus, regional high-tech clusters in industrialized countries, often named 295

296

Emerging clusters

high-tech regions, and their economic history may serve as highly important case studies. Empirical findings in this comparative chapter are based on expert surveys and statistical data from ten regions in the US, the UK, Japan, France and Germany. The regions considered are the so-called ‘Silicon Valley’, the Greater Boston region and the ‘Research Triangle/NC’ in the US, the ‘Western Crescent’ west of London and the Cambridge region in the UK, the Grenoble region, the ‘Cité Scientifique de Paris-Sud’ and Sophia Antipolis close to Nice in France, the Kyushu island in Japan and the Munich region in Germany. This chapter follows a qualitative approach, that is, empirical evidence is – beside in-depth-study of the literature on the historical origins of the high-tech clusters – based upon more than 50 interviews with experts living in and working about the ten regions analysed in this chapter. The creation of a knowledge-intensive regional cluster is interpreted as a process specific in time and space. Thus, theories explaining regional economic growth and development might be helpful. The focus of this chapter is on the origins (but not on the later growth) of emerging high-tech clusters. Thus, the majority of the literature was published quite a number of years ago. One may learn much more from the origins of current high-tech clusters (that is, former nascent clusters) than from their performance today. In that sense my approach is similar to the one Bresnahan and Gambardella (2004a, 2) have chosen: the aim is to learn the deep similarities across places that look quite different. Regional–sectoral knowledge-intensive clusters are defined as a group of industries characterized by above average share of R&D inputs and outputs (for example, patent intensity, R&D employment, R&D expenditure, new knowledge-intensive products) and by a significant degree of both spatial concentration (related to the overall national level) and intraregional cooperation. While this understanding of a high-tech cluster differs slightly from what is meant by high-tech regions during the 1980s and 1990s (Malecki 1991; Hall and Markusen 1985) there is a significant overlapping between both phenomena in terms of determinants of their emergence, local inertia and others. However, a high-tech region may be the location of several knowledge-intensive clusters and not just one. The chapter is structured as follows. In section 2 a selection of theoretical approaches to explain the genesis of knowledge-intensive clusters is presented. In section 3 the case-study regions are briefly described and defined. Section 4 tries to explain the specific reasons for the genesis of each of the ten high-tech clusters. The focus of section 5 is the role of national and regional policies in the emergence of regional–sectoral clusters. In the

How knowledge-intensive clusters emerge

297

final section, the chapter sheds light on the opportunities and threats of a cluster-based strategy of regional development.

2

THEORIZING THE EMERGENCE OF KNOWLEDGE-INTENSIVE CLUSTERS2

Despite many recent publications on the theory of knowledge-intensive clusters (for example, Malmberg and Maskell 2002) a generally accepted cluster theory does not exist. Given the numerous different interpretations of a cluster several scholars (for example Martin and Sunley 2003) criticize the cluster concept as a ‘chaotic’ one. Given these theoretical problems with clusters I will try to learn from some of the theories of knowledgebased regional economic development and growth created in the recent three decades. As there is a significant overlapping of knowledge-intensive regional–sectoral clusters and high-tech clusters we may gain important insights for the emergence of those clusters by searching for the reasons of the genesis (not the later growth) of high-tech clusters. Some of the concepts explaining the genesis of these regions are less disputed than the cluster theories although the explanations are more eclectic by nature than showing characteristics of an overall accepted grand theory. From an empirical perspective some of the regional growth theories have another advantage: they are less hermeneutic than most of the approaches to explain clusters (see Benneworth and Henry 2004). Additionally I will consider an explicitly evolutionary approach, namely the path creation and path dependency concept intensively discussed by Martin and Sunley (2006), Garud and Karnøe (2001) and others. According to the protagonists of the theory of flexible specialization, the emergence of industrial regions is essentially due to the change in forms of organization of new industries (see Piore and Sabel 1984). Hence, industries are able to change the character of ‘their’ region to meet their specific requirements. Consequently, it is not existing spatial characteristics that are responsible for the development of a region but the growth industries themselves. This approach explains the emergence of new industrial regions with the end of an old accumulation regime (Postfordism toward the end of the 1970s) which opens ‘windows of locational opportunities’ to the fast-growing R&D-intensive industries of that era; they utilize these ‘windows’ to make largely arbitrary locational decisions. As a result, new industrial regions or districts emerge far away from the centres of the old regime. Further growth of the new industrial regions is brought about by a process of selective clustering in which some of the new regions grow rapidly, others slowly or not at all. The key process during this phase is

298

Emerging clusters

vertical disintegration. The resulting external ‘economies of scale’ are best realized in those sectors that form the basis of the region. In other words: in order to avoid an excessive rise of the distance-dependent transaction costs of vertical disintegration the growing industry pursues a strategy of spatial agglomeration of production stages previously removed from the region, which leads to re-agglomeration. In a final stage of this dynamic evolutionary model of industrial development paths a dispersion takes place accompanied by the development of growth peripheries. The ideally typical development process ends in a radical shift of core regions as a consequence, for instance, of drastic changes in demand (Storper and Walker 1989). Similarities with the more recent evolutionary stream of economic geography (see Boschma and Martin 2007) are obvious. In his complete works dealing with this subject matter, Scott (1988) cited Silicon Valley, the Cité scientifique de Paris Sud and ‘Third Italy’ as proof of his theory. According to the milieu theory bearing the mark of the francophone GREMI (Groupe de recherche européen sur les milieux innovateurs) group, innovative enterprises are the result of a collective, dynamic process set in motion by many individuals and institutions in a region forming a network of synergy-producing linkages. Hence, the milieu of a region results from interaction between enterprises, institutions and labour, all of which, by way of collective, co-operative learning, help reduce the inherent insecurity during a period of changing technological paradigms. This kind of learning from and with one another is especially enhanced by the mobility of labour, by supply linkages and in the form of face-toface contacts which are facilitated by spatial proximity. Small enterprises profit most from an integration into regional networks (Aydalot 1986; Crevoisier 2004; Camagni and Maillat 2006). The concept of networks has been adapted as an integral part of the milieu theory and, consequently, gained importance as a means of explaining regional innovation processes (Camagni 1991; Giuliani 2007). The majority of empirical milieu studies is confined to regions characterized by innovative networks. Path dependency and path creation are two more recent, interrelated concepts that have been adapted by evolutionary economists and evolutionary economic geographers in order to develop new ideas on regional (and non-regional) causes of regional development, regional economic growth and disparities in the latter processes between regions (see Martin and Sunley 2006). Three different views of path dependence may be distinguished according to Martin and Sunley (2006, 400). Path dependence as technological lock-in means that specific technological fields are locked onto a trajectory. Path dependency as dynamic increasing returns argue that these returns produce positive feedback effects due to various

How knowledge-intensive clusters emerge

299

externalities and learning mechanisms. Finally path dependency as institutional hysteresis means the tendency of institutions to reproduce themselves over time due to socio-economic action. From a regional perspective various locally-bound sources may create regional path dependency. These sources include natural resources, sunk costs of local assets and infrastructures, local external economies of industrial specialization, regional technological lock-in, agglomeration economies, region-specific institutions and cultural traditions as well as interregional linkages (see ibid., 412). While regional path creation to a degree depends on regional path dependencies, of course, the first is much more relevant for the purpose of this chapter. It is especially relevant whether path-dependent regional models have the power to explain the emergence of new knowledge-based clusters. The main argument of most of these models have to do with chance or historical accidents, and most of these events are small, not radical. However, they may have strong and long-lasting effects for regional development. Martin and Sunley (2006, 424ff) distinguish between three approaches of regional path creation: 1. 2.

3.

Path creation as random, that is, as the result of a historical accident. Path creation as a mixture of limiting conditions and chance events, that is, the windows-of-locational-opportunity argument stressed by Scott and Storper (1987), Boschma and Frenken (2003) and others. Path creation interpreted as the dependence between successive (regional) paths, as shown by Zook (2005) for the emergence of the Silicon Valley’s internet industry as a result of the pre-existing venture capital industry in this very region.

The path creation approach, like the evolutionary school of economics as a whole, as the most recent of the approaches presented here, so far suffers from a relative lack of solid empirical studies on specific regions when the reasons for their emergence are considered. None of the approaches described before identifies directly the soughtafter determining factor of an emerging high-tech cluster. However, each of the theories more or less consider some ingredients a region should possess that make cluster emergence at least potential opportunity. Some of these ingredients are related to government policies (national ones, regional ones, with or without regional goals). Others are related to the location factors of new knowledge-based firms like the availability of skilled labour, the proximity to R&D laboratories and universities, or the availability of venture capital. While there are significant differences between the theoretical approaches

300

Emerging clusters

described earlier, they may help to develop a list of determinants which describe necessary or even sufficient characteristics of several or at least of one concrete nascent cluster. A first version of these determinants was discussed during the expert interviews and partly modified afterwards. See Appendix 12A for a list of these experts. All of them have worked and published on high-tech clusters and most of them are also experts for one or several of the high-tech clusters presented in this chapter. The determinants can broadly be assigned into two groups, namely demand factors and supply factors: 1.

2.

Demand factors: ● Private demand for technology-intensive new products ● Public demand for technology-intensive new products (especially military demand) Supply factors: ● Amenities (environment, culture, living conditions and so on) ● Research and educational infrastructure (and, therefore, availability of qualified labour) ● Innovation centres, science parks ● Availability of large enterprises and their attitude toward small and young technology-oriented firms, also intraregional production networks ● Availability of venture capital ● Role of key persons ● Entrepreneurial activities and entrepreneurial attitudes ● Decentralization processes in large agglomerations (the opposite of endogenous development)

Of course, the various facets of government policies must not be ignored. However, it is difficult to clearly assign them either to the demand factors or to the supply side. In fact, government policies can be both. They may provide, via R&D expenditure in favour of public and/or private universities, opportunities to hire highly skilled labour, a supply factor without any doubt. On the other hand, government R&D programmes, for example for military purposes, may create new technologies, new products and new markets. Sometimes, however, federal ministries are the only customers of such products. Given the large variety of different policies (and their varied goals, see Sternberg 1996a) in the countries and regions studied here, it is useful to distinguish between government policies (with explicit regional goals), federal R&D expenditure (with implicit regional impact, for example, contract research), and technology policies of the region (with explicit regional goals).

How knowledge-intensive clusters emerge

3

301

DESCRIPTION OF THE CASE STUDIES

In the following, intentionally not only will the quantitatively most important high-tech clusters of the respective country be considered but also – in order to be able to cover different region types – smaller regions with a high technology intensity will be discussed. A principal precondition is a growth dynamism which lies far above a country’s average. The case studies, without exception, are distinguished by a far above-average growth dynamic over the last decades, but only some of them (in part dependent on the ‘age’ of the region) by a far above-average level of economic activities. Similar to the approach followed by Bresnahan and Gambardella (2004a), I have selected regions in different national environments, with different main clusters and with differences in terms of the age of the high-tech clusters. Different from Bresnahan and Gambardella’s attempt, however, my selection is not restricted to ‘new industrial spaces’ in the sense of Scott (1988), but some of them, like the ‘Western Crescent’ around London, or Greater Boston, are really old industrialized areas (but with new industries!). The ten regions cover the five most important developed countries in terms of GDP/capita, absolute GDP and market size. The ‘Western Crescent’ (14 counties of the standard South East and South West regions, west of London, see Hall et al. 1987), the Munich region (eight counties around the city of Munich, Germany, see Sternberg and Tamásy 1999) as well as the Cité-Scientifique-de-Paris-Sud (four Départements in south-western part of the Ile-de-France around Paris, the economic and technological heartland of France, see Decoster and Tabariés 1993) with their continuing high growth dynamics represent the more or less clearly dominating high-tech clusters of Great Britain, Germany and France. Greater Boston (the CMSA ‘Boston-LawrenceSalem’) is the oldest high-tech region in the USA and an example to support the hypothesis that regions can still be among the high-growth regions in their country even in several ‘long waves of economic development’ and therefore with varying basic innovations (Saxenian 1994). The growth dynamics are by far stronger in Cambridgeshire (one of the three British counties that make up the standard region ‘East Anglia’, see Segal, Quince & Partners (1985) on the ‘Cambridge phenomenon’ and Athreye (2004) for a more recent assessment) and in the Research Triangle, North Carolina, USA (the area between the three university cities Raleigh, Durham and Chapel Hill, including the Research Triangle Park, one of the largest and most successful research parks worldwide; for the origins see Sirbu et al. 1976), but both these regions, by absolute standards, do not belong to the high-tech centres of their respective countries. Silicon Valley (the heart of which is the Santa Clara County in California,

302

Emerging clusters

USA) is the prototype of a high-tech cluster, and by absolute as well as relative criteria and according to statistical and dynamic indicators it numbers among the top technological centres of the USA (see Kenney 2000). In France, the Grenoble region (Département Isère as one of the eight départements in the région Rhône-Alpes) is deemed to possess the most innovative milieu even though by absolute criteria it is clearly only second to the Cité Scientifique de Paris Sud (Dunford 1991). The two case studies of Kyushu, Japan (southernmost of the four big Japanese islands encompassing seven prefectures with about 13 million inhabitants) and Sophia Antipolis, France represent the attempt by central government policy to attract R&D-intensive firms, public research institutions and higher education institutions to relocate in peripheral regions to trigger regional development impulses in the respective regions. Kyushu was a peripheral region both in economic and geographical respects until the mid-1980s when it began to attract attention under the label of ‘Silicon Island’ (see Matsubara 1992) as a high-tech cluster, as it had become home to an unusually high number of semiconductor plants. Situated in the Département Alpes-Maritimes on the Côte d’Azur, Sophia Antipolis is the largest, oldest and, in quantitative terms, most successful technology park in France (Longhi and Quéré 1993).

4 4.1

REASONS FOR THE EMERGENCE OF THE SELECTED HIGH-TECH CLUSTERS Determinants of the Emergence of High-tech Clusters – the Empirical Perspective

In this section, the causes of the genesis or the emergence (not of the subsequent growth) of the selected high-tech clusters will be discussed. The assessment is based on the study of literature and on interviews with regional experts (mean duration 1.5 hours, mainly with the help of nonstandardized questionnaires, see Table 12A.1) all of whom were asked to assess the same catalogue of determinants. Silicon Valley The emergence of Silicon Valley was promoted by an economically and technologically oriented university, by electronics enterprises (for example, Hewlett-Packard) already existing before the development of the semiconductor industry and by the presence of military and aeronautics installations with the resulting demand for semiconductor products (Scott 1988; Lee et al. 2000). Entrepreneurship, especially as local spin-off processes,

How knowledge-intensive clusters emerge

303

was the main mechanism to transfer technological excellence into business related effects for the regional economy (see Sturgeon 2000; Moore and Davies 2004; Klepper 2007). However, important impulses also came from key persons, such as the temporary vice-president of Stanford University, Fred Terman, and the Nobel-Prize-winning co-inventor of the transistor, Bill Shockley. Of major importance in this context was Terman’s untiring endeavour to promote the foundation of enterprises by the university. During its short life span, the Shockley Transistor Corporation, named after its founder, employed scientists of high standing, many of whom started very successful semiconductor companies (see Bernstein et al. 1977). Several of the following determinants of the development are directly or indirectly linked with the two key persons mentioned above. The enhanced reputation of Stanford University under Terman’s vice-presidency helped attract renowned researchers, which increased the quality of research and, hence, the chances of acquiring state R&D funds. The resulting availability of highly qualified labour may have been the chief individual cause of the dynamism of Silicon Valley (see Angel 1991). By contrast, the Industrial Park (founded in 1951) belonging to Stanford University seems to have profited from the boom after it had already been initiated rather than the other way round (see Rees 1986). Government policies were highly instrumental in bringing about the genesis of Silicon Valley in two different ways (SRI 1984). First, government demand for technologically sophisticated and new semiconductor products marked the beginning of the industrial life cycle of this branch in Silicon Valley. Between 1955 and 1963, the share of government institutions in the turnover of Silicon Valley’s semiconductor industry was between 35 per cent and 40 per cent; after that, as civilian markets successively opened up this share continued to reduce. Second, and possibly more important, the state influenced local high-tech development by granting R&D contracts to Stanford University and SME (Leslie 2000). In the final analysis, the development of semiconductor technology between 1940 and 1955 is the result of the military’s effort to miniaturize military electronic equipment. In Silicon Valley, as part of the United States’ ‘gunbelt’ (Markusen et al. 1991), a mutual influence developed between military demand for research projects and the electronic industry which continued to produce technological innovations many of which were suitable for military use. Thus, the orders placed by the Department of Defense (DoD) had a deciding influence on the local diffusion of new technological know-how in Silicon Valley. Finally, two factors difficult to capture by theoretical approaches should not be overseen: ‘the combination of a bit of luck and tremendous hard work’ (Moore and Davis 2004,

304

Emerging clusters

38). In fact, there seems to be a complementary relation between the luck of entrepreneurs (being in the right place at the right time with the right product and technologies) and the skills they have – at least during the early years of the Silicon Valley. Greater Boston Local research institutions and universities (above all MIT) in conjunction with their financing with federal funding are the most important factors to be mentioned when looking for the reasons behind the genesis and later growth of the Greater Boston high-tech cluster (Saxenian 1985). Among the most important causes is also the availability of venture capital. The permanently growing number of technology-based start-ups in particular would not even be thinkable to such an extent without third-party capital rather than credit granted by banks. What would MIT and the high-tech operations in the Greater Boston area be without the R&D expenditure of the federal government, in particular the Department of Defense and NASA? In summary, the impact of governmental R&D expenditure on the Greater Boston high-tech cluster can best be evaluated with the typing of military agglomerations as described by Markusen et al. (1991). Accordingly, Greater Boston belongs to the type ‘Military–Educational Complex’ within which a university (in this case MIT) due to certain historical constellations (also) runs military research and acts as a significant catalyst in state R&D expenditure. Since World War II, MIT has received unusually broad R&D contracts from the Department of Defense and, with its close cooperation with local high-tech operations (often spin-offs of MIT itself), is a technology transfer institution par excellence. The combination of MIT and state R&D expenditures was the decisive factor behind the genesis of this high-tech cluster and one of the most significant for its later development. It should be underlined in summary that the Greater Boston high-tech cluster is not the result of an explicit technology policy. Viewed this way, the high-tech sector did indeed develop in an unplanned and spontaneous fashion. The implicit regional influence of state technology policy in the form of R&D expenditure by the federal government, together with the R&D infrastructure which also depends on this expenditure, is however the decisive factor behind the generation and later development of this high-tech cluster. The ‘Research Triangle’/North Carolina The regional concentration of three universities – North Carolina State University in Raleigh, Duke University in Durham and the University of North Carolina in Chapel Hill – is unusual even by American standards and prompted the name of ‘Research Triangle’ (20 000 graduates p.a. in

How knowledge-intensive clusters emerge

305

an area of an approximate 25 km radius). The economic importance of these local universities becomes evident only in the context of the Research Triangle Park. The Research Triangle Park (RTP) was opened in 1959 and is one of the largest and most successful research parks in the US, employing approximate 37 500 people in over 135 enterprises, research institutions and service facilities in 2002 (see Link and Scott 2003). The significance of the RTP for the emergence and dynamism of the high-tech cluster of the Research Triangle is big. The Research Triangle is, no doubt, the most widely known and most successful ‘example of a science and research area planned to create a high-technology complex’ (Malecki 1986, 58, see also Luger and Goldstein 1992). The initiative for the foundation of the RTP was chiefly owing to the economic and technology policy of the region itself (Governor Luther Hodges). However, the founders were able to win federal authorities, universities and, naturally, economic circles for their concept. Just the same, federal authorities by their locational decisions considerably contributed to the success of the RTP and with it of the Research Triangle. As for the Research Triangle the DoD had no detectable influence on the genesis of the high-tech sector. Nevertheless, the Research Triangle is different from Silicon Valley by its ‘much greater reliance on government contracts and guidance’ (Castells 1989, 101). More recently two industry clusters, medical services, labs/hospitals as well as biotechnology and pharmaceuticals, are relevant elements of the State of NC initiative ‘Vision 2030’, promoted by the Board of Science and Technology (see Feser and Luger 2003). The Western Crescent The combination of two causes, partially interlinked, are predominantly responsible for the emergence of this high-tech cluster: decentralization (of population and industry) and deindustrialization and military-related government R&D spending. Deindustrialization and the ensuing decrease of absolute and relative employment in the secondary sector has taken place in all counties of southern Great Britain since the 1960s. Even in the parts of the Crescent with the highest growth rates massive employment losses were the rule. These losses were heaviest in the agglomeration of Greater London (see Hall 1990). Part of these losses result from the relocation of industries and services into counties west of the metropolis where small towns predominate and where the rate of new enterprise foundations is significantly higher (see Keeble 1991). The process of decentralization progressing westward from the City of London and the western suburbs to the neighbouring counties can be illustrated by the development of the electrotechnical industry during the past 200 years (see Castells and Hall 1994). The migration westward was oriented by radial lines (railroads, roads) and

306

Emerging clusters

skilled or unskilled labour. Strategic considerations during World War II increased the trend of migration out of London. The ensuing phase of drastically boosted dynamism in the Western Crescent was obviously due to political factors which were only indirectly associated with deindustrialization and decentralization. Decentralization, however, does not explain why the Western Crescent has been the highly preferred destination of these industries since the end of World War II. It was the R&D funds of the MoD that one-sidedly favoured the counties of the Western Crescent. The R&D money went where there was already R&D capacity in the form of well-known contracting firms or MoD-owned research institutions. The location pattern of the research institutions goes back to the war and the immediate postwar years, the 1940s, when many of these institutions were founded (see Wells 1987). After their relocation away from the areas that were exposed to German air raids, the share of government-supported and governmentowned research institutions applying for relocation to the South East or South West was 66 per cent and 86 per cent, respectively (see Heim 1987). The military R&D institutes developed into centres of attraction for contracting and subcontracting firms especially of electrotechnology for whom spatial proximity was most important. Frequent personnel transfer between research institutions of the MoD and military-dependent industrial establishments made it increasingly necessary to relocate in the attractive south of the country, because the employees had similar locational preferences. All in all, the Western Crescent is a perfect example of a state-dependent high-tech cluster. Cambridgeshire The renowned University of Cambridge constitutes the main individual cause for the emergence of many high-tech enterprises in this high-tech cluster. For a long time the university has already actively endeavoured to utilize its own technological innovations by way of technological transfer. Cambridge Science Park, initiated by Trinity College, is an expression and a symbol, as it were, of this endeavour (see Castells and Hall 1994). By striving to be interdisciplinary, valuing quality in research and teaching and applying scientific findings in practice, the university generated a climate in which the informal networks could develop that are vital to technology-based industrial establishments. In addition, Cambridgeshire profits from the positive image of Cambridge as a historically grown medium-sized town with a good educational and cultural infrastructure and pleasant surroundings. Therefore, this region became very attractive not only as a place of residence, but also of work especially to scientists from the capital with its agglomeration disadvantages (see Keeble 1989).

How knowledge-intensive clusters emerge

307

In this way, Cambridgeshire, too, profited from the process of decentralization, deindustrialization and recentralization in Great Britain giving rise to new growth regions. These three causes were of more consequence than other aspects like the Cambridge Science Park (CSP) that is more a symbol than a cause of what is called the ‘Cambridge Phenomenon’ (see Crang and Martin 1991). Cambridgeshire is a good example of an unplanned, endogenously developed high-tech cluster in which government influence was negligible. Grenoble The innovative milieu of Grenoble has its origins in the linkages between the local research and education institutions and the SME; in part, these linkages developed organically in the course of many decades, in part they were promoted by regional and national technology policy. The ‘father’ of these linkages was the Nobel laureate in physics, Louis Néel. He brought his influence to bear in order to bring the national Atomic Energy Commission and the National Energy Research Centre to Grenoble in 1956. The Zone pour l’Innovation et les Réalisations Scientifiques et Techniques (ZIRST), numbering among the oldest and most successful technology parks in France, is a visible expression of the linkages between science and economy which are by no means a matter of course. In a cumulative process, the ZIRST, science and the region have mutually stimulated one another. In addition and aside from the historically grown entrepreneurial spirit in the region (Merlin-Gerin invented hydro-electricity) the explicit and widely consensual technology policy of the region itself (the Région Rhône-Alpes, the participating communities) essentially promoted the genesis of a high-tech cluster (Charbit et al. 1991). The technology policy of this region is characteristic of a concerted action of all the essential local ‘actors’ of the innovative milieu (economy, science, education) whose intentions were not diametrically opposed to the implicitly formulated goals of national technology policies. Despite the obvious predominance of endogenous causes, the influence of the government must not be underrated. Without the decentralization of state R&D institutions (located in Paris), important impulses would have been lacking there and the development into a technopolis of this order of magnitude would have been impossible (Brocart 1991). Sophia Antipolis Sophia Antipolis is the largest, oldest and, in quantitative terms, most successful technology park in France and is, in the opinion of many French observers, a technopole. In contrast to the other two high-tech clusters in

308

Emerging clusters

France, the Département Alpes-Maritimes in the south of the country, where Sophia Antipolis is situated, has no tradition as a high-tech cluster. Up until 40 years ago tourism formed the basis of the regional economy (with the exception of Marseille). Today, Alpes-Maritimes advertises the fact that it is situated on the southern French ‘Route des Hautes Technologies’, and an astounding technology-based industrial potential has indeed developed ‘ex nihilo’ in this Département (Longhi and Quéré 1993; Longhi 1999) Sophia Antipolis is almost exclusively attributable to measures of central government and, secondarily, of the Conseil Régional, the Département and the local authorities. The genesis of Sophia Antipolis goes back to the 1960s and the idea of ‘Le Quartier Latin aux Champs’, that is the amalgamation of education, research, industry and living out in the country in what is, as a bonus, a beautiful area. Pierre Laffitte was the key person behind it all. As the then (1968) Director of the Ecole Nationale Supérieure des Mines de Paris who later became Senator des Départements, he possessed a masterful ability to take three national networks that came together in Paris and use them for his own purposes during the birth phase of Sophia Antipolis: academia, high-ranking ministerial civil servants and the managers of research-intensive large firms (Perrin 1988). The development of the park gained pace towards the end of the 1970s when interventions by central government (the arrival of Air France and France Télécom, relocation of research institutions out of Paris) happened. The relocation of the elite Ecole Nationale Supérieure des Mines got the ball rolling and the decentralization laws enacted by President Mitterrand in 1982 represented an additional important impulse. In recent years there has been a shift among political players from the national level to the regional and local level (Garnsey and Longhi 2004). Cité Scientifique de Paris Sud (Ile-de-France) The Région Ile-de-France (hereinafter referred to as IdF) is the core economic, technological and political region in France. It has for a considerable time accounted for roughly half of all researchers employed in manufacturing industries, half of private R&D expenditure and half of the employment in high-tech industries in the country (see Mailfert 1991). In absolute terms, the IdF is undisputedly one of the largest technology regions in Europe (Castells and Hall 1994). The Cité Scientifique de Paris Sud (hereinafter referred to as Cité) proved to be a particularly innovative part of the IdF in the course of the 1980s. There was no industry of any kind in this area until the end of the 1950s, making it a ‘new industrial space’ as defined by Scott. This area is home to around 50 per cent of the IdF total. The high density of high-tech industries is characteristic of the

How knowledge-intensive clusters emerge

309

Cité: those active in technology policy in the ministries, the IdF region and the eight Départements have no doubt aided the growth of the Cité technology region in the past, but they have only recently been able to encourage the development of an innovative milieu. On the other hand, the locational behaviour of Paris high-tech firms can be explained more plausibly with the high prestige of the Cité: no old industries, many green spaces, good transport links, sufficient and attractively priced industrial and residential space. The technology policy of national government pursues the explicit goal of reducing the disparities between the IdF and the rest of the country. In reality, however, the IdF and the Cité within it still benefit from the R&D expenditure of the ministries, although the weight of the IdF region itself and the local communities is relatively low (Benko 1991). The local production system is characterized by the aerospace industry and military electronics/telecommunications. They received massive public support in the past 50 years as part of ‘mission-oriented’ technology policy (see Decoster and Tabariés 1993). This is based on the concept of the ‘filière’ and makes use of the exclusive contacts between the elite representatives of governmental administration and major private enterprises, leading Storper (1993, 445) to conclude correctly: ‘The state and its closely associated (or owned) companies, then, remains the key to the specializations of Ile-de-France’s high technology industries’. The research labs of large firms in particular have benefited from these R&D expenditures and they in turn are in close cooperation with major public research labs (Decoster et al. 2006). It was not spatial proximity, but the prospect of long-term, publicly subsidized large-scale projects that was the cause of this cooperation. That is why central government’s expenditures did not allow an innovative milieu to develop between local large and small enterprises. Munich In this case, the reasons for the emergence of the high-tech cluster can be subdivided into those concerning the whole regional economy and those having to do with locational characteristics particularly favourable to the key industry of microelectronics (see Sternberg and Tamásy 1999). Since the 19th century, there has been traditional industry (automobile industry, railway engineering and mechanical engineering) in Munich. The in-migration of manpower from central and southeastern European countries after World War II was conducive to industrialization in Bavaria. The main entrepreneurial impulse came from the relocation of the Siemens AG headquarters from Berlin to Munich in 1948 for reasons of military policy (fear of dismantling). This was one of the major reasons why Munich became the centre of the electronics industry in Germany. Siemens had a

310

Emerging clusters

direct and indirect strong influence on the growth of both the electrotechnical and the software industry in Munich. As a result of the technology and structural policy of the federal state of Bavaria the industrial ‘climate’ in Munich was favourable to entrepreneurial activities. Four location factors were especially advantageous for the microelectronics industry as the key branch of the Munich high-tech cluster. The first factor is the importance of Siemens AG for the development of the microelectronics industry. The second factor is a considerable regional demand for the products of the electrotechnical industry. Third, the local supply of risk capital no doubt encouraged the concentration of R&Dintensive businesses of the electrotechnical industry, many of them being part of the SME. Munich is also more attractive to investors than any other German region (see Castells and Hall 1994). The availability of highly skilled labour is a further prerequisite for the growth of the electrotechnical industry. The underlying causes of high-tech growth in Munich introduced so far are overlapped and in part surpassed in their significance by the technology and research policy measures of the federal and state government. At the federal level, an important impetus came from the R&D spending of the Defense Department. The larger part of location decisions in favour of state R&D institutions in the Munich region was made in the 1950s and 1960s. Several of the R&D institutions founded in Munich then are part of nuclear energy and/or armament research and owe their location in Munich to the influence of the temporary Defense Minister and Bavarian Prime Minister of many years, F.J. Strauss. The Munich region is not only the German high-tech centre, but also a centre of the likewise R&Dintensive arms industry. Kyushu The significance of the ‘silicon island’ Kyushu as a high-tech cluster is based primarily on one industry, namely the semiconductor industry (Sternberg 1995). Therefore, the causes for the development of Kyushu into a high-tech cluster are to be found in the location standards set by this industry between 1970 and 1990. These standards can be derived from the hypothesis of the spatial division of labour in the Japanese industry which is based on a spatial and an operational hierarchy. The comparative advantages of Kyushu do not only apply to just one high-tech industry branch but essentially also to just one production phase, namely the intermediate assembly of prefabricated parts. During a relatively early phase in the product life cycle of the semiconductor industry – the labour-intensive manufacture of a new product in an increasing number of units – Kyushu possessed the required locational characteristics. This meant, first, the

How knowledge-intensive clusters emerge

311

availability of flexible labour and, second, considerably improved transportation routes (air and railroad traffic), low-cost space and measures of local economic policy (see Sargent 1987). It was only owing to these comparative advantages that Kyushu succeeded in becoming a beneficiary of the decentralization of the high-tech industry. More recent political attempts in the context of the Fukuoka silicon sea-belt project aim to integrate Kyushu’s high-tech industry into international innovations networks within East Asia (see Kitagawa 2005). 4.2

Synopsis of the Case Studies

An analysis of the genesis factors of the ten high-tech clusters reveals that not a single determinant constitutes a necessary or sufficient precondition (see Table 12.1). In the American case study regions and in Cambridgeshire, the research and education infrastructure counts among the major factors of influence, while in the Western Crescent and in Kyushu it is of hardly any significance at all. The same is true of the so-called ‘soft’ locational factors, the amenities, often mentioned in connection with high-tech industries; in most regions they have very little generating impact. The importance of risk capital, of science or research parks and local inventions is also heavily overrated. However, the decentralization processes in other metropolises (Greater London for the Western Crescent, other US metropolises for the Research Triangle, Tokyo–Osaka–Nagoya for Kyushu) and the regional key persons have turned out to be of greater importance than expected. Table 12.1 makes it clear that all the determinants corresponding to a state’s technology policy activities together with the R&D infrastructure, which was in part influenced by those determinants, were the main impacting factors in the emergence of the ten regions.

5

THE ROLE OF GOVERNMENT POLICIES – USEFUL INSIGHTS FROM THE ‘VARIETIES OF CAPITALISM’ APPROACH?3

The various facets of government R&D policies have affected most hightech clusters considerably. The problem is in the multitude of varying effects of government R&D activity. It may have a direct effect or an indirect effect via the above-mentioned determining factors. Government R&D affects, for example, the quality of the research and education infrastructure in a region and it can – to a degree – steer the availability of risk capital or high-tech parks. Direct influence manifests itself essentially

312

Greater Boston (USA) o

•••

o

o

•••

o

Silicon Valley (USA)

o

•••

o

o

•••

•

o

o

o

•••

•

o

•

o

o

o

•••

o

•

o

o

o

o

•••

o

o

o

•••

•

o

•

•••

o

o

•••

o

•••

o

o

o

o

o

o

•

o

•

•••

o

Sophia Grenoble Western Cambridge- Munich Cité Research (D) shire (F) Crescent Triangle Scientifique Antipolis (UK) (UK) (F) (F) (USA)

Qualitative assessment of selected determinants of the genesis of ten knowledge-intensive clusters

Government policies (with explicit regional goals) Federal R&D expenditure (with implicit regional impact, e.g., contract research) Technology policy of the region (with explicit regional goals) Private demand for technology-intensive new products Public demand for technology-intensive new products (especially military demand) Amenities (environment, culture, living conditions, etc.)

Determinant

Table 12.1

o

o

o

•

o

•••

Kyushu (J)

313

••• the three most important factors;

qualitative interviews (see text).

Source:

• important factor;

••• •

••• o

• o

o less important or not relevant.

o •••

o o

o

• o

o

o

•••

o

•

o

•••

•

o

o

o

o

o

o

•

o

•••

•••

•••

o

o

•

•••

•••

•

Notes:

Research and educational infrastructure (and, therefore, availability of qualified labour) Innovation centres, science parks Availability of large enterprises and their attitude toward small and young technologyoriented firms, also intraregional production networks Entrepreneurial activities, start-ups Availability of venture capital Role of key persons Decentralization processes in large agglomerations (the opposite of endogenous development) o •••

o

o

o

o

o •

o

o

•••

•••

••• o

o

•••

o

o

o •••

o

•••

o

o

314

Emerging clusters

in two forms. Government R&D expenditure, for example for contract research at universities and in enterprises or for state R&D facilities, has implicitly regional effects, since it is essentially oriented towards available resources. In part owing to this, early demand for technology-intensive new products, for which no commercial market exists yet, is of vital importance for young enterprises and industries. This variant is important particularly with regard to innovations which can be used for military purposes; Silicon Valley, Greater Boston and the Western Crescent, for example, are proof of this. Both forms of state influence are especially relevant in the very early stages, that is, during the emergence of a hightech cluster, as Saxenian (1994) has shown. In principle, this also applies to a third variant of state technology policy, namely that which explicitly pursues regional goals. Among the case studies selected for this chapter, however, this only applies to Kyushu (and possibly Sophia Antipolis) and may not lead to the desired results there. Our fairly surprising result is that the promotion of high-tech clusters by government technology policies is most pronounced in countries where no technology policy pursuing explicitly regional goals exists. Some years ago Hall and Soskice (2001) developed their approach of ‘varieties of capitalism’ (VoC) stressing the institutional foundations of comparative advantage of nations. Their attempt was strongly considered among regional economists and economic geographers as well. As I have shown, policies, organizations and institutions play a more or less important role for the high-tech clusters studied in this chapter – and this influence differs between the regions considered. Thus it could make sense to analyse whether the VoC approach is useful to explain why policy influence on high-tech cluster emergence differed in the past – and why it eventually will differ in the future as well. There is no doubt that this attempt is a valuable progress beyond three hitherto more or less established perspectives on institutional variation: the modernization approach, the concept of neo-corporatism and the social system of production. Hall and Soskice mainly argue that market economies (they focus chiefly on national market economies, not regional ones) can be compared by reference to the way in which firms resolve the co-ordination problem they face in five spheres: industrial relations, vocational training and education, corporate governance, inter-firm relations and employees. Hall and Soskice differentiate between liberal market economies and co-ordinated market economies. While in the latter firms depend more heavily on non-market relationships to co-ordinate their endeavours with other actors (for example, relational or incomplete contracting, more reliance on collaborative relationships), firms in liberal market economies co-ordinate their activities primarily via hierarchies and

How knowledge-intensive clusters emerge

315

competitive market arrangements. From the perspective of the authors the US economy and the UK economy belong to the liberal market economies whereas Germany and Japan are characterized as co-ordinated economies. France, the fifth economy considered in this chapter, has a more ambigious position, showing some signs of institutional clustering as well, indicating that it constitutes the ‘Mediterranean’ type of capitalism marked by a large agrarian sector and a mix of strong (central!) state interventions in the sphere of corporate finance and more liberal arrangements in the sphere of labour relations (Hall and Soskice 2001, 21). Do the impacts of institutional and, especially, of policy interventions on high-tech cluster emergence differ between the four countries and the ten regions located there? And do they accord with the VoC-typology of liberal versus market economies? Let us look at Table 12.1 again. In the two liberal market economies of the UK and the US (with a total of five regions studied here) the impact of policies with intended regional goals is small; however, unintended spatial effects have played a crucial role in four out of these five regions. For the rest of the indicators, however, differences between these five regions are significantly large. Conversely, the two regions representing the two co-ordinated market economies, Japan and Germany, have several characteristics in common (the role of large enterprises, and of regional technology policies), despite the fact than one of them is a leading high-tech region in the country while the other is definitely not. For the specific case of France, neither a real liberal market economy nor a true co-ordinated market economy, the three regions studied show very different results. Thus, there is neither a clear support nor a rejection of the VoC-typology of national market economies. Possible explanations for this mismatch between empirical evidence and theoretical assumptions are: 1.

2.

3.

Regions differ even within the same nation, although the national policy is the same; regional framework conditions differ and they are eventually more important for cluster emergence than national policies and other national framework conditions. Some of the policies are developed by the national government, others are organized and financed by regional/local governments; several of the latter show attributes different from the attributes of the national policies in the same country. Policies even within the same country may vary over the decades (for example, in Germany today’s national innovation programmes are much more focused on regional/national competitiveness than earlier ones – and this strengthens the strong regions (whereas in the past reducing interregional inequalities in innovation activities has been

316

Emerging clusters

the primary goal); on the other hand, in France more decentralization and deconcentration policies exist than in the past, an outcome of which are Sophia Antipolis and the ‘technopôle’ concept). At the end, there is a clear message for governments in all countries: it makes no sense to expect a new high-tech cluster as a direct (and quick) result of public policies. Wallsten (2004) in his systematic statistical analysis on the role of explicit government policies to support regional technology development in US counties has not find any empirical effect. In the words of Bresnahan and Gambardella (2004b, 355): public government should avoid ‘the foolishness of directive public policy efforts to jumpstart clusters or to make top-down or directive efforts to organize them’. However, this seems to be exactly what happens nowadays in several industrialized countries (see, for example, Kiese and Schätzl 2008, for Germany, or Development Bank of Japan 2007 on the cluster initiatives of government ministries in Japan). My conclusions are not contradictory to the result that public demand, especially military demand, may act as an extremely important factor supporting (or even creating) new technologies and entrepreneurship. The Silicon Valley story has shown this clearly. But this is quite an indirect and long-run, less intended than unintended effect of government policies. Considering the arguments of Meyer and Schubert (2007) I would argue that the emergence of today’s high-tech cluster has only rarely been the result of a purposeful action of policymakers but, nevertheless, this emergence has not occurred by chance. This brings me to more general conclusions on government policies to create clusters.

6

CONCLUSIONS AND POLICY IMPLICATIONS4

To summarize the empirical results this chapter argues that no single determinant constitutes a necessary or sufficient precondition for the genesis of a knowledge-intensive regional cluster. The various facets of government R&D policy affect most clusters to a considerable extent, for instance, the quality of the research and education infrastructure in a region. All the determinants corresponding to a nation’s and/or a state’s technology policy activities (more indirect than direct) together with the R&D infrastructure were the main impacting factors in the emergence of the majority of the clusters. However, most of these clusters are not the planned outcome of explicit government policies with explicit regional goals. Empirical analysis clearly reveals that non government factors (and their combined impact) have played an important role in the process of cluster emerging although not a single determinant constitutes a necessary

How knowledge-intensive clusters emerge

317

or sufficient precondition in the ten high-tech clusters studied (see also Braunerhjelm and Feldman 2006 on various clusters and cluster types). In the American case study regions and in Cambridgeshire, the research and education infrastructure counts among the major factors of influence, while in the Western Crescent and in Kyushu it is of hardly any significance at all. The same is true of the so-called ‘soft’ locational factors, the amenities, often mentioned in connection with knowledge-intensive regional clusters; in most regions they have had very little generating impact. The importance of risk capital, of science or research parks and local inventions is also heavily overrated. However, the decentralization processes in other metropolises (Greater London for the Western Crescent, other US metropolises for the Research Triangle, Tokyo–Osaka–Nagoya for Kyushu) and the regional key persons (for example, Pierre Laffite in Sophia Antipolis, Luther Hodges in the Research Triangle or Fred Terman and Bill Shockley in Silicon Valley) have turned out to be of greater importance than expected. Last not least it seems to be obvious that determinants of cluster growth potentially differ from the determinants of the cluster genesis more the older the cluster is, that is, ‘the processes of starting and sustaining a cluster have different economics’ (Bresnahan and Gambardella 2004b, 334). Obviously, there are different routes and different combinations of determinants that may lead to starting a cluster. Similar to the clusters investigated by Bresnahan and Gambardella (with the exception of the Silicon Valley they are different from the clusters analysed in this chapter) the clusters discussed here have a number of common determinants of their origins. External effects like benefits to particular technology firms that arise from the presence of other firms were not that important during the early years but a sizable and growing demand was: both new technology opportunities and new market opportunities were crucial. From a supplyside of view skilled (often a result of long-term public investment) labour as well as firm-building and market-building capabilities mattered. Finally I would like to shed light on the opportunities and threats of a cluster-based strategy of regional development – a model which is still extremely popular both in developed and in developing countries (see, for example, OECD 2007). The observations made so far have demonstrated that high-tech clusters to date have rarely been the result of explicit technology policies with regional goals. But policy has had and continues to have an influence in most of the regions described, albeit it most usually unintended. For approximately ten years, policies seeking explicitly to create high-tech clusters by specific instruments and strategies have dominated in all industrialized countries and in several emerging countries, with the Silicon Valley model serving as the shining example. The wording has changed slightly since the 1970s and 1980s: high-tech regions are cited

318

Emerging clusters

less frequently and high-tech clusters all the more commonly. Individual elements of a functioning cluster appear indispensable as a precondition of a cluster-oriented economic promotion strategy (see Rosenfeld 1997), such as a critical mass of firms and research institutions, local players willing to cooperate and initial attempts, that can be expanded, to localize value creation chains. These preconditions are not fulfilled in every region. Furthermore, it is essential to identify industries and technology fields that have a touch of the original and novel, and that appropriate to the potential of the region. As the previous sections and the economic history of many high-tech clusters show, some of them emerged without any influence from policy whatsoever. This is the case when the locational decisions of a group of entrepreneurs are the response to positive agglomeration effects and/ or demand effects perceived by these entrepreneurs (for example access to local pools of highly specialized labour and the availability of technological and market opportunities, respectively). But there are also enough examples of policy’s supporting influence in the genesis of clusters, although only ever as one of many decisive factors and/or determinants. Many of today’s high-tech clusters in particular, of which some are still relatively young industrialized regions, owe their generation and their cluster attributes to political strategies of the respective federal government and/ or regional government itself. This effect may have been an explicit goal of the policy and therefore have been intended, as with the Research Triangle Park in North Carolina, USA, or the result of implicit technology policies with unintended regional goals, as in the case of Silicon Valley, where the microelectronics cluster would not have generated without the support of the Department of Defense and NASA.

NOTES 1. An earlier version of this chapter was presented at the SPRIE workshop on ‘High-Tech Regions 2.0: Sustainability and Reinvention’, Stanford University, November 13–14, 2006. 2. Part of this section is based upon Sternberg (1996b). 3. This section is partially based upon Sternberg (1996a). 4. Part of this section is based upon Sternberg et al. (2004).

REFERENCES Angel, D.P. (1991), ‘High-Technology Agglomeration and the Labor Market: The Case of Silicon Valley’, Environment and Planning, 23, 1501–16.

How knowledge-intensive clusters emerge

319

Athreye, S. (2004), ‘Agglomeration and Growth: A Study of the Cambridge HighTech Cluster’, in T. Bresnahan and A. Gambardella (eds), Building High-tech Clusters, Cambridge, New York: Cambridge University Press, pp. 121–59. Aydalot, P. (ed.) (1986), Milieux Innovateurs en Europe, Paris: GREMI (privately printed). Benko, G. (1991), Géographie des Technopôles, Paris: Masson. Benneworth, P. and N. Henry (2004), ‘Where Is the Value Added in the Cluster Approach? Hermeneutic Theorising, Economic Geography and Clusters as a Multiperspectival Approach’, Urban Studies, 41, 1011–23. Bernstein, A., B. DeGrasse, R. Grossman, C. Paine and L. Siegel (1977), Silicon Valley: Paradise or Paradox? The Impact of High Technology Industry on Santa Clara County, Mountain View: Pacific Studies Center. Boschma, R. and K. Frenken (2003), ‘Evolutionary Economics and Industrial Location’, Review for Regional Research, 23, 183–200. Boschma, R. and R. Martin (2007), ‘Constructing an Evolutionary Economic Geography’, Journal of Economic Geography, 7, 537–48. Braunerhjelm, P. and M.P. Feldman (eds) (2006), Cluster Genesis. TechnologyBased Industrial Development, Oxford: Oxford University Press. Bresnahan, T. and A. Gambardella (eds) (2004a), Building High-tech Clusters, Cambridge, New York: Cambridge University Press. Bresnahan, T. and A. Gambardella (2004b), ‘Old-Economy Inputs for NewEconomy Outcomes: What have we Learned?’ in T. Bresnahan and A. Gambardella (eds), Building High-tech Clusters, Cambridge, New York: Cambridge University Press, pp. 331–59. Brocart, M. (1991), La Science et les Regions, Paris: Reclus. Camagni, R. (ed.) (1991), Innovation Networks: Spatial Perspectives, London, New York: Belhaven Press. Camagni, R.and D. Maillat (eds) (2006), Milieux Innovateurs. Théorie et Politiques, Paris: Economica, Anthropos. Castells, M. (1989), The Informational City. Information Technology, Economic Restructuring, and the Urban-Regional Process, Oxford, Cambridge: Blackwell. Castells, M. and P. Hall (1994), Technopoles of the World, London, New York: Routledge. Charbit, C., J.L. Gaffard, C. Longhi, J.C. Perrin, M. Quéré and J.L. Ravix (1991), ‘Systèmes d’Innovation Localisés en Europe – Cohérence Diversité des Systèmes d’Innovation’, FAST Occasional Papers 235, Brussels. Crang, P. and R. Martin (1991), ‘Mrs Thatcher’s Vision of the “New Britain” and the Other Sides of the “Cambridge Phenomenon”’, Environment and Planning D, 9, 91–116. Crevoisier, O. (2004), ‘The Innovative Milieus Approach: Toward a Territorialized Understanding of the Economy?’, Economic Geography, 80, 367–79. Decoster, E. and M. Tabariés (1993), ‘Innovation and Regional Planning: The Ile-de-France Sud Technopole’, in J. Simmie, J. Cohen and D. Hart (eds), Technopole Planning in Britain, Ireland and France: The Planned Regional Acceleration of Innovation, London: University College London, Planning and Development Research Centre, pp. 83–102. Decoster, E., A. Matteaccioli, V. Peyrache and M. Tabariés (2006), ‘Les Réseaux d’Innovation en Région Parisienne: Micro-milieux en Émergence’, in R. Camagni and D. Maillat (eds), Milieux Innovateurs. Théorie et Politiques, Paris: Economica, Anthropos, pp. 219–60.

320

Emerging clusters

Development Bank of Japan (2007), Standortvorteil Cluster. Netzwerke in Deutschland und Japan, Frankfurt/M. Dunford, M. (1991), ‘Industrial Trajectories and Social Relations in Areas of New Industrial Growth’, in G. Benko and M. Dunford (eds), Industrial Change and Regional Development, London: Belhaven, pp. 51–82. Feser, E.J. and M.I. Luger (2003), ‘Cluster Analysis as a Mode of Inquiry: Its Use in Science and Technology Policymaking in North Carolina’, European Planning Studies, 11, 11–24. Garnsey, E. and C. Longhi (2004), ‘High Technology Locations and Globalisation: Converse Paths, Common Processes’, International Journal of Technology Management, 28, 336–55. Garud, R. and P. Karnøe (2001), ‘Path Creation as a Process of Mindful Deviation’, in R. Garud and P. Karnøe (eds), Path Dependence and Creation, Mahwah, NJ: Earlbaum, pp. 1–38. Giuliani, E. (2007), ‘The Selective Nature of Knowledge Networks in Clusters: Evidence from the Wine Industry’, Journal of Economic Geography, 7, 139–68. Hall, P. (1990), ‘Structural Transformation in the Regions of the United Kingdom’, Working Paper 507, Berkeley: Institute of Urban and Regional Development, University of California Berkeley. Hall, P. and A. Markusen (eds) (1985), Silicon Landscapes, Boston, MA: MIT Press. Hall, P.A. and D. Soskice (2001), ‘An Introduction to Varieties of Capitalism’, in P.A. Hall and D. Soskice (eds), Varieties of Capitalism, Oxford, New York: OUP, pp. 1–70. Hall, P., M. Breheny, R. McQuaid and D. Hart (1987), Western Sunrise, London: Allen & Unwin. Heim, C.E. (1987), ‘R&D, Defense, and Spatial Divisions of Labor in TwentiethCentury Britain’, Journal of Economic History, XLVII, 365–78. Hospers, G.-J. and S. Beugelsdijk (2002), ‘Regional Cluster Policies: Learning by Comparing?’, Kyklos, 55, 381–402. Keeble, D. (1989), ‘High Technology Industry and Regional Development in Britain: The Case of the Cambridge Phenomenon’, Environment and Planning, C 7, 153–72. Keeble, D. (1991), ‘Core–Periphery Disparities and Regional Restructuring in the European Community of the 1990s’, in H.H. Blotevogel (ed.), Europäische Regionen im Wandel, Dortmund: Dortmunder Vertrieb für Bau- und Planungsliteratur. Kenney, M. (ed.) (2000), Understanding Silicon Valley. The Anatomy of an Entrepreneurial Region, Stanford: Stanford University Press. Kiese, M. and L. Schätzl (eds) (2008), Cluster und Regionalentwicklung, Dortmund: Rohn. Kitagawa, F. (2005), ‘The Fukuoka Silicon Sea-belt Project – An East Asian Experiment in Developing Transnational Networks’, European Planning Studies, 13, 793–9. Klepper, S. (2007), ‘The Evolution of Geographic Structures in New Industries’, in K. Frenken (ed.), Applied Evolutionary Economics and Economic Geography, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Lee, C.M., W.F. Miller, M. Gong Hancock and H.S. Rowen (eds) (2000), The Silicon Valley Edge. A Habitat for Innovation and Entrepreneurship, Stanford: Stanford University Press.

How knowledge-intensive clusters emerge

321

Leslie, S.W. (2000), ‘The Biggest “Angel” of them all: The Military and the Making of the Silicon Valley’, in M. Kenney (ed.), Understanding Silicon Valley. The Anatomy of an Entrepreneurial Region, Stanford: Stanford University Press, pp. 48–70. Link, A. and J. Scott (2003), ‘The Growth of Research Triangle Park’, Small Business Economics, 20, 167–75. Longhi, C. (1999), ‘Networks, Collective Learning and Technology Development in Innovative High Technology Regions: The Case of Sophia-Antipolis’, Regional Studies, 33, 333–42. Longhi, C. and M. Quéré (1993), ‘Innovative Networks and the Technopolis Phenomenon – The Case of Sophia Antipolis’, Environment and Planning C: Government and Policy, 11, 317–30. Luger, M.I. and H. Goldstein (1992), Technology in the Garden, Chapel Hill: University of North Carolina Press. Mailfert, A. (1991), Recherche et Territoire, Paris: La Documentation Française. Malecki, E.J. (1986), ‘Research and Development and the Geography of HighTechnology Complexes’, in J. Rees (ed.), Technology, Regions, and Policy, Totowa: Allen & Unwin, pp. 51–75. Malecki, E.J. (1991), Technology and Economic Development: The Dynamics of Local, Regional, and National Change, New York: Longman. Malmberg, A. and P. Maskell (2002), ‘The Elusive Concept of Localization Economies: Towards a Knowledge-based Theory of Spatial Clustering’, Environment and Planning, A, 34, 429–49. Markusen, A.R., P. Hall, S. Campbell and S. Deitrick (1991), The Rise of the Gunbelt. The Military Remapping of Industrial America, New York, Oxford: OUP. Martin, R. and P. Sunley (2003), ‘Deconstructing Clusters: Chaotic Concept or Policy Panacea?’, Journal of Economic Geography, 3, 5–35. Martin, R. and P. Sunley (2006), ‘Path Dependence and Regional Economic Evolution’, Journal of Economic Geography, 6, 395–437. Matsubara, H. (1992), ‘The Japanese Semiconductor Industry and Regional Development: The Case of “Silicon Island” Kyushu’, The Economic Review of Seinan Gakuin University, 27, 43–65. Metcalfe, R. (1998), InfoWorld, 2 March, http://www.infoworld.com. Meyer, U. and C. Schubert (2007), ‘Integrating Path Dependency and Path Creation in a General Understanding of Path Constitution. The Role of Agency and Institutions in the Stabilisation of Technological Innovations’, Berlin: Science, Technology & Innovation Studies 3. Moore, G. and K. Davis (2004), ‘Learning the Silicon Valley Way’, in T. Bresnahan and A. Gambardella (eds), Building High-tech Clusters, Cambridge, New York: Cambridge University Press, pp. 7–39. OECD (2007), Competitive Regional Clusters. National Policy Approaches, Paris: OECD (OECD Reviews of Regional Innovation). Perrin, J.-C. (1988), ‘A Deconcentrated Technology Policy – Lessons from the Sophia-Antipolis Experience’, Environment and Planning C: Government and Policy, 6, 415–25. Piore, M. and C. Sabel (1984), The Second Industrial Divide: Possibilities for Prosperity, New York: Basic Books. Rees, J. (1986), ‘The Drawbacks of Research Parks’, Urban Resources, 3, 41–7. Rosenfeld, S.A. (1997), ‘Bringing Business Clusters into the Mainstream of Economic Development’, European Planning Studies, 5, 3–23.

322

Emerging clusters

Sargent, J. (1987), ‘Industrial Location in Japan with Special Reference to the Semiconductor Industry’, Geographical Journal, 153, 72–85. Saxenian, A. (1985), ‘Silicon Valley and Route 128: Regional Prototypes or Historic Exceptions?’, Urban Affairs and Annual Reviews, 28, 81–105. Saxenian, A. (1994), Regional Advantage, Cambridge, MA: Harvard University Press. Scott, A.J. (1988), New Industrial Spaces. Flexible Production Organization and Regional Development in North America and Western Europe, London: Pion. Scott, A.J. and M. Storper (1987), ‘High Technology Industry and Regional Development: A Theoretical Critique and Reconstruction’, International Social Science Journal, 112, 215–32. Segal, Quince & Partners (1985), The Cambridge Phenomenon: The Growth of High Technology Industry in a University Town, Cambridge, Segal, Quince & Partners. Sirbu, M.A., R. Treitel, W. Yorsz and E.B. Roberts (1976), The Formation of a Technology Oriented Complex. North American and European Experience, Cambridge/MA: MIT. SRI International (1984), ‘US Government Programmes and their Influence on Silicon Valley’, Report Prepared for the Ministry of Industry and Research, Government of France, Menlo Park, CA: SRI. Sternberg, R. (1995), ‘Supporting Peripheral Economies or Industrial Policy in Favor of National Growth? An Empirically Based Analysis of Goal Achievement of the Japanese “Technopolis” Program’, Environment and Planning C: Government and Policy, 13, 425–39. Sternberg, R. (1996a), ‘Government R&D Expenditure and Space: Empirical Evidence from Five Advanced Industrial Economies’, Research Policy, 25, 741–58. Sternberg, R. (1996b), ‘Reasons for the Genesis of High-Tech Regions – Theoretical Explanation and Empirical Evidence’, Geoforum, 27, 205–24. Sternberg, R. (1998), Technologiepolitik und High-Tech Regionen – ein internationaler Vergleich, Münster, Hamburg: Lit, 2nd ed. Sternberg, R. and C. Tamásy (1999), ‘Munich as Germany’s No.1 High Technology Region – Empirical Evidence, Theoretical Explanations and the Role of Small Firm/Large Firm Relationships’, Regional Studies, 33, 367–77. Sternberg, R., M. Kiese and L. Schätzl (2004), ‘Clusteransätze in der regionalen Wirtschaftsförderung’, Zeitschrift für Wirtschaftsgeographie, 48, 164–81. Storper, M. (1993), ‘Regional “Worlds” of Production: Learning and Innovation in the Technology Districts of France, Italy and the USA’, Regional Studies, 27, 433–55. Storper, M. and R. Walker (1989), The Capitalist Imperative – Territory, Technology, and Industrial Growth, Oxford: Blackwell. Sturgeon, T.J. (2000), ‘How Silicon Valley Came to Be’, in M. Kenney (ed.), Understanding Silicon Valley. The Anatomy of an Entrepreneurial Region, Stanford: Stanford University Press, pp. 15–47. Wallsten, S. (2004), ‘The Role of Government in Regional Technology Development: The Effects of Public Venture Capital and Science Parks’, in T. Bresnahan and A. Gambardella (eds), Building High-tech Clusters, Cambridge, New York: Cambridge University Press, pp. 229–79. Wells, P. (1987), ‘The Military Scientific Infrastructure and Regional Development’, Environment and Planning, A, 19, 1631–58. Zook, M. (2005), The Geography of the Internet Industry, Oxford: Blackwell.

How knowledge-intensive clusters emerge

323

APPENDIX 12A Table 12A.1

List of interviewees (mostly regional economists, economic geographers)

USA Atkinson, Robert D. (Washington/DC) Chubin, Daryl E. (Washington/DC) Goldstein, Harvey A. (Chapel Hill/NC) Hall, Peter (Berkeley/CA) Hicks, Donald A. (Dallas/TX) Hill, Christopher T. (Washington/DC Kasarda, John D. (Chapel Hill/NC) Kenney, Martin (Davis/CA) Little, William F. (Chapel Hill/NC) Luger, Michael I. (Chapel Hill/NC) Malecki, Edward J. (Gainesville/FL) Malizia, Emil E. (Chapel Hill/NC) Markusen, Ann (New Brunswick/NJ) Moriarty, Barry M. (Chapel Hill/NC) Rees, John (Greensboro/NC) Roberson, James O. (Research Triangle Park/NC) Saxenian, Annalee (Berkeley/CA) Scott, Allen J. (Los Angeles/CA) Siegel, Lenny (Mountain View/CA) Steinmueller, Ed W. (Stanford/CA) Storper, Michael (Los Angeles/CA) Tatsuno, Sheridan (Aptos/CA) Walker, Richard (Berkeley/CA) Japan Kawashima, Tatsuhiko (Tokyo) Matsubara, Hiroshi (Fukuoka) Yada, Toshifumi (Fukuoka) Yamamoto, Kenji (Tokyo)

England Buswell, R.J. (Newcastle) Charles, David (Newcastle) Cooke, Philip (Cardiff) Dunford, Mick (Brighton) Hart, Douglas (Reading) Herriot, Walter (Cambridge) Keeble, David (Cambridge) Lawton-Smith, Helen (Oxford) Martin, Ron (Cambridge) Masser, Ian (Sheffield) Oakey, Ray (Manchester) Rothwell, Roy (Brighton) Sayer, Andrew (Brighton) Simmie, James (London) Tyler, Peter (Cambridge) France Beaujeu-Garnier, Jaqueline (Paris) Benko, Georges (Paris) Bernardy, Michel de (Grenoble) Cavard, Jean-Claude (Paris) Chanaron, Jean-Jacques (Grenoble) Cohen, Jeanine (Paris) Decoster, Elisabeth (Paris) Dézert, Bernard (Paris ) Pecqueur, Bernard (Grenoble) Perrin, Jean-Claude (Aix-enProvence) Quéré, Michel (Sophia Antipolis) Rougier, Henri (Grenoble) Rousier, Nicole (Grenoble) Savy, Michel (Paris) Tabariés, Muriel (Paris)

Note: Locations refer to the affiliation at the time of the interviews; no expert interviews were conducted in Munich.

13.

Policy transfer and institutional learning: an evolutionary perspective on regional cluster policies in Germany* Matthias Kiese

1

INTRODUCTION

Comparatively late, but forcefully, the global cluster hype in economic development policy and practice has taken a firm hold of Germany. Most recently, this is evidenced not only by the federal government’s €600 million Spitzencluster (leading-edge cluster) competition, but also by a growing number of programmes and initiatives launched by state, regional and local governments. Germany is thus jumping on a bandwagon that was set in motion in Anglo-American and Nordic countries in the 1990s and is now rolling full steam over post-socialist, newly industrialising and developing countries (see Sölvell et al. 2003; Ketels et al. 2006). However, the present cluster euphoria is surging far ahead of our current theoretical and empirical knowledge of clusters. Rehfeld (2005) argues that with the surge of cluster policies in the mid-1990s, German academic research has first lost orientation and then track of structural policy. He observes a methodological shift towards case study research, a growing variety of typologies and the quest for identifying generic success factors from best practice cases. At the same time, academic disciplines like regional science or economic geography are still struggling to theorise the alleged benefits of clustering and to prove them empirically. However, they generally shy away from policy analysis and leave it to political scientists who, in turn, do not show a strong inclination towards issues of economic policy at the sub-national scale in general. The emerging phenomenon of cluster policy thus falls into a gap between established disciplines. For instance, economic geographers’ general lack of understanding of political-administrative systems and processes may be seen as a major reason why their policy advice often

324

Evolutionary perspective on cluster policies in Germany

325

remains unheard in policy and practice. Anglo-American economic geographers have recently initiated a lively discourse on the political and practical relevance and engagement of their discipline (see Martin 2001, 2006; Glasmeier 2006; Malecki 2006). Regarding globalisation, Dicken (2004, p. 5) noted that ‘over the years, geographers have developed a disturbing – even dysfunctional – habit of missing out on important intellectual and politically significant debates, even those in which geographers would seem to have a major role to play’. Since clusters are another natural core competency of economic geographers, this observation may well be extended to the issue of cluster policy. In search for greater policy relevance and, hence, public recognition, the embrace of public policy by evolutionary economics offers some fresh potential for cluster policy research (see Metcalfe 1995; Witt 2003; Moreau 2004). This chapter adopts an evolutionary perspective to analyse the diffusion of cluster policy and the accompanying processes of institutional learning. This diffusion may be conceptualised as repeated acts of policy transfer using multiple channels such as manuals, consultants, knowledge communities, or policy tourism. It also involves a significant degree of adaptation that requires institutional learning at the receiving end of the transfer. The spread of cluster policies leads to a convergence of institutional responses to the challenges of globalisation, structural change and increased locational competition, aided by best practice examples and specialised consultants acting as transfer agents. However, complete convergence between cluster policies is unlikely due to the structural and institutional diversity of regional preconditions. An unreflected transfer of experiences from outside thus appears inappropriate. This tension between convergence and diversity provides the background for our comparative study of cluster policies in three western German states and seven regional case studies at the sub-state level. The chapter is organised as follows. First, it is necessary to briefly outline our understanding of cluster policy and the value-added that might be expected from adopting an evolutionary perspective (section 2). It is then necessary to outline the various mechanisms of policy transfer and to develop a model of policy diffusion and institutional learning (section 3). Following an introduction to the project’s research methodology and its case studies (section 4), the main empirical findings on interregional policy transfer and institutional learning will be presented in section 5 to finally draw some conclusions and to identify further challenges for research on cluster policies.

326

2

Emerging clusters

CLUSTER POLICY IN EVOLUTIONARY PERSPECTIVE

Since the mid-1990s, the cluster concept enjoys continuing popularity with academics, politicians and economic development practitioners alike. Porter (1998, p. 197 f.) defines clusters as ‘geographic concentrations of interconnected companies, specialized suppliers, service providers, firms in related industries, and associated institutions (for example, universities, standards agencies, and trade associations) in particular fields that compete but also cooperate’. Despite – or due to – its rather broad character, this definition is the most common and may well serve as a common denominator of most alternatives (see Martin and Sunley 2003). The OECD working group on clusters took Porter’s definition one step further by highlighting the role of cross-business value chains (OECD 1999, 2001). However, this linear focus tends to neglect the systemic nature of production and innovation, as well as the horizontal and diagonal linkages within clusters. Besides the vertical dimension of the value chain, cooperation and competition between firms at the same stage of the chain provide another critical source of innovative dynamics. Furthermore, cluster firms exchange material inputs and knowledge in the lateral or diagonal dimension with service providers, universities and research institutes, as well as other companies and organisations. In sum, clusters may best be conceived as multi-dimensional value systems combining vertical, horizontal and lateral (diagonal) linkages. Within a cluster, the actions and interactions of individuals and organisations are governed by a specific set of norms and rules inscribed in formal and informal institutions, or ‘rules of the game’ according to North (1990). Further to this institutional dimension, any conceptualisation of clusters needs to acknowledge that a varying but potentially significant degree of a value system’s linkages extend beyond the cluster’s spatial confines to form its external dimension (see Bathelt et al. 2004). Leaving the conceptual fuzziness of clusters aside, regional cluster policies comprise all ‘efforts of government to develop and support clusters in a particular region’ (Hospers and Beugelsdijk 2002, p. 382). Cluster policy may hence be seen as a particular form of industrial policy targeting specific regional features and aiming at the development of certain building blocks already in place (for example specialised agglomerations, networks) into clusters, or at growing potential and latent clusters into working ones (see Enright 2003). From an evolutionary perspective, cluster policies emerge at the interface of hitherto largely unconnected established fields, such as industrial policy, science, technology and innovation policy, as well as regional and local economic development policy (see Boekholt and

Evolutionary perspective on cluster policies in Germany

327

Thuriaux 1999; Nauwelaers 2001). It may thus be expected that the interpretation of the cluster approach is critically shaped by past experiences in the respective field. For instance, applying the cluster concept in federal research policy will likely differ noticeably from an application to local economic development. Focusing on public agency, this concept of cluster policy differs from the wider term cluster initiative, in which cluster firms assume centre stage, supported by government and/or research institutes (see Sölvell et al. 2003). Within this broader concept, cluster policy may therefore be seen as a subset characterised by substantial state involvement through initiation, funding and/or governance. Somewhere along the spectrum of public–private partnerships, a line needs to be drawn to set cluster policies apart from private-led initiatives. In addition to this governance dimension that was already highlighted by Fromhold-Eisebith and Eisebith (2005), Kiese (2008b) develops six further dimensions that are needed to characterise and delineate cluster policies in empirical research, namely cluster reference (explicit vs. implicit), cluster orientation, complexity, coherence, institutionalisation and time. The more recent engagement of evolutionary economics with public policy issues has added some fresh insights into the role of the state. Summarised by Moreau (2004), the evolutionary perspective on public policy is both easily applicable and highly relevant to cluster policy. In a neoclassical world, the state is held responsible for stable institutions enabling markets to work properly, and to correct market failures as and when they occur. Good news for interventionists at first glance, evolutionary economics assigns a more pro-active role to public intervention stemming from a high degree of uncertainty about the desirable economic trajectories. Metcalfe (1995) stresses the task of increasing variety through the promotion of experimental behaviour, while Nelson and Winter (1982) demand the monitoring and encouragement of innovation. However, evolutionary reasoning also has some unpleasant news for advocates of public intervention. In neoclassical lines of thought, policy-makers are already hampered by imperfect information, most notably as principals as opposed to agents in constellations of asymmetric knowledge (see Ross 1973; Pratt and Zeckhauser 1991). In the case of cluster policy, policymakers are subject to multiple principal–agent relations, as they are better informed than their voters, but less knowledgeable than both consultants and bureaucrats responsible for policy design and implementation. The two latter groups may employ their superior knowledge to pursue their own goals at the expense of public welfare. While this public choice perspective assumes the existence of at least subjective probabilities quantifying the realisation of future steady states, the evolutionary policy-maker has to deal with bounded rationality and a radically uncertain environment

328

Emerging clusters

with punctual attractors instead of steady states. Further to the incentive problems of bureaucratic capture, lobbying and rent seeking, the evolutionary policy-maker faces a fundamental knowledge problem and has to deal with the irreversible and path-dependent nature of his decision. While the evolutionary perspective on public policy casts additional doubt on the capacity of governments to ‘pick winners’, it also holds a number of general recommendations for cluster promotion (see Boschma and Sotarauta 2007). Evolutionary reasoning advises policy-makers to identify and respect processes of emergence rather than trying to build clusters ex nihilo. They should concentrate on original regional strengths rather than follow the latest fashion in the global cluster promotion game. Cluster policy may increase the circulation of knowledge and the variety of information sources by enhancing connectivity and promoting knowledge exchange between firms, research institutions and policy-makers. To identify original cluster potential, open contests may be a useful approach to tap a region’s distributed knowledge base. In general terms, evolutionary cluster policy should help develop the capacity of regional organisations and strengthen the adaptability of regional institutions to changing environments. Taking the cluster life cycle into account, Menzel and Fornahl (2005) recommend that policy should support the achievement of critical mass through the specialisation around core competencies in the early stage of cluster development, but adjust their focus to increase variety, flexibility and adaptability as a cluster matures. Most and foremost, however, the evolutionary policy-maker must be a learning actor himself, as Groenewegen and van der Steen (2007) demand. To avoid lock-ins, interregional learning should not only be of concern to cluster firms, but also to policy-makers designing, implementing and adapting cluster strategies for their regions.

3

POLICY DIFFUSION AND INSTITUTIONAL LEARNING

The spreading of the cluster approach across time and space may alternatively be analysed from a diffusion or transfer perspective (see Lütz 2007). Policy diffusion focuses on general structural patterns such as the degree to and the speed with which a political innovation spreads, or on the systemic differences between early, late and non-adopters of a new concept (see Tews 2002). The first studies on policy diffusion were done for the states of the US in the late 1960s. Their results stressed the importance of spatial proximity (Walker 1969) and the frequency of interaction with political innovators as well as their embeddedness in networks (Gray 1973)

Evolutionary perspective on cluster policies in Germany

329

as critical determinants for policy diffusion and adoption. Complementary to the macro perspective of policy diffusion, policy transfer focuses on the individual mechanisms, contents and results of the process. It was developed by British policy research in the 1990s, mainly by Rose (1991, 1993) and Dolowitz and Marsh (1996, 2000), and has been in vogue since among political scientists. However, research in this field primarily focuses on international policy transfer and tends to neglect subnational variations and processes of learning that are critical for regional cluster policies. Research on policy transfer should distinguish different degrees or intensities of transfer, mechanisms and channels, as well as the determinants of successful transfer (see Lütz 2007). Both diffusion and transfer can take place vertically between jurisdictions of different spatial scales (top-down or bottom-up), or horizontally between jurisdictions of the same spatial scale. 3.1

Degree of Policy Transfer

An extreme case of policy transfer with the highest possible intensity is the unmodified copying of policies from another country or region, which is hardly conceived as a realistic option (see Rose 1991). It appears more realistic to use someone else’s policy as a best practice model and adapt it to one’s own context. Instead of one single model, policy elements from two or more countries or regions may be combined to suit one’s own demands. Completing this scale of decreasing transfer intensity, policies may be developed largely endogenously with only some inspiration from other nations’ or regions’ experiences. 3.2

Mechanisms of Policy Transfer

The literature on policy diffusion and transfer is divided over the mechanisms of the import and export of policies. Lütz (2007) suggests a classification into five transfer mechanisms, which often appear either simultaneously or sequentially in practice. Policy transfer may be hierarchically enacted when a superordinate political actor forces a subordinate unit to adopt a certain policy. Alternatively, transfer may be caused by locational competition for mobile factors of production, which increases the pressure on national and regional governments to provide efficient and innovative solutions (see Siebert 2006) and may thus accelerate the spread of accepted best practices. Furthermore, policy transfer may result from binding negotiations between national or regional states, which may lead to a combination of models, or to convergence towards one model perceived as superior. A similar but not binding transfer mechanism is

330

Emerging clusters

deliberation, in which the interpretations, preferences and interests of partners may converge during discussion as a result of collective learning. Finally, policy-makers may adopt practices developed elsewhere by way of unilateral policy-shopping, a transfer mechanism akin to ‘voluntary transfer’ (Dolowitz and Marsh 2000), or ‘lesson drawing’ (Rose 1993). The policy transfer literature stresses that especially under ‘evolutionary’ conditions of complexity and uncertainty, policy-makers tend to reduce risk by taking over models that have already proven successful elsewhere. 3.3

Channels of Policy Transfer

While the mechanisms of policy transfer focus on how policy imports and exports are governed, channels of transfer explain how knowledge about cluster policies is actually transmitted. The policy transfer literature distinguishes six different channels (see Lütz 2007): 1.

2.

3.

4.

Knowledge about clusters and cluster initiatives can be obtained from the vast and rapidly expanding cluster literature. As indicated earlier, this literature is biased towards best-practice cases in order to extract critical success factors and to provide practical advice for the management of clusters and networks. Manuals for cluster development are available in most countries with high levels of cluster policy activity nowadays. Alternatively, knowledge relevant to cluster policies may be transmitted through the professional mobility of key individuals, or change agents according to Rogers (2003). These may be either politicians with a particular enthusiasm for the cluster concept, or knowledgeable practitioners promoting a climate of change and highlighting alternatives to exiting solutions (see Lütz 2007). Consultants may also play a key role as transfer agents, as Stone (2004) shows in her study on international relations. In the case of cluster policies, Michael E. Porter obviously assumes a central transfer role by single-handedly creating the ‘cluster brand’ and selling it worldwide (see Martin and Sunley 2003). On the back of his success, some established management consultancies have entered the playing field of regional development advice, competing with an uncharted number of small specialised consultants with predominantly national or even sub-national outreach. Since the mid-1990s, the interregional and international exchange of knowledge is discussed with reference to epistemic communities or communities of practice. The former explicitly aim at adding new knowledge to a highly codified knowledge base, while collective

Evolutionary perspective on cluster policies in Germany

5.

6.

3.4

331

learning rather occurs as a by-product in communities of practice which rely on implicit knowledge to a much greater extent.1 As the clear-cut distinction between codified and implicit knowledge is increasingly called into question, the boundaries between epistemic communities and communities of practice also become increasingly blurred (see Cohendet and Amin 2006), so that both concepts may best be viewed as goalposts marking a continuum of knowledge communities. For cluster practitioners, The Competitiveness Institute has emerged as the leading platform for the exchange of knowledge on cluster management. In 2008, the TCI’s eleventh annual conference drew around 400 participants. Another widespread channel of policy transfer is the travelling of policy-makers, or journeys of larger delegations comprising practitioners as well (see Hospers and Beugelsdijk 2002). Silicon Valley is obviously the most popular destination of such economic development policy tourism, thus contributing to the spread of the Silicon Valley model of economic development: ‘Presidents, ministers, and dignitaries come in pilgrimage here, in well-publicized delegations that aim to capitalize the visit in social prestige or political votes back home’ (Castells and Hall 1994, p. 12). The transfer channels mentioned so far may all lead to the establishment of formal and informal channels of communication, which can be used to exchange knowledge on strategic concepts and instruments of cluster promotion. Determinants of Successful Policy Transfer

Whether policies can be successfully transferred from one national or regional context to another depends on endogeneous and exogenous variables, as well as on properties of the policy that is to be transferred (see Lütz 2007). Endogenous factors refer to the cultural, institutional and socio-economic proximity of the sender and the receiver of the transfer. Ceteris paribus, a smooth and successful transfer may be expected between countries or regions of comparable level of development, similar culture and institutional settings. Conversely, difficulties are likely to occur when transferring policies between developed, newly industrialised, developing and post-socialist countries, or between liberal and co-ordinated market economies in a variety of capitalism framework (see Hall and Soskice 2001). On the other hand, exogenous factors comprise the frequency of interaction between the partners involved, their integration into knowledge communities as well as the number and quality of transfer agents as intermediaries. Finally, transferability depends decisively on the

332

Emerging clusters

characteristics of the policy itself. Policies are easily transferred if they are simple, highly visible and show a clear causality between means and ends. On the other hand, policies bearing high potential for conflict or having distributional impact are considered much more difficult to transfer (see Kern 2000; Knill 2005). This review of the determinants casts more doubts than hopes on the transferability of cluster policies which are characterised by highly complex and hardly measurable links between causes and effects. Cluster policy is poorly visible when dealing with the development of networks, for example: ‘the lay-voter will find it much more difficult to assess a dysfunctional cluster initiative than substandard garbage collection’ (Duranton 2008, p. 26). It is by definition selective and redistributive as it discriminates in favour of cluster members as opposed to non-members which are excluded from promotion. From a public choice perspective, conflicts may be expected whenever organised groups are excluded from the benefits, which in turn creates dangers of cluster policy being captured by rent-seeking minorities (see Olson 1965; Krueger 1974). 3.5

A Model of Interregional Institutional Learning

While the literature on policy transfer and diffusion clearly advances our understanding of the rapid spread of cluster policies, more detailed insights may be gained from conceptualising policy diffusion as separate and sometimes repeated acts of institutional learning. This perspective draws on the concept of the learning region, including its amendments and applications such as learning clusters (see Florida 1995; Hassink 2005). In their knowledge spiral model, Nonaka and Takeuchi (1995) suggest that new knowledge exclusively emerges in implicit form and context-bound. To apply this knowledge in another context, it needs to be decontextualised and codified through terms, models or theories. Once codified, this knowledge can only be used once it is recontextualised and adapted to new circumstances, which in turn requires implicit and context-specific knowledge once again. Based on the knowledge spiral, Hassink and Lagendijk (2001) develop a cyclical model of interregional institutional learning, which Lagendijk (2001) applies to the case of mixed land-use in the Netherlands. Applying their model to cluster policy suggests that regional cluster concepts are developed in close interaction between predominantly implicit and contextualised regional knowledge on the one hand and largely codified conceptual and methodological knowledge on the other. The latter is accumulated through the decontextualisation and codification of experiences from various case studies and thus becomes embodied in scholarly literature,

Evolutionary perspective on cluster policies in Germany

333

generic explicit Cluster approach Decontextualisation Codification

(Re-)Contextualisation Regional cluster concept

Decoding Adaption

accumulated experience, learning by doing (‘laboratory’) local-specific tacit Source:

Based on Hassink and Lagendijk (2001) and Lagendijk (2001).

Figure 13.1

Development of cluster policies as interregional learning

practical guidelines, specialised consultants and their organisations, or in the relations connecting knowledge communities. Applying this stock of knowledge for the development of a new cluster policy requires de-coding and adaptation to a specific context, which is realised through cognitive, social and institutional learning (see Hassink and Lagendijk 2001). At the interface between codified and context-specific knowledge, discourse coalitions form between local change agents and non-local transfer agents, referred to as relay agents by Lagendijk (2001). From an evolutionary perspective, the conceptual action space (‘laboratory’) in which a regional cluster policy develops from the interplay of local and non-local knowledge may be conceived as an arena of variation and preliminary selection. It is the place where conceptual variety emerges, and an initial selection is made between alternative solutions. Alternative channels of policy transfer, addressed as ‘interregional relay centres’ by Lagendijk (2001), cause further selection and convergence by filtering out perceived best practices and extracting generic strategies and conceptual elements from them. To continue the evolutionary analogy, the application of this codified knowledge to develop a cluster policy in a new context may be seen as a mutation of the generic cluster concept. Applying their cyclical concept of interregional institutional to regional development concepts in general, Hassink and Lagendijk (2001) as well as Lagendijk (2001) indicate that some concepts may develop into fashions through selection and convergence (Figure 13.1). The model thus appears well

334

Emerging clusters

suited for the analysis of cluster policies between the countervailing forces of convergence and interregional variety. Hassink and Lagendijk (2001) conclude that some of the mechanisms of learning summarised in their model are still insufficiently understood and thus present ‘a key challenge for future work’ (ibid., p. 81). They specifically address the need to establish how some policies eventually become perceived as successful and turn into objects of policy transfer, and how recipients of policy transfer change their attitudes and institutional environments to increase their capacity for learning. They generally advocate research into how new concepts for regional development diffuse and evolve at different stages of the learning cycle, for example how they are decontextualised and become part of the generic pool of knowledge on regional development. The answers to these research questions should be fed back as reflexive knowledge to advance conceptual development in the future.

4

RESEARCH DESIGN AND CASE STUDIES

Cluster policy is a relatively young, complex and theoretically poorly structured phenomenon that still evades quantitative research by and large. For instance, a comprehensive mapping of all cluster policies in a country like Germany is still absent not just for its high costs, but most importantly due to the lack of a commonly agreed and operational definition of cluster policy. Hence, cluster policy research still needs to rely on a qualitative research design. The interpretation and application of the cluster idea in different spatial contexts as well as the policy transfer and learning processes involved were central objects of a research project including 110 semi-standardised interviews with 134 cluster policy experts, which were conducted between August 2006 and August 2007. The sample of interviewees comprised 60 practitioners in ministries and economic development agencies, of which 19 explicitly classified themselves as cluster managers, 10 consultants and 75 independent observers.2 Bavaria, North Rhine-Westphalia (NRW) and Lower Saxony account for 53, 44 and 35 interviewees, respectively. A further 13 experts are active in more than one state or at the supra-state level more generally. The interview transcripts were complemented by academic literature where available, brochures, political documents such as parliamentary and council notes, as well as unpublished concepts, expertises and evaluations to form the basis for ten case studies of regional cluster policy at the state and sub-state level. Case studies have been selected according to the seven dimensions of cluster policy introduced in section 2. In the governance dimension, the focus on cluster policy rather than cluster initiatives more generally

Evolutionary perspective on cluster policies in Germany

335

requires a significant degree of public agency in the initiation, funding and operational governance of the effort. Despite embracing the cluster notion, policies under study do not have to use the cluster term explicitly – in many German regions there is a tendency to adopt more ‘neutral’ terms like fields or networks of competence instead. Although cluster concepts often cite Porter’s definition of a cluster, there is generally little deeper theoretical grounding, and clusters tend to be understood as organised networks by practitioners (see Kiese 2008b). However, the selected cases are all cluster-oriented when measured by the usage of cluster-specific versus generic economic development tools. They are complex in combining wider sets of instruments for cluster promotion and coherent by uniting different policies and regional stakeholders within a single programme. Institutionalisation may vary from rather loose associations to the dedicated cluster organisations, but all cases are sufficiently mature to allow for some at least preliminary evaluation. The surveyed cases range from the state to the municipal level. At the state level, NRW, Bavaria and Lower Saxony were chosen to roughly represent three economically distinct types of region (see Figure 13.2). While structural policy in NRW was for decades dominated by the challenge to promote structural change in the Ruhr area, Bavaria stands for the opposite case of a late industrialised state with a strong presence of high-tech industries and firms. With its manufacturing sector shaped by Volkswagen (VW) and its supplier network, Lower Saxony appears quite unlike these two extremes but rather falls into the ‘grey mass’ category of regions often neglected in regional studies. This choice of states is meant to create structural, but also institutional and political variety for the interregional comparison of cluster policies. Our choice is furthermore restricted to Western Germany as the new Länder deviate too strongly because of their socialist legacy and a more direct involvement of federal government cluster programmes (see Dohse 2007; Koschatzky and Lo 2007). Within the three states chosen, research focused on state-level cluster policies as well as seven case studies at the sub-state level selected on the basis of published research and initial interviews. 4.1

North Rhine-Westphalia

Based on experience from its regionalised structural policy from the 1980s, the NRW government started promoting its pilot network programme PROFIS in 1993, which is now seen as the antecedent of its fully-fledged cluster policies that were to follow from 2000 after the election of the state’s Minister President Wolfgang Clement (Kompetenzfeldpolitik, see Rehfeld 2006). Following a change in government in 2005, the new

336

Emerging clusters 0

50

100 km

LOWER SAXONY hannoverimpuls GmbH Wolfsburg AG

NORTH RHINEdortmund-project

projekt REGION BRAUNSCHWEIG GMBH

“Triangle”: kompetenzhoch3

WESTPHALIA

Central Franconia

Regensburg

BAVARIA

Cartography:

Figure 13.2

Stephan Pohl.

Case study regions

Evolutionary perspective on cluster policies in Germany

337

conservative-liberal coalition publicised an interministerial cluster policy as part of its innovation strategy in March 2007. During the funding period ending 2013, €635m of European Regional Development Fund (ERDF) Objective 2 funding was earmarked for competitive tenders in 16 statewide clusters predefined by the ministry, an open RegioCluster contest, as well as some cross-sectional competitions (see MWME 2006). Below the state level, the most ambitious cluster policy effort can be found in Dortmund, a major city on the eastern edge of the Ruhr conurbation that had already embraced structural policy since the establishment of its innovation centre and technology park in the mid 1980s. In 2000, the city council approved a cluster strategy devised by international management consultancy McKinsey & Co. targeting IT, micro technologies and e-logistics to compensate for the demise of coalmining, steelworks and breweries.3 As a second sub-state case within NRW, the kompetenzhoch3 collaboration between the city triangle of Wuppertal, Solingen and Remscheid was also driven by the legacy of early industrialisation and a dire need for structural change. Since 2001, the three municipal economic development offices established a division of labour based on five fields of competence, namely automotive, metal processing, product development and design, event management and communication, as well as health and personal care (see Dewald 2006). While Dortmund’s strategy entails a radical break with the past, kompetenzhoch3 includes an injection of design competencies into the remains of the centuries-old cutlery district of Solingen and Remscheid which was famously described by both Marshall and Porter (see Bathelt 1998; van der Linde 1992). 4.2

Bavaria

In Bavaria, the state government launched its recent cluster initiative in February 2006 as a new stage of its technology policy that was initiated in 1994, fuelled by a massive selling-off of stakes in utilities. After €4bn of privatisation revenues were largely invested in R&D infrastructure, the new cluster initiative was endowed with a more modest €50m to establish and fund the management of 19 clusters predefined in top-down fashion. Each of the 19 cluster management units typically consists of a full-time manager, an unsalaried speaker for representation, and a secretary. Public funding was announced to decrease over five years to put pressure on cluster managements to eventually become self-sustaining (see StMWIVT 2006). At the sub-state level, the northern district of Central Franconia devised its first cluster strategy in response to the decline of its dominant electrical engineering sector which culminated in massive employment losses at the traditional company Grundig. Initial efforts were incorporated in the

338

Emerging clusters

more coherent Nuremberg Programme in 1994, which was followed by a consensual perspective report originally drafted and signed in 1998, and renewed in 2005 (see Neumann 1996; IHK Nürnberg 2005). These documents contained a set of fields of competence defined as clusters,4 which are promoted through independent competence initiatives from as early as 1994 (see Heidenreich 2005). Contrasting the industrial decline experienced by Central Franconia, the city of Regensburg witnessed a rather exceptional late industrialisation from the 1980s following the attraction of large manufacturing establishments like BMW or Siemens in what was previously a rural backwater. In Regensburg, municipal cluster policy did not emerge in response to some perceived crisis, but to federal government contests such as the BioRegio initiative5 leading to the establishment of the BioRegio Regensburg in 1996 and the BioPark incubator in 1999. A similar top-down stimulus triggered the establishment of the Strategic Partnership for Sensor Technology in 2003, a concept transferred locally to the field of IT security in 2006 (see Stadt Regensburg 2003; Diefenthal 2006; IT-Speicher 2008). 4.3

Lower Saxony

Unlike NRW and Bavaria, this state’s government does not have an explicit and coherent cluster policy, but adapted a McKinsey & Co blueprint to revamp its regional structural policy in 2004. Its new Regional Growth Concepts (RGCs) are designed to stimulate the bottom-up development of cluster concepts from the regions (see Kiese 2008a). The approach is modelled after the abovementioned dortmund-project and the older AutoVison concept developed by McKinsey on behalf of VW to reverse the economic fortunes of their headquarter and company town in Lower Saxony, Wolfsburg. Next to cutting local unemployment by half within five years, which was achieved with the help of a cyclical upswing, the local cluster organisation Wolfsburg AG aims at transforming the once single-plant location into a self-augmenting cluster of the mobility industry in the long run. To reduce the overwhelming dependency on one single employer, the concept proposed the development of new interlinked clusters of IT, leisure and tourism, as well as health businesses (see Sternberg et al. 2004). As a prototype of its newly-conceived RGCs, the state government teamed up with the city and region of Hanover to fund the development of a cluster-based strategy to improve the competitiveness of its capital region by McKinsey & Co in 2002. In March 2003, local and regional governments jointly incorporated hannoverimpuls to pursue their strategy which focused on the development of automotive, IT, life sciences, optical

Evolutionary perspective on cluster policies in Germany

339

technology and manufacturing technologies into working and interlinked clusters (see Kiese 2008c). Meanwhile, the state co-funded the development of a similar concept by McKinsey & Co. for the Braunschweig region, which has been pursued by the projekt REGION BRAUNSCHWEIG GMBH since early 2005 (see Prätorius 2004).

5 5.1

POLICY TRANSFER VERSUS PATH-DEPENDENT LEARNING: SELECTED FINDINGS State Policies: NRW vs Bavaria

Frequent references in policy documents and interviews show that both NRW and Bavaria have been inspired by Upper Austria in their policy design. The Austrian state made an early decision to invest substantially in cluster promotion and is now regarded as the best practice example within the German-speaking countries. However, Upper Austria’s cluster policy itself was in turn inspired by earlier experiences with automotive cluster promotion in another Austrian state, Styria (see Steiner and Hartmann 1999; Tödtling and Trippl 2004; Fromhold-Eisebith 2007). Although interaction with Upper Austria is more intense in Bavaria due to spatial and cultural proximity as well as cross-border cooperation, NRW ministries have also studied the Austrian case closely, including business trips of practitioners. Despite its status as a role model, Upper Austria’s influence on the cluster policies of NRW and Bavaria can hardly be seen as more than inspirational. With some goodwill, the establishment of cluster management units as platforms for co-operation or the practice of degressive public funding may be interpreted as cases of unilateral policy shopping, had it not already become commonplace in cluster promotion anyway. Furthermore, interviewed practitioners stressed that Austria’s smaller size limits the transferability of experiences, as does the volume of funding committed in Upper Austria that German states are neither willing nor able to match, except for Bavaria due to its rather unique privatisation thrust. The latter questions the assumption of Hospers and Beugelsdijk (2002) that the relatively homogenous per capita income of developed economies implies a comparative resource endowment of their cluster policies. Instead, political philosophies, priorities and opportunities seem to play a critical role here. Further to a common source of inspiration, NRW and Bavaria are also more strongly connected by mutual learning than most other German states. As early as 1997, Bavaria drew on experiences from NRW and other states in the design of its innovation and co-operation initiative for the

340

Emerging clusters

automotive industry, BAIKA. More recently, NRW’s recent cluster policy was inspired by the establishment of cluster management units in Bavaria, but added the rather innovative element of competitive funding based on own experiences and, more importantly, EU funding requirements. However, policy learning between the two largest German states builds not only on mutual observation, but also on direct interaction that can be traced back to 1998 at least when NRW’s minister president Wolfgang Clement came into office. Despite representing competing parties, Clement and his Bavarian colleague Edmund Stoiber agreed on close co-operation and frequent consultations of their leading civil servants, and even held joint cabinet meetings. This partnership included the informal knowledge exchange by ministerial bureaucrats in charge of cluster policy. This unlikely alliance of political entrepreneurs (see Facchini 2006) rested on similar biographies and traits, but also on shared political interests vis-à-vis federal government and the EU. At a more general level, informal meetings of bureaucrats and a joint federal–state committee for research and technology provide opportunities for collective learning at the state level. In sum, horizontal policy transfer between the states is rather weak and mainly limited to mutual inspiration. The two case studies illustrate, however, that path-dependent institutional learning leaves a much greater imprint on the design and implementation of cluster policies. Large ministerial bureaucracies at the state level act as repositories of experiencebased knowledge. This is best evidenced by the breadth and continuity of NRW’s cluster portfolio: When the new state government came into power in 2005, it announced a thorough examination of previous policies and a significant reduction in the number of targeted clusters. After one and a half years of internal discussion and policy formulation, the government came up with a list of 16 clusters in NRW to replace their predecessors’ portfolio of twelve fields of competence for the Ruhr area. It can be shown that some clusters were artificially split up to divide responsibilities between rival ministries. In the end, NRW’s cluster portfolio did not only grow, but also contained all major industries and technologies that were previously supported through the fields of competence policy and other schemes. This strong continuity illustrates the power of the ministerial bureaucracy. State ministries are not only repositories of knowledge and arenas of incremental learning, but also represent the interests of specific industries, fields of science and technology, and political programmes which tend to develop their own dynamics (see Olson 1965). In contrast to NRW with its 2005 change in government, Bavaria’s ruling conservative party CSU has been in power since 1957 without any interruption. Due to this strong political continuity, path-dependent policy learning is even more evident than in NRW. Bavaria’s recent

Evolutionary perspective on cluster policies in Germany

341

cluster initiative builds upon the state’s technology programmes Offensive Zukunft Bayern and High-Tech-Offensive (HTO). Since 1994 and 1999, respectively, the state has invested €4.15bn into its science and technology infrastructure, with a strong emphasis on developing international centres of excellence that had a clear regional focus. To counter concerns about rising spatial inequalities, a certain share of the funds was earmarked for peripheral and lagging regions. At the operational level, Bavaria’s cluster initiative draws heavily on existing organisations from its earlier technology policy thrust, most notably the state-owned Bayern Innovativ society for innovation and technology transfer and its prototype automotive network BAIKA, both established in the mid-1990s. Continuing concerns about spatial equity led to the integration of the regional management instrument from Bavaria’s spatial development policy as a path-dependent supplement to its cluster policy. 5.2

Cluster Policies at Sub-state Level

Below the state level, the analysis of alternative channels, processes and intensities of cluster policy transfer indicates an overall low degree of transfer to start with. Interregional policy learning is generally restricted to inspiration and some elements of combination. As at the state level, path-dependent intraregional learning appears to play a much greater role. However, there is one notable exception to this general pattern. Influenced by management consultancy McKinsey & Co., the cluster policies of Wolfsburg, Dortmund, Hanover and Braunschweig display higher transfer intensities of copying and adaptation. Among the alternative mechanisms of policy transfer, these four cases may be classified as unilateral policy shopping, warranting a more detailed discussion at the end of this section. However, policy transfer was found to be much weaker for the other three sub-state case studies, with mutual observation embedded in locational competition as the main mechanism. As far as the channels of policy transfer are concerned, cluster literature appears to play a negligible role among policy-makers and practitioners alike. References to cluster literature are overwhelmingly limited to Porter’s cluster definition, and manuals dedicated to network and cluster development were largely unknown and did not play a role in the design and implementation of cluster policies. Within our sample, Wolfsburg was the only case where a manual for cluster development was actually used: when reorganising their cluster management, the Wolfsburg AG consciously adapted the functions of Clusterpreneurs and Cluster Engineers from the Cluster Policies Whitebook (Andersson et al. 2004; see Heinecke 2008).

342

Emerging clusters

Compared to literature, the mobility of key personnel appears to be a more relevant channel of policy transfer. Some of the younger cluster organisations surveyed stated that they had consciously hired staff from older cluster organisations, mainly to acquire procedural knowledge, for example on start-up contests. However, policy transfer through the movement of people does not only take place at the operational level. A notable case within our sample is a key individual who worked in the city of Cologne’s urban development department where he acquired cluster policy experience as managing director of the city’s MediaPark. After moving to Nuremberg in 1992, he injected his openness towards the cluster concept into the Nuremberg Programme and thus became one of the founding fathers of Central Franconia’s cluster policy. In 1997, this change agent was appointed head of the city of Dortmund’s economic development division where he focused business support on specific industries and accompanied the development and implementation of the dortmund-project until his retirement in 2004 (see Küpper 2005; Küpper and Röllinghoff 2005). Through his openness and enthusiasm towards the cluster concept, which was certainly not harmed by his academic education as an economic geographer, this change agent thus left a trace of cluster policies linking the different stations of his professional career. However, this illustrative case is exceptional in our sample of cluster policies; the transfer of operational staff is more common, but quantitatively still rather limited.6 Recent literature emphasises the potential role of knowledge communities as channels for policy transfer. Despite some ad-hoc national conferences, professional associations like the annual congress of German economic development professionals have already dealt with cluster policies, albeit not on a frequent basis. Due to spatial and relational proximity, economic development practitioners’ exchange on clusters is more frequent in some states, notably in NRW where the government’s cluster contests create additional demand for cluster strategies in the regions to qualify for ERDF funding. However, there is no national platform to institutionalise knowledge exchange on cluster promotion, and German practitioners are clearly underrepresented in international communities of practice such as TCI, highlighting the largely untapped potential of this learning channel for cluster policies in Germany. Interviewed practitioners and observers cite budget constraints and language barriers as the main obstacles to a more active participation in international communities of practice. Despite these constraints, business trips by economic development professionals to learn from successful cluster policies first-hand are quite common during the concept development stage. However, practitioners

Evolutionary perspective on cluster policies in Germany

343

stated problems justifying the costs of such policy tourism to their local governments and the public. Furthermore, there is a general scepticism towards the transferability of cluster policy experiences made elsewhere and hence towards the potential for interregional policy learning in general. To complete our survey of transfer channels, the low overall intensity of interregional policy learning means that the secondary channel of informal exchange based on communication between individuals plays a certain role between state ministries (see section 5.1), but rarely occurs between our regional case studies. 5.3

Consultants as Transfer Agents: The McKinsey & Co. Case

In our sample, the cases of Wolfsburg, Dortmund, Hanover and Braunschweig provide an outstanding example of policy diffusion via consultants as transfer agents. When international management consultancy McKinsey & Co. was commissioned by VW to develop a concept to revitalise its ailing company town of Wolfsburg in 1998, they could draw on relevant experiences from consulting projects in the US. This knowledge was decontextualised and transferred to Wolfsburg via the consultancy’s knowledge management system. For the development of Wolfsburg’s AutoVision concept, this codified knowledge was combined with the accumulated experience of local experts who helped the consultants draft the concept in a joint team over a few months. However, the local knowledge base was not entirely implicit in that it included earlier reports and economic development strategies such as the vision of the Braunschweig region as a centre of competence for transportation (Verkehrskompetenzregion, see Lompe et al. 1996). When the AutoVision concept coincided with a favourable business cycle, McKinsey & Co. went on to sell it as a ‘plan for German job creation’ (Heuser et al. 2001) and as a showcase for the acquisition of further projects. When ThyssenKrupp was pressured to compensate the city of Dortmund for the closure of its steel plant, their key customer VW demanded that they set up a manufacturing facility in Wolfsburg (see Ziesemer 2004). This is how the steelmaker became aware of the AutoVision approach which entails the attraction of suppliers to Wolfsburg. In October 1999, ThyssenKrupp commissioned McKinsey & Co. to develop a similar concept in Dortmund to strengthen regional competitiveness through cluster development, which led to the establishment of the dortmundproject in May 2000. When the state ministry for the economy of Lower Saxony became aware of McKinsey’s work in Wolfsburg and Dortmund, it developed plans to apply this approach with its capital region of Hanover as a testing ground. When developing the cluster concept for Hanover, the

344

Emerging clusters

McKinsey-led project team of local experts went to Dortmund for a presentation of the dortmund-project. Some practitioners who were involved in this early stage confirm the consultants’ strong inclination to follow their blueprints applied in Wolfsburg and Dortmund (see Kiese 2008d). Originally, the government of Lower Saxony had planned to extend McKinsey’s cluster approach to all its regions. However, elections in February 2003 handed the Ministry of Economic Affairs back to the liberal democrats who declared the RGCs as a non-binding offer open to, but no longer mandatory for, all regions. Three of them accepted the offer of state co-funding for concept development, which led to the establishment of new cluster-oriented economic development agencies in the Weserbergland, Braunschweig and Süderelbe regions (see Kiese 2008a). The Braunschweig project purposefully learned from McKinsey’s showcase projects in Dortmund and Hanover how to employ and guide the consultants more effectively to accommodate their local interests. Nevertheless, practitioners again report the consultants’ strong push to apply their blueprints, indicating a systematic struggle between generic and context-specific knowledge. Figure 13.3 summarises our evidence of path-dependent cluster policy learning in lines and interregional policy transfer indicated by arrows. Of the degrees, mechanisms and channels of transfer discussed in the literature, only a few could be detected in our case study research of cluster policies. Inspiration as the lowest possible degree of transfer is most common, while initial attempts at copying by consultants were gradually eroded in the process of implementation. Although policy tourism does play a certain role, consultants and the mobility of key individuals are the most effective channels for transfer, but they are rather singular phenomena. As far as mechanisms are concerned, the top-down impetus of policy contests does not fit into the classification proposed by the policy transfer literature (see Lütz 2007). However, the latter draws on analyses of international policy transfer, while the emerging mechanism of policy contests is confined to the federal and state level, and has indeed been applied to cluster policy at the sub-state level by the Stuttgart region (see Sautter 2004). Compared to path-dependent policy learning, however, interregional transfer is relatively sporadic and offers policy and practice ample opportunities for further learning.

6

IMPLICATIONS FOR POLICY AND RESEARCH

Our theoretical discussion has shown that evolutionary policy-makers should first and foremost be learning actors. In the case of cluster policy,

345

Figure 13.3

Policy transfer and path-dependent learning in cluster policy: selected German case studies

Notes: Governance levels: A – supranational, B – national/federal, C – state, D – sub-state 1 SP = Strategic Partnerships pilot project, 2 SP = Strategic Partnership Path-dependent learning connects programmes parallel to the time axis in each line.

346

Emerging clusters

however, the transferability of experiences is restricted by the complex, largely invisible and potentially discriminating nature of the policy. Case study evidence illustrates that despite the variety of mechanisms and channels available for transfer, interregional policy learning in German cluster policy is rather limited. If policy transfer occurs, it is mainly confined to unilateral policy shopping which uses best practice examples like Upper Austria as a source of inspiration. In contrast, path-dependent learning appears much more influential on the way cluster policies are designed and implemented at the state and sub-state level. At first sight, the series of cluster concepts developed by McKinsey & Co. appears as an exceptional case of intensive policy transfer through copying and adaptation. These cases serve well to illustrate not just policy diffusion through consultants as transfer agents, but also the interactive development of policy concepts at the interface of contextspecific local and codified generic knowledge. However, evidence also shows that especially at the sub-state level, interaction with non-local generic knowledge is not a continuous phenomenon. The channel of policy transfer through consultants can only be afforded for a few months during concept development, and they were expected to develop a concept to guide cluster policy for a decade. In reality, considerable adjustments had to be made right from the start once the consultants’ job was done, which means that learning from generic knowledge soon became diluted by path-dependent experience. As policy-makers committed insufficient resources to the proper analysis of original cluster potential in their regions, cluster organisations resorted to generic instruments or even acquired new tasks7 which led to a gradual loosening of their initial cluster focus. These findings have important repercussions on the model of interregional institutional learning proposed by Hassink and Lagendijk (2001). Its application to case studies of cluster policy in Germany suggests that interaction with generic non-local knowledge is discontinuous, and interregional knowledge inputs soon tend to become dominated by pathdependent learning. Hence, the cyclical model should be ‘evolutionised’ by embedding the local knowledge base in path-dependent processes of incremental and cumulative institutional learning. Path-dependency of institutional change is a main theme of institutional economics (see North 1990) and corresponds with the structural inertia and preferences of bureaucracies for ‘proven solutions’ (Franke 2000, p. 104) stated by public choice theory. The relative importance of the learning cycle and the arrow of path-dependent incrementalism may vary but in our case studies the arrow tends to dominate either throughout, or once an impulse from outside has faded.

Evolutionary perspective on cluster policies in Germany

347

Particularly at the local scale, the practice of cluster promotion is shaped by the accumulation of implicit knowledge through cumulative learning by doing, or ‘coagulated experience’ (Brödner 2003, p. 150). Practitioners emphasise that their day-to-day duties leave little room for continuous knowledge sourcing and concept development. As a consequence, incrementalism and muddling through take reign. Regional cluster policies tend to be inward-looking and are thus susceptible to lock-ins and rent seeking. The potential offered by interregional learning remains largely untapped due to budget constraints and a lack of awareness. Policy transfer literature does support the widely held impression that the transferability of cluster policies is limited, but weak forms of inspiration and combination do exist and indicate that there is significant potential for interregional institutional learning on cluster policies. Evolutionary policy makers should thus promote awareness for the benefits of interregional and institutional learning without concealing the limits to strong forms of transfer like copying or adaptation. Practitioners need incentives to engage more actively in international communities of practice, and a national platform for continuous exchange could facilitate knowledge transfer. The decision to host a first national cluster congress in Leipzig in autumn 2008 is certainly a step in the right direction, as is the increasing use of (open) contests as an ‘evolutionarily correct’ means of cluster identification. This chapter could only marginally indicate the potential that evolutionary thinking offers for the analysis of cluster policies. More research of this kind should clearly be encouraged to increase methodological variety. For instance, more detailed single-case studies are needed to explore the potential of this novel perspective more fully. On the other hand, international comparisons of cluster policies and the analysis of international cluster policy transfer could increase the variety of contexts in which senders and receivers of policy transfer are embedded. In between, the working of knowledge communities like TCI and their actual relevance for the transfer of knowledge on cluster promotion remains another uncharted territory for research. Evolutionary or not, the advancement of cluster policy research critically depends on opportunities for interdisciplinary research. For instance, conceptual synergies could be exploited by joining the evolutionary perspective with hitherto essentially static public choice concepts like asymmetric knowledge in principal–agent constellations, or political and bureaucratic rationality. In any case, the ongoing cluster euphoria in policy and practice needs some timely and constructive scholarly critique which in turn offers fresh opportunities for academics to overcome their loss of orientation and to underscore their social relevance.

348

Emerging clusters

NOTES * 1. 2.

3. 4.

5. 6. 7.

Helpful comments from workshop participants and two anonymous referees are gratefully acknowledged. The author assumes full responsibility for all remaining shortcomings. See, for example, Brown and Duguid (1991), Wenger (1999) and Amin and Cohendet (2005). This residual category includes all experts not directly engaged in the concept development, governance or consulting of a cluster policy. However, they are qualified to judge the respective initiative through their education as economic geographers, regional economists, spatial planners and so on, their professional experience and their local context. Some of these experts are assigned to more than one state or category; hence the total of the breakdown given here exceeds 134. Ziesemer (2004) provides a detailed inside report of the dortmund-project, while preliminary appraisals can be found in Küpper (2005), Küpper and Röllinghoff (2005) and Röllinghoff (2008). The original vision of the economic development of Central Franconia included five fields of competence, namely transportation and logistics, ICT, medical technology and healthcare, energy and the environment, as well as new materials. In its 2005 update, automation and manufacturing technology was added as a sixth field, and innovative services as a generic competence (see IHK Nürnberg 2005). For the BioRegio contest, see Cooke (2002) and Dohse (2007), as well as Eickelpasch and Fritsch (2005). A notable exception is Dortmund’s neighbouring city of Bochum which filled key positions in its cluster management unit with former dortmund-project staff. Established five years later in 2005, Bochum2015 follows the neighbour’s example closely. For instance, hannoverimpuls managed to secure its survival until 2012 not on the account of its contribution to cluster development, but primarily by acquiring the task to administrate ERDF funding for the Hanover region.

REFERENCES Amin, A. and P. Cohendet (2005), ‘Geographies of Knowledge Formation in Firms’, Industry and Innovation, 12, 465–86. Andersson, T., S. Schwaag Serger, J. Sörvik and E. Wise Hansson (2004), The Cluster Policies Whitebook, Malmö: International Organisation for Knowledge Economy and Enterprise Development. Bathelt, H. (1998), ‘Regionales Wachstum in vernetzten Strukturen: Konzeptioneller Überblick und kritische Bewertung des Phänomens “Drittes Italien”’, Die Erde, 129, S. 247–71. Bathelt, H., P. Maskell and A. Malmberg (2004), ‘Clusters and Knowledge: Local Buzz, Global Pipelines and the Process of Knowledge Creation’, Progress in Human Geography, 28, 31–56. Boekholt, P. and B. Thuriaux (1999), ‘Public Policies to Facilitate Clusters: Background, Rationale and Policy Practices in International Perspective’, in OECD (ed.), Boosting Innovation: The Cluster Approach, Paris: OECD, pp. 381–412. Boschma, R.A. and M. Sotarauta (2007), ‘Economic Policy from an Evolutionary Perspective: The Case of Finland’, International Journal of Entrepreneurship and Innovation Management, 7, 156–73.

Evolutionary perspective on cluster policies in Germany

349

Brödner, P. (2003), ‘The Internet: Universal Information Infrastructure for the Emerging Knowledge Society’, in E. Helmstädter (ed.), The Economics of Knowledge Sharing: A New Institutional Approach (New Horizons in Institutional and Evolutionary Economics), Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 147–69. Brown, J.S. and P. Duguid (1991), ‘Organizational Learning and Communities of Practice: Toward a Unified View of Working, Learning, and Innovation’, Organization Science, 2, 40–57. Castells, M. and P. Hall (1994), Technopoles of the World: The Making of 21st Century Industrial Complexes, London, New York: Routledge. Cohendet, P. and A. Amin (2006), ‘Epistemic Communities and Communities of Practice in the Knowledge-based Firm’, in C. Antonelli, D. Foray, B.H. Hall and W.E. Steinmueller (eds), New Frontiers in the Economics of Innovation and New Technology: Essays in Honor of Paul A. David, Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 296–322. Cooke, P. (2002), ‘Biotechnology Clusters as Regional, Sectoral Innovation Systems’, International Regional Science Review, 25, 8–37. Dewald, U. (2006), Clusterpolitik als Instrument der Regionalentwicklung am Beispiel des Bergischen Städtedreiecks (SPACES – Spatial Aspects Concerning Economic Structures, 2006–02), Marburg: Faculty of Geography, Philipps-University. Dicken, P. (2004), ‘Geographers and ‘Globalization’: (Yet) Another Missed Boat?’, Transactions of the Institute of British Geographers, 29, 5–26. Diefenthal, T. (2006), ‘BioRegio Regensburg: Biotechnologische Revolution und Gründungsboom an der Donau’, in A. Sedlmeier and J. Vossen (eds), Stadtatlas Regensburg, Regensburg: Pustet, pp. 110–11. Dohse, D. (2007), ‘Cluster-based Technology Policy: The German Experience’, Industry and Innovation, 14, 69–94. Dolowitz, D.P. and D. Marsh (1996): ‘Who Learns What from Whom: A Review of the Policy-Transfer Literature’, Political Studies, 44, 343–57. Dolowitz, D.P. and D. Marsh (2000): ‘Learning from Abroad: The Role of PolicyTransfer in Contemporary Policy-Making’, Governance, 13, 5–24. Duranton, G. (2008), ‘California Dreamin’. The Feeble Case for Cluster Policies’, http://individual.utoronto.ca/gilles/Papers/Cluster.pdf, accessed 9 April 2009. Eickelpasch, A. and M. Fritsch (2005), ‘Contests for Cooperation: A New Approach in German Innovation Policy’, Research Policy, 34, 1269–82. Enright, M.J. (2003), ‘Regional Clusters: What We Know and What We Should Know’, in J. Bröcker, D. Dohse and R. Soltwedel (eds), Innovation Clusters and Interregional Competition (Advances in Spatial Science), Berlin, Heidelberg, New York: Springer, pp. 99–129. Facchini, F. (2006), ‘L’Entrepreneur Politique et son Territoire’, Revue d’Économie Régionale et Urbaine, 2, 263–80. Florida, R.L. (1995), ‘Toward the Learning Region’, Futures, 27, 527–36. Franke, S.F. (2000), (Ir)rationale Politik? Grundzüge und politische Anwendungen der Ökonomischen Theorie der Politik, 2nd edn (Grundlagen der Wirtschaftswissenschaft, 8), Marburg: Metropolis. Fromhold-Eisebith, M. (2007), ‘Which Mode of (CLUSTER) Promotion for Which Aspect of Entrepreneurship? A Differentiating View on Institutional Support of Automotive Clusters’, in J.D. Gatrell and N. Reid (eds), Enterprising Worlds: A Geographic Perspective on Economics, Environments and Ethics (The GeoJournal Library, 86), Dordrecht: Springer, pp. 13–28.

350

Emerging clusters

Fromhold-Eisebith, M. and G. Eisebith (2005), ‘How to Institutionalize Innovative Clusters? Comparing Explicit Top-down and Implicit Bottom-up Approaches’, Research Policy, 34, 1250–68. Glasmeier, A.K. (2006), ‘On the Intersection of Policy and Economic Geography: Selective Engagement, Partial Acceptance, and Missed Opportunities’, in S. Bagchi-Sen and H. Lawton Smith (eds), Economic Geography: Past, Present and Future (Routledge Studies in Economic Geography), London: Routledge, pp. 208–20. Gray, V. (1973), ‘Innovations in the State: A Diffusion Study’, American Political Science Review, 67, 1174–85. Groenewegen, J. and M. van der Steen (2007), ‘The Evolutionary Policy Maker’, Journal of Economic Issues, 41, 351–8. Hall, P.A. and D. Soskice (2001), ‘An Introduction to Varieties of Capitalism’, in P.A. Hall and D. Soskice (eds), Varieties of Capitalism: The Institutional Foundations of Comparative Advantage, Oxford: Oxford University Press, pp. 1–68. Hassink, R. (2005), ‘How to Unlock Regional Economies From Path Dependency? From Learning Region to Learning Cluster’, European Planning Studies, 13, 521–35. Hassink, R. and A. Lagendijk (2001), ‘The Dilemmas of Interregional Institutional Learning’, Environment and Planning C, 19, 65–84. Heidenreich, M. (2005), ‘The Renewal of Regional Capabilities: Experimental Regionalism in Germany’, Research Policy, 34, 739–57. Heinecke, H. (2008), ‘Clusterentwicklung in Wolfsburg: Ein Blick aus naher Distanz’, in M. Kiese and L. Schätzl (eds), Cluster und Regionalentwicklung: Theorie, Beratung und praktische Umsetzung, Dortmund: Rohn, pp. 151–6. Heuser, T., P. Kraljic and M.R. Stuchtey (2001), ‘A Plan for German Job Creation. A Partnership Between Volkswagen and the Regional Government in Wolfsburg, Germany, Shows How to Build an Economic Cluster’, The McKinsey Quarterly, 16–18. Hospers, G.-J. and S. Beugelsdijk (2002), ‘Regional Cluster Policies: Learning by Comparing?’ Kyklos, 55, 381–402. IHK Nürnberg (2005), Entwicklungsleitbild der Wirtschaftsregion Nürnberg, Nuremberg: Nuremberg Chamber of Industry and Commerce for Central Franconia. IT-Speicher (2008), ‘Strategische Partnerschaft IT-Sicherheit’, www.regensburg.it/ kompetenz/5543-106,1,0.html, accessed 21 March 2008. Kern, K. (2000), Die Diffusion von Politikinnovationen: Umweltpolitische Innovationen im Mehrebenensystem der USA (Gesellschaftspolitik und Staatstätigkeit, 17), Opladen: Leske und Budrich. Ketels, C.H.M., G. Lindqvist and Ö. Sölvell (2006), Cluster Initiatives in Developing and Transition Economies, Stockholm: Center for Strategy and Competitiveness. Kiese, M. (2008a), ‘Cluster Approaches to Local Economic Development: Conceptual Remarks and Case Studies from Lower Saxony, Germany’, in U. Blien and G. Maier (eds), The Economics of Regional Clusters: Networks, Technology and Policy, (New Horizons in Regional Science), Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 269–303. Kiese, M. (2008b), ‘Mind the Gap: Regionale Clusterpolitik im

Evolutionary perspective on cluster policies in Germany

351

Spannungsfeld von Wissenschaft, Politik und Praxis aus der Perspektive der Neuen Politischen Ökonomie’, Zeitschrift für Wirtschaftsgeographie, 52, 129–45. Kiese, M. (2008c), ‘Stand und Perspektiven der regionalen Clusterforschung’, in M. Kiese and L. Schätzl (eds), Cluster und Regionalentwicklung: Theorie, Beratung und praktische Umsetzung, Dortmund: Rohn, pp. 9–50. Kiese, M. (2008d), ‘Vom Hannover-Projekt zu hannoverimpuls: Clusterorientierte Wirtschaftsförderung in der Region Hannover’, in M. Kiese and L. Schätzl (eds), Cluster und Regionalentwicklung: Theorie, Beratung und praktische Umsetzung. Dortmund: Rohn, pp. 199–230. Knill, C. (2005), ‘Introduction: Cross-national Policy Convergence: Concepts, Approaches and Explanatory Factors’, Journal of European Public Policy, 12, 764–74. Koschatzky, K. and V. Lo (2007), ‘Promoting Regional Networking and Cluster Formation in East Germany: A Chance for Setting up New Regional Growth Regimes in an Economically Volatile Environment?’, International Journal of Entrepreneurship and Innovation Management, 7, 462–81. Krueger, A.O (1974), ‘The Political Economy of the Rent-seeking Society’, The American Economic Review, 64, 291–303. Küpper, U.I. (2005), ‘Zwischenbilanz des “dortmund-projects” aus der Sicht des Wirtschaftsförderers’, Informationen zur Raumentwicklung, 9/10, 627–36. Küpper, U.I. and S. Röllinghoff (2005), ‘Clustermanagement: Anforderungen an Städte und regionale Netzwerke’, German Journal of Urban Studies, 44, 60–93. Lagendijk, A. (2001), ‘Regional Learning Between Variation and Convergence: The Concept of ‘Mixed Land-Use’ in Regional Spatial Planning in The Netherlands’, Canadian Journal of Regional Science, 24, 81–100. Lompe, K., A. Blöcker, B. Lux and O. Syring (1996), Regionalisierung als Innovationsstrategie: Die VW-Region auf dem Weg von der Automobil- zur Verkehrskompetenzregion, Berlin: Ed. Sigma. Lütz, S. (2007), ‘Policy-Transfer und Policy-Diffusion’, in A. Benz, S. Lütz, U. Schimank and G. Simonis (eds), Handbuch Governance: Theoretische Grundlagen und empirische Anwendungsfelder, Wiesbaden: VS Verl. für Sozialwissenschaften, pp. 132–43. Malecki, E.J. (2006), ‘Technology, Knowledge, and Jobs’, in S. Bagchi-Sen and H. Lawton Smith (eds), Economic Geography: Past, Present and Future (Routledge Studies in Economic Geography), London: Routledge, pp. 244–50. Martin, R.A. (2001), ‘Geography and Public Policy: The Case of the Missing Agenda’, Progress in Human Geography, 25, 189–210. Martin, R.A. (2006), ‘Economic Geography and the New Discourse of Regional Competitiveness’, in S. Bagchi-Sen and H. Lawton Smith (eds), Economic Geography: Past, Present and Future (Routledge Studies in Economic Geography), London: Routledge, pp. 159–72. Martin, R.A. and P. Sunley (2003), ‘Deconstructing Clusters: Chaotic Concept or Policy Panacea?’, Journal of Economic Geography, 3, 5–35. Menzel, M.-P. and D. Fornahl (2005), ‘Unternehmensgründungen und regionale Cluster: Ein Stufenmodell mit quantitativen, qualitativen und systemischen Faktoren’, Zeitschrift für Wirtschaftsgeographie, 49, 131–49. Metcalfe, J.S. (1995), ‘Technology Systems and Technology Policy in an Evolutionary Framework’, Cambridge Journal of Economics, 19, 24–46.

352

Emerging clusters

Moreau, F. (2004), ‘The Role of the State in Evolutionary Economics’, Cambridge Journal of Economics, 28, 847–74. MWME (2006), NRW-Ziel 2-Programm (EFRE) 2007–2013: Operationelles Programm (EFRE) für das Ziel ‘Regionale Wettbewerbsfähigkeit und Beschäftigung’ für Nordrhein-Westfalen, Düsseldorf: Ministry for the Economy, SMEs and Energy of North Rhine-Westphalia. Nauwelaers, C. (2001), ‘Path-Dependency and the Role of Institutions in Cluster Policy Generation’, in A. Mariussen (ed.), Cluster Policies – Cluster Development? A Contribution to the Analysis of the New Learning Economy (Nordregio Report, 2001–2), Stockholm: Nordregio, pp. 93–108. Nelson, R.R. and S.G. Winter (1982), An Evolutionary Theory of Economic Change, Cambridge, MA and London: Belknap Press. Neumann, G. (1996), ‘Regionales Change-Management: Das Nürnberg-Programm – ein exemplarischer Ansatz zur Verknüpfung von Regional-, Wirtschafts- und Arbeitsmarktpolitik’, WSI-Mitteilungen, 49, 754–63. Nonaka, I. and H. Takeuchi (1995), The Knowledge-creating Company: How Japanese Companies Create the Dynamics of Innovation, New York: Oxford University Press. North, D.C. (1990), Institutions, Institutional Change and Economic Performance, Cambridge: Cambridge University Press. OECD (ed.) (1999), Boosting Innovation: The Cluster Approach, Paris: OECD. OECD (ed.) (2001), Innovative Clusters: Drivers of National Innovation Systems, Paris: OECD. Olson, M. (1965), The Logic of Collective Action: Public Goods and the Theory of Groups, Cambridge, MA: Harvard University Press. Porter, M.E. (1998), ‘Clusters and Competition: New Agendas for Companies, Governments and Institutions’, in M.E. Porter, On Competition (The Harvard Business Review Book Series), Boston: The Harvard Business School Publishing, pp. 197–287. Pratt, J.W. and R.J. Zeckhauser (1991), ‘Principals and Agents: An Overview’, in J.W. Pratt and R.J. Zeckhauser (eds), Principals and Agents (Research Colloquium Harvard Business School), Boston, MA: Harvard Business School Press, pp. 1–35. Prätorius, G. (2004), ‘“Projekt Region Braunschweig”: Kooperatives Wachstumskonzept für die Region’, Neues Archiv für Niedersachsen, 2, 55–9. Rehfeld, D. (2005), ‘Grenzen wissenschaftlicher Politikberatung – Überlegungen zur zeitlichen Dimension am Beispiel der Strukturpolitik’, in U. Jens and H. Romahn (eds), Glanz und Elend der Politikberatung, Marburg: Metropolis, pp. 129–48. Rehfeld, D. (2006), ‘Kompetenzfeldwirtschaft im Ruhrgebiet’, Zeitschrift für Wirtschaftsgeographie, 50, 245–57. Rogers, E.M. (2003), Diffusion of Innovations, 5th edn, New York, NY: Free Press. Röllinghoff, S. (2008), ‘Clusterpolitik im Strukturwandel: Das dortmund-project’, in M. Kiese and L. Schätzl (eds), Cluster und Regionalentwicklung: Theorie, Beratung und praktische Umsetzung, Dortmund: Rohn, pp. 157–82. Rose, R. (1991), ‘What is Lesson-Drawing?’, Journal of Public Policy, 11, 3–30. Rose, R. (1993), Lesson-drawing in Public Policy: A Guide to Learning Across Time and Space, Chatham, NJ: Chatham House.

Evolutionary perspective on cluster policies in Germany

353

Ross, S.A. (1973), ‘The Economic Theory of Agency: The Principal’s Problem’, American Economic Review, 63, 134–9. Sautter, B. (2004), ‘Regionale Cluster: Konzept, Analyse und Strategie zur Wirtschaftsförderung’, Standort, 28, 66–72. Siebert, H. (2006), ‘Locational Competition: A Neglected Paradigm in the International Division of Labour’, The World Economy, 29, 137–59. Sölvell, Ö., G. Lindqvist and C.H.M. Ketels (2003), The Cluster Initiative Greenbook, Gothenburg: Ivory Tower AB. Stadt Regensburg (2003), Strategische Partnerschaft zwischen Unternehmen, Hochschulen und der Stadt Regensburg im Bereich der Sensorik: Kooperationsvereinbarung, www.sensorik-regensburg.de/downloads/Sensorik_ Vereinbarung.pdf, accessed 15 January 2008. Steiner, M. and C. Hartmann (1999), ‘Interfirm Co-operation and Learning within SME Networks: Two Case Sudies from the Styrian Automotive Cluster’, in M.M. Fischer, L. Suarez-Villa and M. Steiner (eds), Innovation, Networks and Localities (Advances in Spatial Science), Berlin, Heidelberg, New York: Springer, pp. 85–106. Sternberg, R., M. Kiese and L. Schätzl (2004), ‘Clusteransätze in der regionalen Wirtschaftsförderung: Theoretische Überlegungen und empirische Beispiele aus Wolfsburg und Hannover’, Zeitschrift für Wirtschaftsgeographie, 48, 159–76. StMWIVT (2006), Allianz Bayern Innovativ: Eckpunkte bayerischer Clusterpolitik, Munich: Bavarian Ministry of the Economy, Infrastructure, Transportation and Technology. Stone, D. (2004), ‘Transfer Agents and Global Networks in the “Transnationalization” of Policy’, Journal of European Economic Policy, 11, 545–66. Tews, K. (2002), Der Diffusionsansatz für die vergleichende Policy-Analyse: Wurzeln und Potenziale eines Konzepts – Eine Literaturstudie (FFU Report, 02–2002), Berlin: Free University of Berlin. Tödtling, F. and M. Trippl (2004), ‘Like Phoenix From the Ashes? The Renewal of Clusters in Old Industrial Areas’, Urban Studies, 41, 1175–95. van der Linde, C.M. (1992), Deutsche Wettbewerbsvorteile, Düsseldorf, Wien and New York: Econ. Walker, J.L. (1969), ‘The Diffusion of Innovations Among American States’, American Political Science Review, 63, 880–89. Wenger, E. (1999), Communities of Practice: Learning, Meaning and Identity (Learning in Doing: Social, Cognitive and Computational Perspectives), Cambridge: Cambridge University Press. Witt, U. (2003), ‘Economic Policy Making in Evolutionary Perspective’, Journal of Evolutionary Economics, 13, 77–94. Ziesemer, A. (2004), Strategische Stadtentwicklungsplanung im Ruhrgebiet: Eine Analyse am Beispiel der Städte Duisburg und Dortmund (Duisburger Geographische Arbeiten, 25), Dortmund: Dortmunder Vertrieb für Bau- und Planungsliteratur.

Index academic and non-academic occupations 120–21, 130 ‘accidents’ 3, 10, 76 Achurra, M. 175, 177–8, 180 adjustment policy (Prozesspolitik) 44–5 agglomeration 102 disadvantages 306 economies 2–3, 5, 193, 195, 239–40, 261, 299 critical mass of companies 76 spinoff dynamics 191 effects 260 cluster-specific advantages 49 externalities 1 military 304 positive 318 specialised 326 Akron, tyre cluster 5, 240 Alborg, NorCom wireless communications 30, 38 Alborg University 26 allies, broadcasting models in occupied zones 58–9, 65, 106 Amsterdam banking cluster 192, 207 increase since (1850) 202, 206, see also Dutch Banking Central Bank location 198, 207 demise of diamond industry 91 hazard rates of banks 205–6 survival curves of spinoffs 206–7 takeovers 209 Twentsche Bank 199–200 WPO closed 208 Amsterdamsche Bank 199–200 Amsterdamsche Bank voor Belgie 80 Andreessen, Marc 219, 231–2 Angel, D.P. 103, 104, 127, 303 Anglo-American economic geographers 325

Antwerp 74–6, 83, 92 datasets of registers van aankomst en vertrek 78 diamond cluster 6–7, 80, 92 re-emergence of 74, 78 diamond district until 1940 79–81 evolution of competing centres 83–4 migrants to diamond district (1948–65) 88 some Jews returned (1940s) 81, 92 World War I interruption 80 World War II, still existing growth centres 75 see also diamond bourses Antwerp City Archive (Stadsarchief Antwerpen) 78 Antwerp diamantaires 84 attracted diamond dealers 88 dispersal of a major problem 85 fled from Nazis 89–90 Antwerp World Diamond Centre 74 APPENDIX 5A 132–9 APPENDIX 11A, case studies 284–90 APPENDIX 12A 300, 323 APSTC 174, 177–9 Arhus, fibreglass wind turbine blades 27 Arhus University 26 Arstiderne organic Food Network 38 Arthur, W.B. 2–3, 75–6, 193–5, 239, 266–7 Asia 27, 180 Association of Chilean Salmon Producers 181 Association of Salmon and Trout Producers of Chile see APSTC Association of University Technology Managers (AUTM) 224 Audretsch, D.B. 4, 143, 191, 217, 239, 242

355

356

Emerging clusters

automobile industry, cycle and coach-making 194 automobile and tyre industries 192, 209 AVM clusters in Germany 100, 105–110, 133–4 Cologne and Rhein-Erft 108 employees and directions of labour flow 125–6 employment stock 116 firms and employees (1980 TO 2007) 106–7 industrial origins of job movers 122, 125 intra-industry changes 122 job changes at local level 116 job rankings in cluster regions 111–12, 135 large firms in 138–9 location Kreis 108–9 make-up of employment stock 129 shares of job-changers and non-job-changers 112–15 sub-sectors 106, 132 top occupations, 110–111, see also APPENDIX 5A Avnimelech, G. 140–41, 143–5, 154, 160 Baden-Baden academic occupations 122 AVM 108 French occupying army 59 job-changers 112, 114 non-job-changers 112, 114 Baden-Württemberg 23, 244 Basic Constitutional Law 51 Bathelt, H. competencies in Jena and Berlin 258 emerging clusters 1, 3, 44–5, 48–9, 76, 91, 170 regional clusters 281, 326, 337 Bavaria 336, 337–8 automotive industry, BAIKA 340–41 industrialization 309–310 Offensive Zukunft Bayern and High-Tech-Offensive (HTO) 341 privatisation thrust 339 Tele 5 (later DSF) 63–4

Bavaria and Lower Saxony 335–6 Bavarian Broadcasting (Bayerischer Rundfunk BR) 58 Bayh Dole Act (1980) 215, 225 Beck, H. 52, 57, 61 Be’er Sheva, railway connection 18 Belgian and Antwerp authorities, against anti-Semitism 86 Belgian exile government in London 90 Ben Gurion University, Negev in Israel Cleantech Ventures 18 Berlin 53, 55–7 biochip industry 8, 240, 242, 244, 246–7, 252–5, 261 first phase 254–6 second phase 256–7 third phase 257–8 BioHytec network 243, 256 medicine 258 radio 58–9 Berlin Wall, after fall Jenoptik 100 spin-offs 248 Berlin-Buch 256, 258 Beugelsdijk, S. 45, 295, 326, 331, 339 Beurs voor Diamanthandel 79, 82 Beutenberg Campus 248, 249–51, 258 Bi-national Industrial R&D see BIRD biochip industry, development 248–9 ‘biochip’ or ‘microarray’ 244 biochip technology, definition 242 Biodot 268, 271, 272, 276 BioHytec 243–4, 256–7 BioPark incubator 338 BioRegio 243–4, 249–50, 338, 348 biotechnology clusters 105, 219–20 biotechnology industry 4, 19, 141 Birch index, ‘balanced’ index 107–8 BIRD program, US and Israel 148–50 Board of Science and Technology 305 Boggs, J.S. 3, 76, 91, 258 Bone Care International 230–31 born global companies 161 Boschma, R.A. automotive industries 39 clusters 1, 47, 239, 266, 328 Dutch banking 191–6, 208 economic geography 298 incubators 49–50 path creation 299

Index Third Italy 20 WLO and 76 Boston 159, 220 bounded rationality, evolutionary policy-maker 327 Box 6.1 New ITP programs in Israel During (1990s) 147 Brachfeld Archive, Jews in diamond trade 79 Brandenburg 243, 252 Braunerhjelm, P. 2, 11, 75, 166, 169, 214, 317 IJP 265–6 Brazil 38, 81, 85 Bremen Broadcasting (Radio Bremen RB) 58–9 Brenner, T. 4, 49, 143, 191, 194, 242 Breschi, S. 105, 140, 166–8, 181 Bresnahan, T. 2, 11 case studies 301, 316–17 cluster development 165, 168, 239 high-tech clusters 295–6 knowledge flows 170 networks and VC 217 regional agglomeration 140 British automobile industry 196 broadcasters under public law market pioneers 65 protected monopolies 59–60 vertical-integrated systems of TV 60, 65 broadcasting, consumers’ freedom 51 Broadcasting Berlin-Brandenburg (Rundfunk Berlin-Brandenburg RBB) 55 broadcasting law of North-Rhine Westphalia, RTL Plus 63 Brunschweig region, transportation 343 Buch 252, 257 cable networks 61 calcitriol, used for osteoporosis 230 California 6, 18–19, 37 agro-food, a key industry 36 Carlsbad, alloy golf cluster 38 cleantech clusters, juxtaposition of 35 Jacobian clusters 25 semiconductor industry 40 windmill blade inferiority 27, 37

357

Cambridge Acorn Computers 267 Auto ID Centre 280, 282 centre for R&D 279, 281 IJP cluster 9, 265, 272, 280 M&A 276–7 international marketing of IJP 276 job mobility 104 job turnover 127 Cambridge Consultants Ltd see CCL Cambridge Display Technology 280–81 ‘Cambridge phenomenon’ 301, 307 Cambridge region (UK) 296 high-tech centre 265 Cambridge Science Park 306–7 Cambridge University Engineering Department 268, 279 Cambridgeshire 312–13, 317 case study 306–7 IJP companies 272 SMEs 127 Canada 171, 174 Carbon Capture and Storage (CCS) 35 Cardiff microprocessor firm IQE 35 Carpathian Jews 87 Castells, M. 104, 305–6, 308, 310, 331 Catalonia, SEAT and SME ‘district’type innovation relations 23 CCL 268, 269–70, 272–3, 280 cell phones, Xennia and CIT 280 Central Fraconia 337–8, 342 Champaign-Urbana, Illinois 220–21, 234 chance 76–7, 87, 91, 92 Charité, the 253, 258 CheckPoint, global ICT firm 154 Chile 166 ‘Code of Standards for Salmon’ 178 connecting to advanced country markets 180–81 copper exports 171 CORFO 174–5, 179, 181, 183 diversification of products 181 farmed salmon business (1980s) 7, 171–2 government help 174–6, 183 fish mortality and fish diseases 179 foreign firms as exporters of salmon 173 Fundación 174–5, 177–8, 181, 183

358

Emerging clusters

IFOP 174, 183 lessons from 181–2 ocean ranching 174–5 quality seal certification 177–8 quality standards 176–9, 182 salmon aquaculture cluster 167 salmon aquaculture in 171–4 SERNAPESCA 174, 175, 178 Sociedad de Pesqueria Lanquihue 175 supermarket chains 180 Tenth Region 172–4, 176, 183 Chile’s Economic Development Corporation (Corporacion de Fomento) see CORFO Chile’s National Fisheries Service (Servico Nacional de Pesca) SERNAP 174 China 74, 180 Cité Scientifique de Paris Sud see Ile-de-France cleantech sectors 18, 22, 35–7, 40, 141 Clement, Wolfgang (NRW minister) 335, 340 cluster development and entrepreneurship 143 cluster effect, MNCs in ICT areas 154 cluster emergence connected to spin-offs 260 institutional mechanism 182 two insights 239 cluster emergence process 141, 144 cluster firms, core activities 102–3 cluster formation 50, 53, 60, 64 co-evolutionary process 220 cluster growth dynamics 168 cluster initiatives, knowledge-intensive industries 295 cluster mutation, discovery of 18–19 cluster policies analysis 334 institutional responses 325 sub-state level 341–3 cluster policies (Clusterpolitik) 44 Cluster Policies Whitebook 341 cluster policy assigned to structural policy 44 case studies 334–5 economic policy 44, 67 industrial policy (Industriepolitik) 44

cluster region, growth stage 102, 160 cluster ‘species’ multiplication 35, 37 cluster structure, patterns of job mobility 102 cluster-life cycles 99–100, 108, 328 cluster-related occupations 110 cluster-specific human capital base 101–2 cluster-specific jobs 110 cluster-specific labour markets 104 clustering processes 55 seedbeds for 65 clusters definition 1, 241 dynamic approaches rare 266 early phases 168 emergence of 2, 5 accidents, path dependency and strategic action 2–3 human capital base 99 path-dependent phenomenon 75–7 political support of 44 co-evolutionary links, types of 146, 160 co-ordinated economies, Germany and Japan 315 co-ordination problem, five spheres 314–15 codification 332 COFDI 84–7, 89–90, 93 Cognac, relocation from Antwerp to 81, 90 collective behaviour 144 collective knowledge flows 166, 171, 181 collective learning 165–7, 169, 182, 266, 330–31 Cologne 46, 53, 55–7, 59, 64, 66, 108, 342 municipal savings bank 63 non-job changers 112, 114 competence base 266 competencies 248, 258, 260–61 patents and 273 qualitative interviews and 247 spin-offs and 259 competition-oriented cluster concepts 102 competitive advantages 48, 76 competitiveness 170

Index Competitiveness Institute 331 Comptoir Diamantaire Anversois SA 80, 92 computer aided design (CAD) drawings 275 Comverse (precursor Efrat) 150 Concentration Index (CI) 107–8 concept of ‘filière’ 309 concept of path creation 77 Conductive Inkjet Technology 280 Congo, raw diamond extraction 85 continuous IJP 269, 278 Cooke, P. 18, 21–3, 26, 35, 37, 75 coordination issues 144 core competencies 257, 328 core regions, radical shift of 298 core–periphery model 3 COROP Groot-Amsterdam 201–2 Correspondence Office for the Diamond Industry see COFDI Corus Colours, Solar Paint 35 Council for mutual Economic Assistance (COMECON) 248 Cox regressions 203, 205 critical mass 76, 141, 144–5, 151, 161, 193, 328 ICT start-ups 151, 161 ‘cumulative causation’ 21–2 cutthroat competition 52 Czechoslovakia 89 Dahl, M.S. 104, 194, 217 Danaher IJP (USA) 274, 277, 279 Danish Technological Institutes (DTI) 26, 32 Danish Wind Energy Association database 26 De Beers Consolidated Mines Ltd 84–5 de noveo high technology firms 221, 236 decentralization 305–8, 311, 317 decisions in regulation policy, spatial effect 50 decontextualisation 332, 334 deindustrialization 305, 307 DeLuca, Professor Hector, Vitamin D 230–32 Denmark 18 Vestas Wind Systems of Randers 27

359

Départment Alpes Maritimes 308 Detroit, automobile cluster 5, 240 Deutsche Stunde Gesellschaft für drahtlose Belehrung und Unterhaltung mbaH 57 developing country firms 166–7, 170–71 Devisenschutzkommando, raided diamond bourses 82 Dewald, U. 44–5, 337 Diamantclub 79, 82 Diamantkring 79 diamond bourses close on Jewish holidays 81 raided by Devisenschutzkommando and Sipo-SD 82, 93 Diamond Corporation Ltd (Dicorp) 84 diamond polishers 83, 93 diamond sector, monopolistic structures 91 diamond workers before the war and in 1945 83 Flemish origin 81, 89 returning to Antwerp 86–7 differentiation, between cluster and agglomeration 18 diversification, firms in Jena 251 diversifiers 255–6, 258 Domino 268, 270–71, 272, 274–80 purchased laser companies 278 Dorenkamp, Ansgar 47–50, 55, 57, 59, 64, 107 Dortmund area 246, 337, 344 dot matrix printing 269 Dresden 59 drop-on-demand IJP 269, 278 Dulas, green engineering firm mid-Wales 35 Dutch banking 8 906 banks (1850–1993) 198, 209 ABN-AMRO Bank 200 cohorts 205–6 entrants and exits 198–9 evolution of banks (1850–1993) 200, 210 four large banks in Amsterdam 201, 210 Geat Depression and 198–9, 209 innovations 199

360

Emerging clusters

mergers and acquisitions 199, 203, 209 oligopoly (ABN-AMRO, ING and Rabobank) 201 rebirth (1850) 197 related activities 204 spinoff entrants (1850–1993) 204 Dutch Central Bank, network of offices 198 Dutch colonies, new capital needed 197 Dutch diamond district 88–9 dynamic growth cluster characteristics 75 regions attract mobile production 49 Eastern Europe, anti-Semitism 91 Ecole Nationale Supérieure des Mines de Paris 308 ecological processes 266, 273–8 economic geographers, core competency 325 economic policy framework 67 economies of scale 48, 155 and scope 144 ecosystem 216 effective organizational routines 48–50, 192–3, 261 electronic engineering and computer software (EE&CS) 224–5 ELISA tests 256 elite universities, clusters in Illinois and Wisconsin 8 Elscint Law 149 Embraer (Brazil) 165 embryonic clusters, environment and 91 ‘emergence’, definition 2 emergence of clusters, factors of 166 emerging clusters 241 no specialised labour pool 122 emerging markets, one pioneer 47 empirical analysis, cluster emergence and 316–17 empirical evidence, theoretical assumptions 315–16 empirical research, cluster policies 327 endogenous cluster dynamics 4, 7, 10–11

endogenous processes, clusters and 49 Engineering and Physical Sciences Research Council (EPSRC) 279, 281 Enterprise Integration Technologies (EIT) 231 ‘entrepreneurial’ (ERIS) 23 entrepreneurship, successful clusters 143 environmental contamination 179 Europe bolt for biofuels 33 cluster concept 44 demand for Danish Wind Systems 27 ‘institutional’ (IRIS) 23 IPOS 156 salmon 170, 176–7 Second German Television (Zweites Deutsches Fernsehen ZDF) 55, 59, 62, 107 European Patent Office 268 European Regional Development Fund (ERDF) 337 NRW and 342 European Union 21, 51, 107 evolutionary concept of variety 24 evolutionary economics, state and 327 evolutionary policy-makers bounded rationality 327 learning actors 344 evolutionary process 153 ex-post events 10 external economies 168 externalities 1, 105, 144 extra-regional supply of labour large firms 127 (Thesis 5) 127–8 West German labour market regions 109–110 face-to-face contacts 170, 178, 298 Federal Agency of Employment (Nuremberg) 106, 111 Federal Constitutional Court 51 private broadcasting 60 Fein, A.J. 192, 197, 209 Feldman, M. 2–5, 11, 75, 214, 220, 260, 267 Feldman, M.P. biochip industries 239–42

Index clusters 143–4, 166, 168–9, 191, 265–6, 317 labour pool 102 regional agglomeration 140 university clusters 217, 220, 225, 233 fields of competence, defined as clusters 338, 348 firm founding, quantitative and qualitative effects 143 firm-specific routines 192–3 first-mover advantages 47, 50–51, 53, 65 Fisheries Development Institute (instituto de Formento Pesquero) see IFOP Flensburg 112, 115 flexible specialization 297 Florida, R. 142, 148, 216, 332 fluid sub-phase (1993–95) 153, 160 foreign firms, source of foreign demand 181 foreign VC companies, Israeli start-up and 157 Forminier and De Beers contract 91 Forminiere company (Societe Internationale Forestiere et Miniere du Congo) 85 Fornahl, D. 3 agglomeration 49, 140 clusters 91, 143–4, 191, 239, 241, 328 human capital 100 IJP 265 labour pool 122 networks 104 Fortunia 79, 82 France 9, 87, 93 Archives Nationales 78 Credit Mobilier 197 innovative regions 23 ‘Mediterranean ’ type of capitalism 315 Francis, J.L. 52, 168, 169, 191, 217, 220, 233 Frankfurt/Mainz/Wiesbaden 55 Fraunhofer IBMT 244, 253, 256–7 Fraunhofer Institute for Biomedical Engineering see IBMT French sector, radio Koblenz 58 Frenken, K. 1, 24, 49–50, 191, 195, 241, 299

361

Frynas 47, 48, 50, 65 Fujifilm 279 Fukuoka silicon sea-belt project 311 Gambardella, A. 140, 295–6, 301, 316–17 Garnsey, E. 265–6, 269, 308 GDR 256 Central Institute (ZIMET) 248, 253 geographical mobility 104 highly-qualified human capital 120–22 geography, concept of path creation 77 German Cancer Research Centre (DKFZ) 243–4 German Democratic Republic, Carl Zeiss Jena 248 German Federal Research Ministry 243 German Human Genome Project, expiration of 251 Germany 9, 39, 45, 316 airship cluster in Friedrichshaften 91 biochip industry 243 spatial pattern 244–5 broadcasting corporations under public law (öffentlichrechtliche Rundfunkanstalt) 52 broadcasting monopoly for military use 57 Bundesarchiv 78 case studies 334–9, 345 clusters in broadcasting industry 6, 55 clusters and spatial systems 46–53 economic development policy 324 economic policy (Wirtschaftspolitik) 44 evolution of spatial pattern of TV 57–64 spatial effects on cluster formation 66 federal states (Bundesländer) focused on TV industry 46 freelancers in 106 generic non-local knowledge 346 global cluster hype 324 infrastructure for telecasts 65–6

362

Emerging clusters

licensing Home Order Television (1995) 64 media policy after World War II 51 private broadcasters 53–5, 61 licences and 61 location and economic policy 60, 63–5 private broadcasting (1984) 52–3, 106 social insurance system 106 sound and television broadcasting 46, 51 spatial pattern of TV industry 53–6, 66 supply Danish wind energy input market 27 surrender May 1945 58 VC industry 7, 141, 145, 148, 151–3, 157 Germany’s audio-visual media industry see AVM Germinal Holdings, AberDart strain of rye-grass 33 Geschiedenis van de Algemene Banken in Nederland 1860–1914 197 Geschiedenis van de Nederlandsche Bank 197 Giuliani, E. 165, 170, 181, 298 global ICT markets 154 global knowledge 170–71 globalization 79, 142, 152, 325 Great Depression, impact on diamond industry 80 Greater Boston region 296, 301, 304, 312–13 ‘green clusters’ 23–4, 27, 34–5, 37 ‘green’ innovation 19–20, 26, 37–8, 40 greenhouse gas (GHG) 37 Grenoble region, the ‘Cité Scientifique de Paris-Sud’ see Ile-de-France Group de recherche européen sur les milieux innovateurs (GREMI) 298 Grundig 337 Hague region 202 Hall, P. 296, 301, 305–6, 308, 310, 331 Halle 59 Hamburg 53, 55–9, 108 Hanover 59

Hans-Knöll-Institute for Research of Natural Products (HKI) 249 Hassink, R. 77, 332–4, 346 Hazard Analysis of Critical Control Points (HACCP) 170, 178 Heinrich, J. 51–2, 60–61, 66 Henry, N. 3, 5, 10, 104, 127, 142, 297 Hessian Broadcasting (Hessischer Rundfunk HR) 58, 62 high technology manufacturing (HTM) 21–2 high-qualified human capital, extra-regional labour pools 129 high-tech clusters 296, 301–2, 307, 318 highly-skilled employees, human capital base (Thesis 2) 120 highly-skilled workers, knowledge-and technology-based industries 101 Historic Firm Panel 112 Historical Employment Database 112, 120 Hitachi (Japan) 274 Hitler, NSDAP used for propaganda 58 Hoffmann-Riem, W. 46, 60–61, 77 Hollywood, film cluster 37 Hospers, Gert-Jan 44, 45, 295, 326, 331, 339 hub-and-spoke district 124, 129 human agency, clusters and 76 human biotherapeutics 220 human capital 101–2 human capital theory 120 IBMT 243–4, 256–7 ICT start-ups 151, 160 IFOP 174–5, 183 IJP companies outsource parts 274 different techniques 269 wide applications 278–80 IJP research centre (University of Cambridge) 278–9 IJP technologies, applied to packaging 275–6 Ile-de-France 296, 298, 301–2, 307–9, 312–13 Illinos 228 Imaje (France) 271–2

Index immigrant Jews, bourse members 80–81 implicit knowledge, local human capital 103 in vivo analysis 261 in-house job careers 102 in-house training, costly for small firms 127 Inbal 147, 152, 154 Inca 268, 270–71, 274, 279 incentives, relocation of new firms 49 incubators 49, 53, 259 CCL 280 failure to become 60 left in Jena 252 networks 259 regions 216 spin-offs and 240 India 85 Tata Steel 35 Wipro and Infosys 165 industrial agglomerations 105, 107 industrial district, definition 142 industrial districts, emblematic examples 124 industrial dynamics 192 industrialisation 337 industries created regional resources 101 establish remote locations 76 industry life cycle 108 industry-specific clusters, specialization 216 Ink Jet Academy 273 ink jet printing see IJP InnoProfile 243 InnoRegio 243 innovation finance (VC) 141 innovation platform 20–21 innovation processes 104, 142 innovation and technology policy see ITP innovative start-ups 143 Institute of Employment Research 105 Institute for Grassland & Environmental Research (IGER) 32–4 Institute for Microbiology and Experiments Therapy (IMET) 248

363

Institute of Molecular Biotechnology (IMB) 249 Institute of Photonic Technology 249 Institute of Physical High Technologies (IPHT) 249, 251 institutionalisation, variations of 335 Integrated Protein Chips for Point of Care Diagnostics (iPOC) 243 Intel Capital 157 inter-firm job mobility 104 inter-firm networks 181 inter-organizational learning 193 Internationaal Instituut vor Sociale Geschiedenis 197 International Public Offerings see IPOs Internet sources 221, 235–6 interregional institutional learning 332–4 interregional labour inflows, human capital base (Thesis 1) 116 interregional labour market, large firms and 129 interregional mobility cluster-specific human capital 102 definition 110 interregional policy learning 343 interregional policy transfer, institutional learning 325 interregional transfer, sporadic 344 INTESAL 179 intra-and interregional rates of labour inflow 116–19 intra-cluster diffusion of knowledge 104 intra-regional mobility 102 definition 109 knowledge transfer 99, 106 IPOs in NASDAQ 154, 156, 160 ISO 9000 170 Israel 18, 21, 83, 87, 93 cluster emergence 144–5 co-evolutionary processes 149–50, 151–2, 154–5 co-evolutionary summary 150–51, 152–3 companies, IPQ in NASDAQ 157 defense industries 151, 153, 158 Dimona nuclear facility Science Park 18 emergence of high tech cluster

364

Emerging clusters

phase III co-evolutionary processes 154–5, 160 start-up intensive ICT cluster 158–9 VC industry 141, 148, 155–7, 159–60 Government’s ITP (1969) 146–7 Grants to Business Sector R&D 147, 161 high-tech clusters 153–4, 158 ICT cluster 7, 145–6, 148 development 141, 143 ICT cluster, co-evolutionary links 145–6 ICT multinational companies 160 immigration from Soviet Union 151, 154, 157, 162 IPOs and M&As 157 ITP programs 141, 146–7 program portfolio 146–7 R&D 141, 143, 146–53, 159–61 Ramat Gan 84 start-ups, foreign VC companies 157 Yozma program 147, 152, 154–5, 156–7, 161 Italy 39 ITP 141, 146–7, 150–51, 160 Jack, A 82, 85, 89 Jacobian cluster mutation 19, 32 focus on ‘the city’ 21 Jacobian cluster path dependencies 31 Jacobian clusters, assets for 39 Jacobian dimension product of its emergence 24–5 railroadization and 18, 30 Jacobs, J. 4, 19, 23, 38, 208, 241 Japan 9, 85 Chile and 177, 180 Development Bank of 316 Japan International Cooperation Agency (JICA) 174 Jena 8, 34 attenuated second generation growth 261 biochip industry 240, 242–4, 246–7 development of 248–9, 261 first phase 250

second phase 250–51 third phase 251–2, 251–3 IPHT 249 optics and precision engineering 258 variety generation 255 Jena Biochip Initiative 252 Jewish community, diamond sector and 80, 92 Jewish diamantaires, Antwerp trade and 86 Jewish Museum of Deportation and Resistance 79 job hopping, Motorsport Valley 127 job mobility 100 clusters and 101–2 Germany 105 labour pooling and 116 job movers academics and 122, 124 definition 106 mobility patterns 129 region-specific origins 127–8 Johns Hopkins University 225 Jutland (Denmark) 350 community schools 17, 32 clusters 37–8 food path dependence 35 social capital SME-based entrepreneurship 32 Wind Industry Association 26–7 wind turbine technology 38 Kaiser-Wilhelm-Institut für Gehirnforschung (institute for brain research) 252 Kaiserslautern, public broadcasting station 116 Karnøe, P. 37, 77, 297 Kenney, M. 1, 142, 148, 216, 224, 302 key knowledge 101, 104–5 Kiese, Matthias 44–5, 316, 327, 335, 338–9, 344 Kinsbergen, A. 79–81, 84, 89 Klepper, S. 4–5, 8–10, 39, 49, 76, 143–4, 261 ‘anchor firms’ 267 cluster emergence 239, 241, 260 Detroit automobile cluster 240 Dutch banking 191–2, 194–5, 203–6 hypothesis 247, 258–9, 262

Index knowledge 194 acquisition 166 asymmetric 347 capabilities 21 capital markets and 153 communities 347 channels for policy transfer 342 related to learning 169 economy 21 flows 166, 170, 177 standards and 182 generation, localization of 4 radical innovations 259 society, Germany 46 transfer 38, 99, 101, 273, 276, 347 knowledge spillovers 1, 24, 26, 37–9, 105, 191 knowledge-intensive industries 142–3 localised 103 knowledge-intensive activities, face-to-face contacts 142 business services (KIBS) 21–2 clusters 297–300, 317 service sector 208 kompetenzhoch 337, 348 Korean War (1950–53) 88, 91 Kreis, M. 116, 117, 129 local district 107–9 Krugman, P. 2, 103, 239 Kymmell, J. 197–9, 201, 204 Kyushu island (Japan) 296, 302, 310–11, 312–13, 314, 317 labour flows 127–8 labour inflows 116–19 labour market pool 281 labour market region, several districts 109 labour mobility 105, 108, 129, 298 labour poaching 103, 105, 116 labour pooling 103 labour-mobility 99, 101 Lagendijk 333–4, 346 Lambooy, J. 20, 47, 49, 193 laser printing 269 late-mover advantages 48 Laureys, Eric 74, 78–82, 84–6, 88 learning by doing 9, 247, 333

365

learning processes, developing countries 170 Lehman Brothers 152 Leibniz Institute for Age Research 249 Leipzig 3, 59, 106, 347 liberal market, US and UK economy 315 licences 52–3, 57, 60–65 Lieberman, M.B. 47, 48, 191 Linton, Bill, Promega of Madison 219 Linx 268, 270–71, 274, 276, 279 literature review, cluster emergence 142–3, 144–5 lobbying and rent seeking 328 local competition 1 local environment 193 local industrial structure, labour mobility 129 local labor force 4 local labor market pools 7, 99–100 local labour market 103, 120 local mobility definition 108 strongest in clusters 129 location policies (Standortpolitik) 43 London-based de Beers 84 Longhi, C. 4, 127, 241, 308 Los Angeles, Jacobian clusters 26 Lower Saxony 336, 338–9 Lunar Corporation 230 Lütz, S. 328–31, 344 McCann, P. 1, 103, 191 McKinsey & Co 241, 337, 338, 339, 343–4, 348 macropolitical economic stimulus packages 66 Madison, biotechnology cluster 219 Madison, Wisconsin 221, 234 Maggi, C. 171, 173, 176–9 magnet organizations 267 Magnet Program 147, 154 Mainz 53, 55, 59, 61–2, 66, 107–8 non-job changers 112–13 Mainz/Wiesbaden/Frankfurt 55–6 Malecki, E. 23, 296, 305, 325 Malerba, F. 140, 166–8, 181 Malmberg, A. 1, 239, 260, 297 market entry barriers 51 monopoly 64–5

366

Emerging clusters

market exit 65 market failure blocked start-ups 151 media contents are public goods 52 market patterns, vertical splitting 50 market pioneers, first-mover advantages 47–8, 65 Markusen, A. 80, 124, 193, 296, 303–4 Marshall, A. 1, 80, 103, 116, 142, 193, 337 Marshall-Arrow-Romer (MAR) spillovers 103 thinking 38 Marshallian agglomeration economies 194 Marshallian district 124 Marshall’s labour pooling thesis 103 Martin, R. 1, 3, 5, 75–8, 167, 194, 240, 293 ‘Cambridge phenomenon’ 307 clusters 1, 167, 241, 297–8, 326, 330 economic geography 194, 325 path dependence 3, 5, 75–8, 298–9 Maskell, P. 1, 140, 142, 144, 217, 239, 260, 297 mass-media, social and cultural importance 52 Massachusetts 228 Max-Delbrück-Centre for Molecular Medicine (MDC) 252 Max-Planck-Institute for Molecular Genetics see MPIMG MDC, within BioHytec 253 media goods, economic and cultural goods 52 Menzel, M.-P. chance and 77 cluster emergence 91, 122, 140, 143–4, 239–40 cluster formation 3 cluster life-cycle 100, 328 evolution of clusters 191 firm foundation 241, 247, 249 geographical proximity 260 IJP 265 incubator networks 259 mergers and acquisitions (M&A) 157 Metcalfe, J.S. 153, 325, 327 Mexico 85 micro-arrays, minor focus 258

Milford Haven refineries, SugarGrass biorefineries 33–4 Military Educational Complex 304 mindful deviation, two kinds 78 ‘mission-oriented’ technology 309 MNCs 154, 166, 267 advanced country 170 clustering and 166 source of ideas 181 mobile telephony 32 mobility, local employees and 100 mobility industry 338 mobility patterns, region-specific characteristics 130 mobility of talent 39 Modern Archief 79 monopolistic abuse 52 companies, under public law 64 regional leadership 53 structures 91 diamond sector 91 Montero, C. 171, 173, 175–6, 178 Mosaic, Internet browser 230–31 Mosaic Communications Corporation (later Netscape) 232 Mossig, Ivo 46–50, 55, 64, 107, 122 MPIMG 243–4, 253, 255–7, 258 Munich AVM 46, 53, 55–8, 64, 108–9 biochip industry 243–4, 246 case study 296, 301, 309–310, 312–13 ‘mutation agents’ identifiable 20 Myrdal, G. 21, 193, 195 Myrdal-Hirschman theses, ‘cumulative causation’ 21–2, 40 NACE categories, biodiesel or bioethanol 24 nanotechnology 19, 35, 216 NASDAQ, foreign start-ups 152 NASDAQ Index 157 National Center for Supercomputing Applications (NCSA) 230–32 national system policy 45 Nationale Vereniging van Banken 197 natural monopoly 50 Nazi rule diamond sector during 78 end of trajectory 81–3

Index NCSA 230–32 NE England, fewer Jacobian clusters 38 Nederlandsch Economisch-Historisch Archief 197 Nelson, R.R. 141, 145, 160, 166, 193, 327 neo-Schumpeterian, theory of regional evolution 21 neoclasical world, the state and 327 Netherlands banking sector 192 foreign banks and 202 list of banks 196–7 spinoff entrants 206 Netscape and Paypal 219, 228, 230 ‘network regions’ 23 networks 1, 166, 181, 217–18 of competence 335 local informal 104 regional interactions between 260 social capital and 39 university 233 New York 84, 86 non-academic job movers 122, 124 non-start-up companies, IPOs in NASDAQ 152 North American firms 170, 176 North Carolina 228 North German Broadcasting (Norddeutscher Rundfunk NDR) 55 North Jutland 6, 19, 26, 37 Jacobian cluster 30–31 mobile telephony 40 railroadization 30, 39 solar thermal energy cluster 29 wind-turbine cluster 26–8 North Rhine-Westphalia (NRW) 335–7, 340 North-West Baden-Württemberg 246 Northern California, Jacobian clusters 25 Northwest German Broadcasting (Nordwestdeutscher Rundfunk NWDR) 59, 63–4 Norway 6 cleantech 35–6, 40 Mongstad, CO2 capture testing facility 35–6

367

non-cluster case 19 REC firm for solar energy 36 salmon 171–2, 174–6, 180 Nuremberg Programme (1994) 338, 342 OCS 141 policy promoted demand for R&D 148 regular R&D support program 146, 149–50 Office of the Chief Scientist, Ministry of Industry and Trade see OCS Office of Corporate Relations (OCR) 225 on-the-job-learning 193 open systems 23 organizational routines 48–50 original equipment manufacturer (OEM) 274–5 Oslo peace agreements 157 Owen-Smith, J. 1, 3, 170, 220 Oxford, job mobility 104 Oxfordshire British motorcar sports industry 3 job turnover low 127 Paderborn (Germany), IT cluster 102 Palestine 81, 83, 85, 87 patents Cambridgeshire IJP 268, 273, 276, 280 knowledge transmission 215, 231 Vitamin D 225, 230 path creation 298–9 role of entrepreneurs 77 path dependency 298–9 clusters and 74 path-dependent institutional learning 340–41, 345 path-dependent process, key characteristic 75 Perez-Aleman, P. 165–6, 170–71, 174, 176 photovoltaics 34–6 Physical-Technical-Institute (PTI) 248–9 Physics Cavendish Laboratories 280 Pinch, S. 3, 5, 10, 104, 127, 142 pioneer disadvantages 48

368

Emerging clusters

pioneers’ temporary monopoly 51 Piore, M.J. 166, 168, 181, 297 PKS, private broadcaster 61 Plastic Logic 280–81 platform concept 39 Poland, anti-Semitism 89 policy tourism 331, 343 policy transfer 325, 341–2, 345 channels 330–31 degree of 329 determinants of 331–2 mechanisms 329 Porter, M.E. 1–2, 26, 37, 78, 102, 142, 167, 239 cluster policies 330 definition of clusters 241, 326, 335, 341 positive external economies 103 positive externalities 144 labour mobility 105 post ministry (Reichspostministerium) 57 Postbank (later ING Postbank) 200 Potsdam 59 Potsdam/Babelsberg 106 Powell, W.W. 1, 3, 166, 170 pre-emergence processes 141 preferential attachment 260 primate cities 21 printed circuit boards 275, 280 process of emergence, dynamic economies of scale 144 product quality standards 182 productivity contributions, railroads and 18 proximity 19, 217, 260, 266, 275–6 geographical 19, 260 R&D 299 spatial 298, 306, 309, 328, 339 and relational 342 to incubators 49 public choice theory 346 public sector, start-ups and 175 public television station ARD 107 Puppis, M. 51–2, 60, 66 Quaker Oats, WARF licence (1927) 225 qualitative changes, quantitative changes and 153

qualitative co-evolution 160 qualitative dimension, all three phases 145 qualitative interviews, competencies and 247 qualitative method, dynamics 261 ‘Quartier Latin aux Champs, Le’ 308 R&D Bavaria 337 Cambridge 279 expenditure 300 Greater Boston 304 infrastructure 9, 316 labs 160 MoD and 306 Munich 310 performing firms, high-tech activities 141 policy 316 the ‘regional state’ and 22 university 215 Rad Group 150 radio frequency identification (RFID) 280 Radio Television Luxemburg (RTL) 53, 62–4, 66 ‘railroadization’ 6, 17–19, 30 Jutland and 32 Rau, Johannes (PM of North-Rhine Westphalia) 63 re-emergence of a cluster, strategic action and ‘accidental’ events 90 Reagan administration, cut alternative energy research budgets 27 recent advances in RIS research 23–5 Jacobian clusters 25–32 Regenburg city, industrialisation 338 REGION BRAUNSCHWEIG GMBH 339 region-specific characteristics, mobility patterns 130 regional agglomeration, global leading cluster 140 regional clusters definition 165 emergence of 75 regional economic opportunity (Schumpeterian) 20 ‘regional evolution’ 17

Index Regional Growth Concepts (RGCs) 338, 344 ‘regional innovation networks’ 22 regional innovation systems (RIS) 22–3 regional labour pool, small firms and 127 regional learning processes 261 regional stock corporations 57 regularities, clusters and 168, 182 ‘related variety’, industrial districts 20, 24, 30, 38–9 relating railroads, related variety 19–22 relevance of human capital and labour mobility 100–105 reproduction processes, organizational routines 49 reverse spin-off process 127 Rhine-Neckar triangle 243 Rhineland 38, 57 biochip industry 243 Rhineland Palatinate media agency (Anstalt für Kabelkommunikation AKK) 61 Rhineland Palatinate prime minister 62, 67 RIS framework, ‘varieties of innovation’ 23, 39 Romanelli, E. 4–5, 220, 239–40, 260 Rotterdam 198, 202 Rotterdamsche Bank, strategy 198–9 ‘Route des Hautes Technologies’ 308 Russia 89 Russian River valleys, clusters 25 Sabel 166, 168, 171, 181, 297 Sacramento river valleys, varieties of horticulture 25 St Asaph, UK-Dutch steel manufacturer Corus 35 Salmocorp 180–81 Salmofood 181 salmon, water quality and disease transmission 179 salmon cultivation, complex 179 Salmon Technology Institute (INTESAL) 179 Salmones Antártica, salmon aquaculture 175

369

salmon’s lifecycle, technological challenges 175 San Diego 25–6, 220 San Francisco 220 San Francisco Bay Area biotechnology cluster 220 San Joaquin river valleys, varieties of horticulture 25 Sat 1 company 43, 55, 61–2, 64 Cologne and 66 licence in Berlin 62–3 satellite towns 21 Saxenian, A. 1 biochip industry 240–41 California and 26, 39 case studies and 301, 314 clusters in developing countries 165–6 Greater Boston 304 networks 181 Silicon valley and 168 start-ups 143 university clusters 216 Scheldt city 74 Schmitz 165, 169, 181 Schott-Zeiss Institute (1944), penicillin 248 Schumpeter, J. 17, 30 railroadization 39–40 Schumpeterian and Jacobian, relationship 19–20 Schumpeterian literature, co-evolution 160 Schumpeterian logic, clusters and 267 Schumpeterian ‘new combinations’ 6, 37–8 ‘science city’, German reunification and 254 science, technology and higher education infrastructure (STE) 148, 159 Scitex 152 Scotland, salmon 171, 175–6 Scott, A.J. 2, 193, 298–9, 301–2, 308 second generation growth 5, 8, 267 biochip industries 240, 242, 252, 257–8 MPIMG firms 258 processes leading to 260–61 second-mover advantages 48

370

Emerging clusters

seedbeds 49–50, 53, 65 failure to establish 60 RTL and 66 seeding event 214 seeds 216–17, 233 Sequoia 157 Sericol 279 Sharp, Japanese electronics corporation 35 Shockley, Bill, co-inventor of the transistor 303 Siemens 102 Siemens AG 309 Silicon Valley cluster 4, 39, 105, 140, 159, 167–8, 218 attempts to copy 239, 295–6, 298, 301 case study 302–4, 312–13, 317–18 individuals from UIUC 227, 228 Netscape drew managers 233 Netscape and Paypal 219, 233 policy tourism 331 Stanford University and California 216 Singapore 180 small firms, local labour market pool (Thesis 5) 127–8 small and medium-sized enterprises see SMEs SMEs 124 Denmark and 26–7 Germany and 243, 246 Grenoble 307 Stanford University 303 smolts (juvenile salmon), cages in the sea 175, 183 social capital 32, 39 social market economy (soziale Marktwirtschaft) 45 ‘social structure of innovation’ 216 software, development 151–2 Sophia Antipolis (Nice in France) 127, 296, 302, 307–8, 312–13, 314, 317 sourcing of human capital 110 industrial origins of labour inflows 122–4 intra-and interregional labour mobility 116–20 labour inflows into selected clusters 112–16

local labour market pool and its development 110–112 mobility patterns and cluster structure 124–9 South Africa 81 South German Broadcasting (Süddeutscher Rundfunk SDR) 59 South-African diamond deposits, rough diamonds 79 Southern California, Jacobian clusters 26 Southwest Broadcasting (Südwestfunk SWF) 59 Southwest Broadcasting (Südwestrundfunk SWR) 55, 62, 67 Soviet Union, immigration to Israel 151 Spain, Silkeborg turbines 27 spatial agglomeration 298 spatial concentration, system policy and 50 spatial effects 50, 57 spatial proximity, electrotechnology and 306 specialised high-tech firms 266 spin-off dynamics 8 spin-off firms, further spin-offs 277 spin-off processes 8, 9, 10, 49–51, 64, 66, 143, 239 firms formed in first phase 258–9 firms in second phase 267 spin-offs 5, 144, 239–40 competencies 260–61 definition 204 EE&CS 224 firm routines 260 from early entrants 267 IJP sector 267 mechanisms 246 source of variety 241 university 8 spin-outs 252, 257, 265 spinoff companies 195–6 must innovate 272 spinoff dynamics 192, 194, 195 agglomeration economics 191, 194–6 spinoff model 194–5 spinoff processes 143, 195

Index Spitzencluster (leading-edge cluster) competition 324 spotting devices 242, 254–5 spread effects 21 Stam, Erik 217, 266, 272 stand-off situation 103 standardization 7 standards 165, 170 institutional mechanism 182 international markets and 168 Stanford, EE & CS spinoffs 224–5 star scientists, industrial agglomerations 105 start-up-intensive high-tech cluster life cycle model 140 start-ups based on university research 217 excess demand for support 154–5 inventor companies 145 technology categories 221–2 state policies: North Rhine-Westphalia vs. Bavaria 339–41 STE 159 Steenbock, Professor Harry (UW-M) 225, 230, 232 Sternberg, R. 107–8, 295, 300–301, 309–310, 338 stochastic approaches 2 stochastic concepts, mathematical models 75–6, 92 Stoerring, D. 27, 30, 38 Stoiber, Edmund (Bavarian minister) 340 Storper, M. cluster growth dynamics 168–9 Dutch banking 191, 193 hub-and-spoke district 124 labour and 101–2, 116, 120 path creation 299, 309 ‘shifting centres’ 75, 83, 93 spatial clustering processes 47–50, 53 university clusters 216–17 WLO approach 2–4, 76, 239 strategic action chance and other factors 87–9 COFDI and 85 former cluster structures and 90 generating advantages on supply side 84–5 recovery of labour 85–7

371

strategic action and path creation 77–8 Strategic Partnership for Sensor Technology 338 sub-system of firms, ‘exploitation’ 23 SugarGrass, prospect to replace oil 33 Sun Chemical 279 Sunley, P. clusters 1, 167, 239–40, 241, 297, 330 path dependence 3, 5, 75–8, 298–9 Sweden forestry in biofuels cluster in Örnsköldsvik 35 Xaar relocated to 277 Switzerland 275 Syndicate of the Belgian diamond industry (Syndicaat der Belgische Diamantnijverheid) 79 synthetic vitamin D pharmaceuticals, Tetronics 230 Syria 85 system policy (Ordnungspolitik) 44 market entry and exit 51 regulation to promote public interest 52 system policy and adjustment policy, relationship hierarchical 45 tacit knowledge 195, 216–17 Taiwan 180 information technology in Hsinchu region 165–6 targeting VC, Yozma program 147, 160 Technological Incubators 154 Technological Incubators’ Program 147, 152 technological knowledge 165 technology classification, MG&E 221 technology licensing offices (TLOs) 225 technopole 307 telegraphy (Reichtelegraphenverwaltung) 57 televisor satellite ECS 1 60–61 Terman, Fred 303 Teubal, M. 140–41, 143–7, 154, 160 Thailand 74 Third Italy 20, 23, 298 Thyssen Krupp 343

372

Emerging clusters

‘traded interdependencies’ of a cluster 217 training on the job, small firms 127 triggering event 214, 221 Turkey 85 UC Berkeley 224–5 UIUC 215, 218 engineering and physical sciences 228–9 high technology set-ups 226–7 information technology and engineering start-ups 222 Lotus Notes and Eudora 218–19 not retaining entrepreneurs 228 seeds and 233 Silicon Valley and 219 start-ups 228–9 UK 81 diamonds 83, 87, 93 high cost of capital disincentive 275 salmon exports 174 uncodified knowledge, mobility rates and 104 University of Wisconsin-Madison see UW-M University of California, $500 million for Climate Change research from BP 33 University of Illinois at UrbanaChampaign see UIUC university licensed technology 224 university research-centric cluster research university 232–3 specialization 216–17 start-ups 230 technological innovations 233 university spin-offs 216 University of Wisconsin-Madison see UW-M Unix 231 Upper Austria, cluster policy 339 USA 9, 39, 81, 83 biotech clusters 105, 267, 295 biotherapeutics industry 5, 240 bolt for biofuels 33 Cambridge/Boston cluster 220 emigrants in and Antwerp 86–7 Frontier West in nineteenth century 32

IJPs and pumps 275 National Archives and Records Administration 78 railroadization 17 Research Triangle Park (RTP) 305 Research Triangle/NC 296, 301, 304–5, 312–13, 317–18 salmon 171, 174, 177, 180 Sunbelt states 193 university-based clusters 214–15, 234 UW-M 215, 218–19, 222–4 biological and medical sciences 228–9, 235 high technology set-ups 226–7 start-ups 228–30 University Research Park (1984) 225 Vitamin D 230 WARF and Office of Corporate Relations 233 Vanden Daelen, V. 81–2, 86–8 ‘varieties of capitalism’ (VOC) 314–15 variety creation, competencies 247 venture capital see VC Vereniging voor Vrije Diamanthandel 79 vertical disintegration 298 vertically integrated district 124 Videojet (USA) 270–71, 274, 276–7 Virginia 228 Vitamin D 225, 230 Voelzkow, H. 53, 63–5 Volkswagen (VW) 335 VW, Wolfsburg and McKinsey & Co 343 Wales 6, 18–19, 23, 26, 38 bioenergy from crops 32–5, 38 food path dependence 35 solar energy equipment manufacturers 34 Walker, R. Dutch Banking 191, 193 hub and spoke district 124 labour and 101–2, 116, 120 ‘shifting centres’ 75, 83, 93 spatial clustering processes 47–50, 53 WLO 2–4, 76, 239

Index WARF 225, 230, 232 patents 231 Weimar Republic 57–9 Wenting, R. 39, 47, 49–50, 196, 208, 239 clusters 266 fashion industry 192 West German Broadcasting (Westdeutscher Rundfunk WDR) 53 West Germany, sound results from 106 ‘Western Crescent’ west of London 296, 301, 305–6, 312–13, 317 Western Germany, new Länder deviate 335 Willett (Corby UK) 272, 276 window of locational opportunity approach see WLO wine clusters, overlap horticultural zones 37 Wisconsin 220 Angel Network 231 Entrepreneurs’ network 231 Technology Network 231 Wisconsin Alumni Research Foundation see WARF WLO 2, 4, 10, 20, 59, 75, 208 Antwerp 76

373

biochip industries 239 Dutch banking 192–3 new 84 R&D-intensive industries 297 Wolfsburg AG 341 Wolfsburg’s AutoVision concept 343 World Bank, Projects of National Importance program 149 World War II 74 Wowereit, Klaus (Berlin’s mayor) 43 www.biochipnet.com 244, 261 www.biotechnologie.de 244, 261 Xaar 268, 270–71, 272, 276–9 Xennia 268, 272–3 application in RFID 280 taken over by Ten Cate (Netherlands) 277–8 Yiddish, mazl un brokhe (good luck and blessing) 81 Yozma, critical mass (1993) 161 Zeiss 34, 248, 250, 251, 255, 258 ZIMET 248–9 Zone pour l’Innovation et les Réalisations Scientifiques et Techniques (ZIRST) 307 Zucker, L. 1, 4, 90, 105, 216–17