Joern Block Long-term Orientation of Family Firms
GABLER RESEARCH Innovation und Entrepreneurship Herausgegeben von Pr...
25 downloads
1767 Views
1MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Joern Block Long-term Orientation of Family Firms
GABLER RESEARCH Innovation und Entrepreneurship Herausgegeben von Professor Dr. Nikolaus Franke, Wirtschaftsuniversität Wien, Professor Dietmar Harhoff, Ph.D., Universität München, und Professor Dr. Joachim Henkel, Technische Universität München
Innovative Konzepte und unternehmerische Leistungen sind für Wohlstand und Fortschritt von entscheidender Bedeutung. Diese Schriftenreihe vereint wissenschaftliche Arbeiten zu diesem Themenbereich. Sie beschreiben substanzielle Erkenntnisse auf hohem methodischen Niveau.
Joern Block
Long-term Orientation of Family Firms An Investigation of R&D Investments, Downsizing Practices, and Executive Pay With a foreword by Prof. Dr. Joachim Henkel
RESEARCH
Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.
Dissertation Technische Universität München, 2009
1st Edition 2009 All rights reserved © Gabler | GWV Fachverlage GmbH, Wiesbaden 2009 Editorial Office: Claudia Jeske | Jutta Hinrichsen Gabler is part of the specialist publishing group Springer Science+Business Media. www.gabler.de No part of this publication may be reproduced, stored in a retrieval system or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the copyright holder. Registered and/or industrial names, trade names, trade descriptions etc. cited in this publication are part of the law for trade-mark protection and may not be used free in any form or by any means even if this is not specifically marked. Coverdesign: KünkelLopka Medienentwicklung, Heidelberg Printed on acid-free paper Printed in Germany ISBN 978-3-8349-1959-5
Foreword Family firms play an important role in most economies. This fact is due not only to family firms’ share of overall revenues or jobs, but also because they are commonly assumed to be more long-term oriented than comparable non-family firms. This view, however, is largely based on examples and case studies, while large-scale quantitative studies are so far missing. In more detail, we have very limited knowledge about how family and non-family firms compare along various dimensions of long-term orientation such as R&D activity, human resource policy, and incentive pay for executives. Furthermore, it is an open question how the hypothesized long-term orientation of family firms depends on the exact definition of this type of firm, in particular on management by family members and on the extent of family ownership. Jörn Block addresses the above questions in this pioneering book. Well-grounded in theory, he develops hypotheses regarding various dimensions of long-term orientation. Using an impressive data set on US stock-listed firms that draws on several data sources, the author then tests his hypotheses using leading-edge empirical methods. Noteworthy, he not only proceeds by classical null hypotheses testing, but also employs Bayesian econometrics, a so far rather rarely used method. Jörn Block complements his comprehensive empirical work by an elegant principal-agent model of paying a non-family manager in a family firm. The study lends clear support to the hypothesis of a stronger long-term orientation of family firms. At the same time, it becomes clear that this result hinges both on the exact definition of ‘family firm,’ in particular on the distinction between family ownership and family management, and on the respective dimension of long-term orientation under consideration. I leave it to the reader to explore the numerous interesting and important results. This book is Jörn Block’s doctoral thesis at Technische Universität München, and it lays the foundation for a promising academic career. The study provides valuable insights for management as well as for research, and I strongly recommend it to practitioners and academics alike.
Prof. Dr. Joachim Henkel
Preface Family firms represent a large number of businesses in the economy. They are considered a source of patient capital. Some family firms exist for more than 200 years and have been successfully transferred from one generation to the next generation. Some of them have never been owned or managed by non-family members. Still, they are often market leaders in their field. My dissertation analyzes this phenomenon and investigates whether and under which conditions family firms have more long-term oriented strategies relative to other firms. My dissertation would not have been completed without the help of others. I take this opportunity to thank those who have helped me in various ways in creating this dissertation. First of all, I would like to thank Prof. Joachim Henkel, my dissertation advisor, who provided me always with his full support throughout the entire process. His fascination for research and his interdisciplinary approach to social sciences has led to an environment at his chair which is highly stimulating and supportive and which ultimately sets the ground for cutting-edge research. I am also indebted to him for the idea of analyzing employment downsizing in the context of this dissertation (Chapter 7). I am also thankful to Prof. Ann-Kristin Achleitner, my second advisor. She was highly supportive throughout the entire process. She also encouraged a fruitful conversation between me and members of her chair who work on similar topics. Her openness for cooperation contributed very much to the successful joint bid for a DFG grant about the strategies of family firms. Finally, I would also like to thank Prof. Isabell Welpe, who chaired my dissertation committee. I thank her in particular for her insights about the ‘world of academics’ offered to me at various conferences. The selection of the topic and the first steps of the dissertation are of crucial importance. In addition to the intense discussions with my advisors, I also discussed the topic with experts in the fields of entrepreneurship and family business. I would like to thank Prof. Marc Gruber, Prof. Peter Jaskiewicz, and Prof. Sabine Klein for the highly constructive and inspiring discussions about my research ideas. I would also like to thank Prof. Sabine Klein and her team at the EBS European Family Business Center for the great hospitality during my research visits. For similar reasons, I would like to thank Prof. Philipp Köllinger and Prof. Roy Thurik, who hosted me as I worked on my dissertation during my longer research stay at Erasmus University Rotterdam from March to May 2008.
VIII
A large part of this dissertation is empirical in nature. I would like to thank Thomas Daffner, Friedrich Große-Dunker, Andreas Riemann, Frank Spiegel, and Marc Weiglein for their help in collecting the dataset. I also acknowledge support from the SFB 649 “Economic Risk” at Humboldt-University Berlin, in particular the Data Service Center who provided very helpful support with the financial databases. Throughout the entire dissertation, I had various discussions with other researchers working on similar topics. I would like to thank Markus Ampenberger, Oliver Klöckner, and Dr. Eva Nathusius for their inspiring discussions. I am also thankful to Andreas Thams who made me familiar with the Bayesian approach of thinking about statistics. He is also a coauthor of a paper that is included in this dissertation (Chapter 6). Finally, I owe great thanks to Philipp Sandner, with whom I cooperated on various side projects during the time of my dissertation. The work with him has been a great experience and rewarding in many ways. With our joint papers, we had a steep learning curve, from which my dissertation benefited strongly. I also thank him for giving very valuable and constructive feedback on later versions of my dissertation. I am also grateful to my colleagues Oliver Alexy, Florian Jell, Simone Käs, Stefanie Pangerl, Manuel Sojer, Dr. Marcus Wagner, Johannes Wechsler, and Evelin Winands for a very supportive team environment. Particular thanks go to Oliver Alexy and Dr. Marcus Wagner who engaged in many constructive discussions about the details of my dissertation. Above all, I wish to thank my parents and my brother for their continued support (my father has read and commented the entire dissertation). I am also indebted to my girlfriend and future wife Maria Cristina for her love and patience. This book is dedicated to you.
Jörn Hendrich Block
Table of contents Foreword .......................................................................................................................V Preface ....................................................................................................................... VII Table of contents......................................................................................................... IX List of figures .......................................................................................................... XIII List of tables ...............................................................................................................XV List of abbreviations...............................................................................................XVII Zusammenfassung .....................................................................................................XX 1
Introduction........................................................................................................... 1
1.1 Motivation ...................................................................................................................................... 1 1.1.1 Example of the Freudenberg Group ................................................................................................ 3 1.1.2 Discussion in the media ................................................................................................................ 4 1.2
Research goals ................................................................................................................................ 6
1.3
Structure of the dissertation.............................................................................................................. 6
2
Family firms .......................................................................................................... 9
2.1 Definition of a family firm ................................................................................................................ 9 2.1.1 Background information ............................................................................................................... 9 2.1.2 Definitions used in this thesis ...................................................................................................... 15 2.2
Family influence ............................................................................................................................ 19
2.3 Theories about family firms ............................................................................................................ 21 2.3.1 Agency theory .......................................................................................................................... 21 2.3.2 Stewardship theory .................................................................................................................... 36
3
Long-term orientation of firms ......................................................................... 42
3.1 Key terms ..................................................................................................................................... 42 3.1.1 Intertemporal choice .................................................................................................................. 42 3.1.2 Short-termism, myopia, and long-term orientation ........................................................................... 43 3.2
Dimensions of long-term orientation ................................................................................................ 44
X
Table of contents
3.3 Long-term orientation in family firms: a discussion ........................................................................... 46 3.3.1 Management practices................................................................................................................ 46 3.3.2 Managerial opportunism ............................................................................................................. 47 3.3.3 Executive compensation ............................................................................................................. 48 3.3.4 Information asymmetry and signaling ........................................................................................... 49 3.3.5 Impatient capital, stock market myopia.......................................................................................... 50 3.3.6 Further explanations from family business theory ............................................................................ 51 3.3.7 Summary of the discussion ......................................................................................................... 53 3.4
Research questions ........................................................................................................................ 54
4
Data and method................................................................................................. 58
4.1
Construction of the dataset ............................................................................................................. 58
4.2 Descriptive analysis of the dataset ................................................................................................... 60 4.2.1 Types of family firms and dimensions of family influence ................................................................ 60 4.2.2 Characteristics of firms .............................................................................................................. 62 4.2.3 Governance and CEO characteristics ............................................................................................ 64 4.2.4 Industry classification ................................................................................................................ 66 4.3 Method: Bayesian analysis.............................................................................................................. 68 4.3.1 Introduction to Bayesian analysis ................................................................................................. 68 4.3.2 Differences between Bayesian and classical analysis ........................................................................ 69 4.3.3 Motivation to use Bayesian methods ............................................................................................. 72 4.3.4 Description of the Bayesian model estimated .................................................................................. 74
5
Performance of family firms.............................................................................. 76
5.1
Literature review........................................................................................................................... 76
5.2 Family firms versus non-family firms .............................................................................................. 77 5.2.1 Univariate analysis .................................................................................................................... 78 5.2.2 Multivariate analysis.................................................................................................................. 81 5.3
Impact of family management and family ownership ......................................................................... 89
5.4
Summary and conclusions .............................................................................................................. 92
6
Family firms and R&D spending ...................................................................... 94
6.1 Introduction ................................................................................................................................. 94 6.1.1 R&D spending, innovation, and long-term orientation ...................................................................... 94 6.1.2 What does R&D spending measure? ............................................................................................. 96 6.1.3 Studies using R&D expenditures as a proxy to measure long-term orientation....................................... 97 6.2 Theory and hypotheses ................................................................................................................... 98 6.2.1 Impact of family ownership on R&D spending ............................................................................... 98 6.2.2 Impact of family management on R&D spending .......................................................................... 100 6.2.3 Impact of family firm age on R&D spending ................................................................................ 102 6.3 Data ........................................................................................................................................... 103 6.3.1 Sample .................................................................................................................................. 103 6.3.2 Measures ............................................................................................................................... 106
Table of contents
XI
6.4 Results ....................................................................................................................................... 107 6.4.1 Univariate analysis .................................................................................................................. 107 6.4.2 Multivariate analysis................................................................................................................ 109 6.5 Discussion................................................................................................................................... 126 6.5.1 Implications for theory ............................................................................................................. 126 6.5.2 Implications for practice ........................................................................................................... 130 6.6
Summary and conclusions ............................................................................................................ 131
7
Family firms and employment downsizing .................................................... 133
7.1
Introduction ............................................................................................................................... 133
7.2
Family firms and their employees: a literature review ...................................................................... 135
7.3 Theory and hypotheses ................................................................................................................. 136 7.3.1 Family management and its influence on downsizing ..................................................................... 136 7.3.2 Family ownership and its influence on downsizing ........................................................................ 138 7.4 Data ........................................................................................................................................... 139 7.4.1 Sample .................................................................................................................................. 139 7.4.2 Measures ............................................................................................................................... 139 7.5 Results ....................................................................................................................................... 140 7.5.1 Univariate analysis .................................................................................................................. 140 7.5.2 Multivariate analysis................................................................................................................ 144 7.6 Discussion................................................................................................................................... 150 7.6.1 Implications for theory ............................................................................................................. 150 7.6.2 Implications for practice ........................................................................................................... 151 7.7
Summary and conclusions ............................................................................................................ 152
8
Family firms and executive pay....................................................................... 154
8.1
Introduction ............................................................................................................................... 154
8.2
Related literature ........................................................................................................................ 156
8.3 Data ........................................................................................................................................... 157 8.3.1 Sample .................................................................................................................................. 157 8.3.2 Measures ............................................................................................................................... 158 8.3.3 Regression models .................................................................................................................. 158 8.4 Results ....................................................................................................................................... 159 8.4.1 Descriptive statistics and univariate analysis................................................................................. 159 8.4.2 Multivariate analysis – full sample ............................................................................................. 164 8.4.3 Multivariate analysis – only non-family CEOs .............................................................................. 171 8.5 Discussion................................................................................................................................... 172 8.5.1 Implications for theory ............................................................................................................. 172 8.5.2 Implications for practice ........................................................................................................... 173 8.6
Summary and conclusions ............................................................................................................ 174
XII
Table of contents
9
Executive pay in family firms: a principal-agent model............................... 176
9.1
Introduction ............................................................................................................................... 176
9.2 Literature review......................................................................................................................... 178 9.2.1 Non-family executives in family firms ........................................................................................ 178 9.2.2 Executive pay in family firms .................................................................................................... 179 9.2.3 Principal-agent models on executive pay ..................................................................................... 180 9.2.4 Principal-agent models on managerial myopia .............................................................................. 181 9.3
The model................................................................................................................................... 181
9.4 Analysis...................................................................................................................................... 186 9.4.1 General compensation rules ...................................................................................................... 187 9.4.2 Compensation of a non-family executive ..................................................................................... 190 9.4.3 Family versus non-family executives .......................................................................................... 191 9.5 Discussion................................................................................................................................... 193 9.5.1 Implications for theory: avenues for further empirical research ........................................................ 193 9.5.2 Implications for practice: some guidelines for family business owners ............................................... 195 9.6
Summary and conclusions ............................................................................................................ 196
10 Summary, implications, and outlook .............................................................. 197 10.1 Summary.................................................................................................................................... 197 10.2 Limitations ................................................................................................................................. 200 10.3 Implications ................................................................................................................................ 200 10.3.1 Theoretical implications ........................................................................................................... 200 10.3.2 Practical implications............................................................................................................... 204 10.4 Directions for further research ...................................................................................................... 206
Bibliography.............................................................................................................. 209 Appendix ................................................................................................................... 237 List of firms in the dataset ..................................................................................................................... 237 List of variables .................................................................................................................................... 238 MatlabTM code for Bayesian random-effects model ................................................................................... 247 Correlation table .................................................................................................................................. 250
List of figures 2-1
Trade-off between firm value and private benefits for an owner-manager
2-2
Owner-manager sells a fraction of her equity
25
2-3
Rational expectations of the equity market
26
2-4
Agency costs of equity
27
2-5
Role of monitoring activities
28
2-6
Family business group organized in a pyramidal structure
35
4-1
Number of family firm observations in the sample
60
5-1
Bayesian regression of the firm’s market-to-book value
5-2
Effect of family firm characteristics on the firm’s market-to-book value (Bayesian analysis)
6-1
Family versus non-family firms – Bayesian RE regression of R&D/assets
6-2
Effect of family management and family ownership on R&D/assets (Bayesian model)
117
6-3
Effect of different degrees of family ownership on R&D/assets
118
6-4
Interaction effect
119
6-5
Impact of firm age on R&D/assets in family and non-family firms
121
6-6
Family versus non-family firms – Bayesian RE regressions on R&D/sales
122
6-7
Effect of family management and family ownership on R&D/sales (Bayesian model)
123
7-1
Histogram of workforce decrease by family ownership
145
8-1
Box plot of total pay
160
8-2
Median CEO pay over time
161
8-3
Histograms regarding the structure of executive pay
162
9-1
Sequence of events
185
9-2
Sensitivity analysis
189
A-1
TM
Matlab
code
24
86, 87 91 112, 113
247-249
List of tables 2-1
Family firm definitions used in the literature
11
2-2
Family firm definitions used in this thesis
18
2-3
Dimensions of family influence used in this thesis
21
2-4
Differences between agency and stewardship theory
38
3-1
Concept of long-term orientation in this thesis
44
3-2
Arguments regarding long-term orientation of family firms
54
4-1
Changes between the different firm categories
61
4-2
Channels through which the family influences the business
62
4-3
Firm characteristics by family firm definition
63
4-4
Governance and CEO characteristics by family firm definition
65
4-5
Family and non-family observations by industry
67
4-6
Differences between classical and Bayesian analysis
72
5-1
Arguments regarding superior performance of family firms
77
5-2a Financial performance in family and non-family firms (group of non-family firms remains fixed)
79
5-2b Financial performance in family and non-family firms (group of non-family firms changes with the family firm definitions used)
80
5-3
Random-effects GLS regressions of return on assets
83
5-4
Random-effects GLS regressions of the firm’s market-to-book value
85
5-5
Effect of family firm characteristics on the firm’s market-to-book value
6-1
Family and non-family firm observations by R&D-intensive industry
6-2
Characteristics of (R&D-intensive) family and non-family firms
108
6-3
Family versus non-family firms – classical RE regressions of R&D/assets
110
90 105
6-4
Family versus non-family firms – Bayesian RE regressions of R&D/assets
111
6-5
Effect of family firm characteristics on R&D/assets (classical RE model)
115
6-6
Effect of family firm characteristics on R&D/assets (Bayesian RE model)
116
6-7
Effect of family firm age on R&D/assets (Bayesian RE model)
120
XVI
List of tables
List of tables (continued) 6-8
Effect of being a family firm on R&D/assets (alternative family firm definitions included in one model)
125
7-1
Summary statistics and correlations
142
7-2
Univariate analysis
143
7-3
Random-effects logit regressions of workforce decreased (>0.1%, >5%, >6%, >8%, and >10%)
146
7-4
Heckman model
149
8-1
Level and structure of executive pay
161
8-2
Structure of executive pay – family versus non-family firms
163
8-3
Regressions of total pay (in $000s)
166
8-4
Median regressions of the structure of executive pay – full sample
167
8-5
Median regressions of the share of base salary (in %) – family versus non-family firms
168
8-6
Median regressions of the share of annual bonus (in %) – family versus nonfamily firms
169
8-7
Median regressions of the share of stock option pay (in %) – family versus nonfamily firms
170
8-8
Median regressions of the structure of executive pay – only non-family CEOs
171
9-1
Notation used in the model
186
9-2
Numerical example comparing a family and a non-family executive
A-1
List of firms in the dataset
238-242
A-2
List of variables used in the empirical analyses
243-246
A-3
Correlation table
193
250
List of abbreviations AG
Aktiengesellschaft
ASQ
Administrative Science Quarterly
Avg.
Average
BLUE
Best linear unbiased estimator
Bn
Billion
Bzw.
Beziehungsweise
CARA
Constant absolute risk aversion
CD
Compact disc
CEO
Chief executive officer
CFO
Chief financial officer
COO
Chief operating officer
Corp.
Corporation
Co.
Company
CSR
Corporate social responsibility
DVD
Digital versatile disc
DEF 14A
Definitive proxy statement 14A
DFG
Deutsche Forschungsgemeinschaft
DM
Deutsche Mark
Dr.
Doktor
EBITDA
Earnings before interest, taxes, depreciation, and amortization
Ed./eds.
Editor/ editors
Et al.
Et alii (and others)
Etc.
Et cetera (and so on)
E.g.
Exempli gratia (for example)
FBN
Family Business Network
FEDC
Financial and Economic Data Center
FGF
Förderkreis-Gründungsforschung e.V.
F-PEC
Family – Power, Experience, and Control
F&E
Forschung und Entwicklung
GAAP
Generally accepted accounting principles
GLS
Generalized least squares
XVIII
List of abbreviations
List of abbreviations (continued) IFERA
International Family Enterprise Research Academy
IFERA
International Family Enterprise Research Academy
Inc.
Incorporated
I.e.
Id est (that is)
ISO 9002
International Organization for Standardization 9002
LR-test
Likelihood-ration test
Max.
Maximum
MCMC
Markov chain monte carlo techniques
Min.
Minimum
MTB
Market-to-book value
Mn
Million
M&A
Mergers & acquisitions
NASDAQ
National Association of Securities Dealers Automated Quotation System
No.
Number
NYSE
New York Stock Exchange
N/a
Not available
Obs.
Observations
OLS
Ordinary least squares
P./ pp.
Page/ pages
PPE
Property, plant, and equipment
Prob. (ß), P(ß)
Probability of ß
RE
Random-effects
ROA
Return on assets
ROE
Return on equity
ROI
Return on investment
SE
Standard error
SEC
Securities and Exchange Commission
SFB
Sonderforschungsbereich
SIC
Standard Industrial Classification
STATA
Stata is a general-purpose statistical software package
Std. dev.
Standard deviation
S&P 500
Standard & Poors 500 stock market index
S&P 1500
Standard & Poors 1500 stock market index
List of abbreviations
XIX
List of abbreviations (continued) S&P Global 1200
Standard & Poors Global 1200 stock market index
UK
United Kingdom
US
United States
VIF
Variance inflation factor
Vs.
Versus
VW
Volkswagen
Yrs
Years
WHU
Wissenschaftliche Hochschule für Unternehmensführung
Z.B.
Zum Beispiel
Zusammenfassung Eine große Anzahl von Unternehmen weltweit befindet sich in Familienbesitz. In Deutschland z.B. sind mehr als 60% aller Unternehmen mit einem Umsatz von mehr als 1 Million Euro der Gruppe der Familienunternehmen zuzurechnen. Eine wachsende Zahl von Untersuchungen vergleicht Familien- und Nichtfamilienunternehmen bezüglich ihres finanziellen Erfolges. Eine langfristiger ausgerichtete Unternehmensstrategie wird in diesem Zusammenhang als ein Wettbewerbsvorteil von Familienunternehmen im Vergleich zu Nichtfamilienunternehmen angesehen. Es existieren bereits einige qualitative Arbeiten insbesondere über US-amerikanische Familienunternehmen zu diesem Thema. Evidenz im Rahmen großzahlig-quantitativer Untersuchungen gibt es bisher jedoch nicht. Diese Dissertation leistet einen Beitrag zur Schließung dieser Lücke und untersucht mit Hilfe von großzahligen Paneldaten aus den USA, ob und unter welchen Bedingungen Familienunternehmen langfristiger handeln als Nichtfamilienunternehmen. Langfristigkeit ist ein multidimensionales Konzept und zeigt sich in vielen Aspekten der Unternehmensstrategie. Diese Arbeit fokussiert auf Aspekte der Langfristigkeit im Bereich des Innovationsmanagements, des Personalmanagements sowie der Bezahlung von Führungskräften. Es wird empirisch untersucht, ob Familienunternehmen mehr Ressourcen in F&E investieren, seltener (in großem Maßstab) Personal abbauen und ihre Führungskräfte weniger stark anhand kurzfristiger Erfolgsindikatoren bezahlen. Die empirische Untersuchung findet Evidenz für die Vermutung, dass Familienunternehmen eine langfristiger ausgerichtete Unternehmensstrategie verfolgen als vergleichbare Nichtfamilienunternehmen. Die Ergebnisse unterscheiden sich jedoch sehr stark nach der verwendeten Familienunternehmensdefinition. Je enger die Gruppe der Familienunternehmen definiert ist, desto eher lässt sich Evidenz für eine langfristiger ausgerichtete Unternehmensstrategie finden. Unternehmen, in denen die Familie einen hohen Eigentumsanteil hält (mehr als 30%), investieren mehr Ressourcen in F&E als andere vergleichbare Unternehmen. Diese Unternehmen bezahlen ihre Manager auch seltener in Form von Aktienoptionen. Bei Verwendung einer weiter gefassten Familienunternehmensdefinition, lassen sich diese Zusammenhänge jedoch nicht finden. Die Ergebnisse der Untersuchung zeigen auch, dass sich die beiden Familienunternehmensdimensionen Familienmanagement und Familieneigentum in ihrer Wirkung auf die Langfristigkeit der Unternehmensstrategie deutlich unterscheiden. Familieneigentum hat einen positiven Einfluss auf die Höhe der F&E-Ausgaben und verringert die Wahrscheinlichkeit großzahligen Personalabbaus. Des Weiteren sinkt der Anteil von Aktienoptionen in der Managervergütung. Familienmanagement hingegen hat keinen (oder nur ei-
Zusammenfassung
XXI
nen geringen) Einfluss auf die Höhe der F&E-Ausgaben, das Durchführen von Personalabbauprogrammen und der Struktur der Managervergütung. Interessanterweise zeigt sich auch, dass das Alter eines Familienunternehmens einen Einfluss auf die Langfristigkeit der Unternehmensstrategie hat. Ältere Familienunternehmen investieren im Durchschnitt weniger Ressourcen in F&E als junge Familienunternehmen. Dieser Zusammenhang konnte in der Gruppe der Nichtfamilienunternehmen nicht gefunden werden. Die Arbeit schließt mit einem theoretischen Kapitel. Mit Hilfe eines formalen PrinzipalAgenten-Modells wird untersucht, wie ein Manager eines Familienunternehmens, der nicht zum Kreis der Familieneigentümer zählt, entlohnt werden sollte. Es wird dabei unterstellt, dass die Familie das Unternehmen langfristig entwickeln möchte und dass der Fremdmanager seinen Wert auf dem Markt für Führungskräfte steigern will. Das Modell erlaubt eine Reihe von normativen Aussagen zur Gestaltung von Vergütungsverträgen in einer solchen Situation. Das optimale Niveau an kurzfristigen Erfolgsanreizen verringert sich in dem Ausmaß, in dem der Fremdmanager Erfolgssignale an den Markt für Führungskräfte senden möchte. Des Weiteren hängt das optimale Niveau an Anreizen für kurzfristigen Erfolg von der Risikoeinstellung des Managers, der Beobachtbarkeit seiner Anstrengungen für kurzfristigen Erfolg und seiner Reaktion auf Anreize im Allgemeinen ab. Die Ergebnisse der empirischen Untersuchung und des theoretischen Modells sind relevant für die Theorie und Praxis. Aus Sicht der Theorie liegt der Wert der Arbeit darin, dass die vielfach unterstellte Annahme einer stärkeren Langfristigkeit von Familienunternehmen gegenüber Nichtfamilienunternehmen mit empirischen Daten überprüft wird. Es zeigt sich, dass die Aussage „Familienunternehmen handeln langfristiger“ nicht für alle Familienunternehmen gleichermaßen zutrifft und Fallunterscheidungen nötig sind. Die Ergebnisse dieser Arbeit stellen das Wissen zu dieser grundlegenden Annahme somit auf eine solidere Basis. Die Arbeit trägt zur Management- und Ökonomieliteratur gleichermaßen bei. Aus Sicht der Praxis ist diese Arbeit hilfreich für Eigentümer von Familienunternehmen und ihren Vertretern gegenüber der Politik. Die Ergebnisse der Arbeit zeigen, dass Familienunternehmen, sofern eng definiert, in der Tat eine langfristiger ausgerichtete Unternehmensstrategie verfolgen und somit für das Wohl einer Region von besonderer Bedeutung sein können. Familienunternehmen können die Ergebnisse der Arbeit in ihrer Darstellung gegenüber anderen wichtigen Stakeholdergruppen (z.B. Mitarbeiter, Kunden oder Zulieferer) verwenden. Diese wiederum gewinnen ein umfassenderes Bild über die unternehmerischen Ziele von Familienunternehmen. Zu guter Letzt können Familienunternehmenseigentümer die Ergebnisse des theoretischen Modells zur Bezahlung von Fremdmanagern verwenden, um ihre bestehenden Vergütungsverträge ggf. anzupassen bzw. kritisch zu hinterfragen.
1 Introduction 1.1
Motivation Many firms around the world, both small and large, are family firms. For example, more than
60% of all German firms with a revenue greater than one million Euro belong to the group of family firms (Klein, 2000). In the US, 20% of Fortune 500 firms have a family as the largest holder of voting rights (Villalonga and Amit, 2006).1 Family firms and their role in the economy have been widely debated in the academic literature. In his seminal works The Visible Hand: The Managerial Revolution in American Business2 and Scale and Scope: The Dynamics of Industrial Capitalism3, the renowned business historian Alfred DuPont Chandler argues that large firms run by a cadre of salaried managers are managed better and care more about their long-term competitive advantage as compared to ‘personal enterprises,’ which he refers to as “firms managed by individuals or by a small number of associates, often members, of founder’s families, assisted by only a few salaried managers” (Chandler, 1990, p. 235). Chandler attributes Britain’s industrial decline relative to the US and Germany before World War II to the strong persistence of family firms in that country.4 He further argues that the shift toward salaried managers running large enterprises is responsible for the growth and strength of US industries in the early years of modern capitalism (1850-1920). Referring to British family firms, Chandler states that their “goal […] appears to have been to provide a steady flow of cash flow to owners – owners who were also managers” (1990, p. 390). Short-term income was preferred to long-term growth, dividend payments were high and retained earnings were low. By contrast, “in American managerial firms the basic goals appear to have been longterm profit and growth” (Chandler, 1990, p. 390). He argues that salaried managers are more capable of dealing with ‘dividend-hungry owners’ than managers who are also members of the owner-
1
2 3 4
For further evidence regarding the importance of family firms, see the following studies: US (Shankar and Astrachan, 1996; Shleifer and Vishny, 1986), Western Europe (Andres, 2008; Becht and Roell, 1999; Faccio and Lang, 2002; Franks and Mayer, 2001; IfM Bonn, 2007), Asia (Claessens et al., 2000, 2002), and 27 industrialized countries (La Porta et al., 1999). See Chandler (1977). The book received the Pulitzer Prize in 1978. See Chandler (1990). This argument has been criticized by business historians for overstating the number of family firms in Britain while understating the number of family firms in the US. For criticism from various perspectives, see, e.g., Alford (1994), Church et al. (1990), or Church (1993).
2
1 Introduction
ship family. The latter are more vulnerable to pressure from family members uninvolved in the business, and thus favor high dividends. Michael Porter, author of the seminal work The Competitive Advantage of Nations5, takes a different point of view. He argues that the US system of allocating investment capital is failing (Porter, 1992). US firms invest too little, in particular into the assets and capabilities required for competitiveness, including R&D, employee training, and supplier relations. This underinvestment places US firms at a competitive disadvantage relative to German or Japanese firms. He argues that the main reason for this development in the short-term relationship between US firms and external capital providers is as follows. About 60% of total equity is owned by institutional investors (e.g., mutual funds or pension funds), which have highly diversified portfolios with only small stakes in each firm. The average holding time of shares is only 1.9 years.6 Investors focus on financial goals over the near term. They base their investment choices on limited information oriented towards predicting short-term stock price movements, focusing on easily measurable figures, such as current earnings. Ultimately, the funds supplied by the institutional investors move rapidly between firms. In contrast, this is not the situation in Japan or Germany. Porter argues that firms in these two countries have ‘dedicated owners’ who act as principals rather than as agents. These owners “hold significant stakes, rather than small fragmented positions […] and seek long-term appreciation of their shares, which they hold in perpetuity” (Porter 1992, p. 70). Although Porter did not explicitly refer to family owners as a group, family owners resemble in many aspects Porter’s group of ‘dedicated owners.’7 In particular, family owners possess the desire to pass the firm on to the next generation, leading them to think long-term (Casson, 1999; Guzzo and Abbot, 1990; James, 1999; Tagiuri and Davis, 1992). For these and other reasons explained below, scholars in the fields of business history (e.g., Berghoff, 2006; Casson, 1999), economics or finance (e.g., Anderson and Reeb, 2003; Bertrand and Schoar, 2006; Böttcher and Linnemann, 2008; James, 1999), and management (e.g., Le BretonMiller and Miller, 2006; Miller and Le Breton-Miller, 2005; Simon et al., 2005; Zellweger, 2007) have followed Porter’s view that family firms are more oriented towards the long-term than nonfamily firms.8 However, apart from some qualitative research, no large-scale empirical study has analyzed this basic assumption. This dissertation aims to close this gap, analyzing to what degree 5 6 7
8
See Porter (1998). The numbers are taken from Porter (1992) and refer to the year 1990. Other important blockholders in Germany have been banks and insurance companies. In Japan, suppliers and customers own large stakes in each other in order to cement their business relationship. See Section 3.3.
1 Introduction
3
and under which conditions family firms are oriented towards the long-term. I used the S&P 500 firms as of July 31st, 2003 and collected detailed information about each firm’s ownership and management structures for the years 1992-2003, allowing me to identify family and non-family firms. The resulting dataset was then used to analyze the long-term orientation of family firms along several dimensions. Before describing the research goals and objectives in detail, the topic is motivated using an example of a family firm oriented towards the long-term, as well as through media statements regarding the long-term orientation of family firms. 1.1.1
Example of the Freudenberg Group The Freudenberg Group is an interesting example of a family firm oriented towards the long-
term. The example is a clear illustration of the link between family ownership and long-term orientation. On their corporate website, the firm portrays itself as a “family company offering its customers technically challenging product solutions and services.”9 The website also adds that “as a family company, we are guided by our long-term orientation, financial solidity and the excellence of our people in 53 countries around the globe.”10 The Freudenberg Group is a diversified company with a broad spectrum of products, including seals and nonwovens, household products, lubricants, and IT services. The firm was founded in 1849 by Carl Johann Freudenberg and his partner Heinrich Christian Heintze in the Müllheim valley near Weinheim, Germany.11 Today, the firm is a conglomerate consisting of more than 400 individual business groups. It has more than 34,000 employees and reports sales of more than € 5.3 million. Still, in most of its characteristics, the firm resembles a family firm. The parent company that controls the conglomerate is governed by a board of partners which “consists of between seven and a maximum of 13 members […], the majority of which must be members of the Freudenberg family or the spouse of a member of the Freudenberg family.”12 Dr. Wolfram Freudenberg, a member of the founding family, acts as chairman of the board. The firm is not listed on the stock market and is owned by more than 300 individual family shareholders. I am interested in the measures taken by the family members involved with the firm to achieve long-term orientation.
9
See http://www.freudenberg.com (accessed January 18th, 2009).
10
http://www.freudenberg.com/ecomaXL/index.php?site=FCN_EN_portrait (accessed January 18th, 2009).
11 12
Weinheim is a town in the northwest of Baden-Würtemberg, Germany. See http://www.freudenberg.com/ecomaXL/index.php?site=FCO_EN_board_of_partners (acces-sed January 18th, 2009).
4
1 Introduction In an interview with the economic magazine Brand Eins13, Dr. Wolfram Freudenberg, the
chairman of the group, explained the relationship between the family and the firm as follows. Concerning priorities, he pointed to “a golden rule: the firm is first priority, followed by the family, and then the individual.”14 This can be seen in the firm’s dividend policy. The partnership agreement between the members of the family specifies that dividends are capped at 20% of earnings. Regarding the sale of ownership shares, the partnership agreement states that shares can only be sold to other family members. In-laws have to sell their ownership rights if a family member’s marriage ends in divorce. The agreement is valid until the year 2030. Regarding the role of non-family managers, Dr. Wolfram Freudenberg states that “competence is the only thing that counts – regardless of where it comes from.” The CEO of the Freudenberg Group, just like the CEOs of the various business groups, is not a member of the Freudenberg family. Freudenberg states that “the firm has been growing so fast […] there were not enough talents in the family. We needed knowledge from outside.” No family member has the automatic right or privilege to a career in the firm, since family members have to fulfill the same requirements as non-family candidates. Regarding corporate governance, Freudenberg argues that “in its internal relationships, the firm is not different from any publicly listed company. […] Yet, in its external relationships, the firm works like a private company.” That is, the firm does not have to follow any ‘management fashion’ that emerges. For example, the firm has a strongly diversified business portfolio, which is against the recent trend of focusing on the core business alone. All strategic decisions are taken by the board of partners, in which the family has the majority of votes. The family speaks with one voice and aims to avoid family conflicts, which is not easy with more than 300 individual family owners. To this end, the family involves all members through regular meetings and tries to keep them informed of the firm’s developments. The goal is to establish an emotional link to the firm.15 1.1.2 Discussion in the media In recent years, the media has become interested in the group of family firms (e.g., Business Week, 2003; Handelsblatt, 2007a, 2007b; Manager Magazin, 2008a, 2008b; Stuttgarter Zeitung, 2008).16 In this discussion, the issue of the long-term orientation of family firms has played a
13 14 15 16
This interview led to an article, see Sywottek (2008). This statement, as well as those that follow, were translated from German by the author of this thesis. For a deeper discussion of the firm’s unique culture and cohesion, see Pieper (2007). Since I am more acquainted with the German-speaking press, I refer to examples from the German-speaking area. The leading German-language business newspaper, Handelsblatt, devotes a weekly section to the issue of family firms. It has also launched a hall of fame of successful German family entrepreneurs. Also, The Süd-
1 Introduction
5
prominent role (e.g., Handelsblatt, 2007a, 2007b). There are three major ongoing debates in this literature. The first is linked to the stock market performance of owner-dominated firms. The development of style indices representing the performance of owner-dominated, publicly-quoted firms such as the German entrepreneurial index (GEX®)17 or the Swiss Entrepreneurial Index has led to a discussion of whether the particular characteristics of owner-dominated firms lead to stronger stock market performance.18 In this context, a stronger long-term orientation is considered one of several advantages of owner-dominated firms. The second debate concerns ‘hidden champions’ and their importance for the German economy (e.g., Handelsblatt, 2008). Hidden champions are lesserknown companies that have succeeded quietly, and most of them are family-owned (Simon, 2007, 2009). Apart from being relatively unknown and family-owned, hidden champions are often market leaders in their specialized markets. Hermann Simon, the main promoter of the concept, argues that hidden champions are more oriented towards the long-term than other firms. To this end, he cites Karsten Ottenberg, CEO of Giesecke & Devrient19: “We are not thinking about the firm’s development in the next quarter, but about sustainability in terms of generations”20 (Simon 2007, p. 82). Finally, there has been an active debate on the benefits of private equity investments.21 Some family firms and their owners have undertaken investments that appear similar to private equity strategies. However, the public debate does not view them as private equity investment funds. Often, family firms are viewed as the ‘good guys,’ whereas private equity funds such as KKR22 or Permira23 are viewed as the ‘bad guys’24 (Manager Magazin, 2008b). In the public’s view, private equity investors are focused on the short-term and do not care about long-term prospects of the firms in which
deutsche Zeitung, another important daily newspaper, has a family firm rubric on its website. See http://www.sueddeutsche.de/wirtschaft/456/301453/uebersicht (accessed February 1st, 2009). 17
18 19
20 21 22 23 24
The (GEX®) has been developed by the Center for Entrepreneurial and Financial Studies (CEFS) at Technische Universität München together with the Deutsche Börse AG, a stock market exchange. See Achleitner et al. (2005a, 2005b) for the five selection criteria of the GEX® and their theoretical foundation. See Achleitner and Ampenberger (2006) for a comparison of the Swiss and German entrepreneurial indices. Note that the GEX® excludes firms that have been listed on the stock market for more than 10 years, which is why the GEX® is not representative of the entire population of family firms. For a non-academic book on this issue, see Sander (2008). Giesecke & Devrient is a family-owned technology group that produces banknotes, among other products. See http://www.gi-de.com (accessed January 20th, 2009). This statement is translated from German. For a collection of papers dealing with this issue, see World Economic Forum (2008). See http://www.kkr.com (accessed January, 19th, 2009). See http://www.permira.com/en/index.html (accessed January 19th, 2009). They were even referred to as ‘locusts’, only interested in quick returns on their investments.
6
1 Introduction
they invest. Contrary to that, family firms and their owners are viewed as investors that have a strong interest in developing portfolio firms.25
1.2
Research goals The long-term orientation of family firms has both theoretical and practical relevance. Regard-
ing theory, this dissertation aims to contribute to the general management literature on the issue of long-term orientation, to the finance literature on the performance of family firms, and to the family business literature. Further, as the level of R&D spending and the structure of executive compensation are used to measure long-term orientation, I also aim to contribute to the innovation literature on the determinants of R&D spending and to the finance literature on the determinants of executive compensation. From a practical perspective, these results may be relevant for family firms, family business owners, and stakeholders that deal with family firms (e.g., employees, customers and suppliers). A firm’s long-term orientation is a concept with several dimensions and involves many aspects, such as brand building, innovation, and human resource management (see Section 3.2). In order to reflect this multidimensionality, this dissertation studies the long-term orientation of family firms in several ways. I analyze whether family firms spend more on R&D (an indicator of longterm orientation) and whether they are less likely to engage in large-scale employment downsizing (an indicator of corporate myopia). Furthermore, this study investigates to what degree executive pay is linked to short-term measures of performance. Finally, using a principal-agent model, I aim to understand how a non-family manager in a family firm should be compensated given that the ownership family pursues long-term goals. The specific research questions are formulated in Section 3.4. The next section describes the structure of this dissertation.
1.3
Structure of the dissertation This dissertation comprises ten chapters. The following chapter deals with the theoretical
foundations of family firms. Chapter 3 discusses the concept of long-term orientation and presents the specific research questions to be analyzed. Chapter 4 introduces the data and the methods used. Chapter 5 compares the financial performance of family and non-family firms. Chapter 6 aims to understand whether family firms spend more on R&D than other firms. Chapter 7 analyzes whether family firms are less likely to engage in short-term job cuts. Chapter 8 compares the structure of 25
Note that this hypothesis is not proven. There is little empirical research on this issue. For papers that deal with this issue, see Achleitner et al. (2006, 2008a, 2008b, 2008c).
1 Introduction
7
executive pay in family and non-family firms, with a particular focus on incentive pay. Chapter 9 uses a formal principal-agent model to derive the optimal compensation contract of a non-family executive employed by a family firm. Finally, Chapter 10 summarizes the key findings and discusses the main theoretical and practical implications. More specifically, Chapter 2 reviews the extant literature on the definition of a family firm and discusses the particularities of the US corporate governance system. I introduce the specific family firm definitions used in the empirical chapters of this dissertation. Next, the concept of family influence is presented as an alternative to the dichotomous view of family vs. non-family firms, and the specific measures used in this dissertation are described. This chapter closes with a review of agency and stewardship theory, the two most widely used theories in the field of family business research. Chapter 3 discusses a firm’s long-term orientation, the main concept underlying the research questions of this dissertation. Key terms are defined, such as intertemporal choice, myopia, and sustainability, and are distinguished from the concept of long-term orientation. Next, I present my own understanding of a firm oriented towards the long-term and give concrete examples of longterm orientation in various settings, such as brand building or innovation. In a next step, I review the extensive literature on the causes of corporate myopia and relate this literature to the discussion of long-term orientation in family firms. This chapter closes with the specific research questions to be analyzed. Chapter 4 describes the construction of the dataset used in the empirical chapters. Next, the dataset is analyzed in a descriptive way. I rely on univariate statistics to compare family and nonfamily firms according to firm, CEO, industry, and governance characteristics. Then, I provide an introduction to Bayesian methods, which I use in addition to classical statistical methods. Chapter 5 analyzes the financial performance of family and non-family firms. The chapter starts with a literature review of studies on the performance of family firms. Following this review, I present my own results using the data described in Chapter 4. I perform univariate and multivariate analysis in order to determine whether family firms outperform non-family firms and to learn about the impacts of both family management and family ownership on financial performance. I use both classical and Bayesian methods. Chapter 6 focuses on the innovation dimension of long-term orientation and analyzes whether family firms spend more on R&D than non-family firms. The chapter starts with an example that illustrates the link between long-term orientation and R&D spending. After that, I summarize extant work using the level of R&D spending as a proxy for long-term orientation. I refer to family business theory and the literature on long-term orientation to derive hypotheses about the
8
1 Introduction
impacts of family management, family ownership, and family firm age on the level of R&D spending. The empirical investigation starts with a description of the sample and the measures used. The hypotheses are analyzed using univariate and multivariate analyses as well as Bayesian and classical statistical techniques. Finally, I discuss the theoretical and practical implications of these results. Chapter 7 focuses on the relationship between family firms and their employees, another dimension of long-term orientation. More specifically, I analyze whether family firms are less likely to engage in deep job cuts. After some introductory examples regarding the link between corporate myopia and deep job cuts, I summarize what is known about the roles of employees in family firms. I derive theoretical hypotheses on the impacts of family management and family ownership on the likelihood of deep job cuts. The empirical investigation starts with a description of the sample and the measures used. Next, I test the hypotheses with univariate and multivariate analyses. The chapter ends with a discussion of the theoretical and practical implications. Chapter 8 deals with the structure of executive pay in family and non-family firms. After a short introduction, a summary of the related literature, and a description of the data, I present the results of univariate and multivariate analyses. I focus on stock option pay in particular, which has been criticized for its adverse effects on long-term thinking. The final section highlights the theoretical and practical implications of these results. Chapter 9 does not make use of data. Instead, I use a formal principal-agent model to derive the optimal compensation contract of a non-family manager working in a family firm. The chapter is normative, and it is linked to the main research question through the implications for how longterm-oriented family owners should reward short-term-oriented non-family executives. After an introduction to this management problem, I present a review of the literature. The review summarizes research about (1) the role of non-family executives in family firms, (2) the size and structure of executive pay in family firms, (3) principal-agent models on executive pay, and (4) principalagent models explaining managerial myopia. Based on this review, I construct a model to derive the optimal compensation contract. In a next step, I analyze the model and the optimal compensation contract and formulate compensation guidelines resulting from the model. Finally, Chapter 10 concludes with a summary of the main results and contributions of this work. From the key findings, I derive recommendations for family firms, family business owners, and stakeholders that deal with family firms. Furthermore, I discuss the implications of the key findings for management research and present the future research questions resulting from this dissertation.
2 Family firms The goal of this thesis is to analyze the performance and long-term orientation of family firms. Yet, before I can proceed with the empirical analysis of this topic in Chapters 5 to 8, some basic knowledge about family firms is required. This chapter summarizes what is known about family firms, focusing on research regarding the definition of a family firm (Section 2.1), the concept of family influence (Section 2.2) and theories that are used to explain the behavior and performance of family firms (Section 2.3).
2.1 2.1.1
Definition of a family firm Background information The definition of a family firm is a complex issue. The academic literature so far has not
found a satisfactory definition of the family firm (Astrachan et al., 2002; Handler, 1989; Klein et al., 2005). Habbershon and Williams (1999) note that at least 44 different definitions of family firms were used in the literature between 1989 and 1999. Flören (2002) provides an overview of more than 50 definitions; Miller et al. (2007) give an overview of 28 definitions used in the literature until now. Why is it so complicated to define a family firm? As an illustration consider the following three examples that show the broad spectrum of what can be regarded as a family firm. The first example concerns a founder-dominated firm. EBay Inc.26, an American Internet company that manages amongst others an online auction website, was founded in 1995 by Pierre Omidyar. As the founder is still active in the business in the year 2003, it is considered a (founderowned) family firm by Anderson and Reeb (2003) and Villalonga and Amit (2006).27 The second example is about a firm which was founded by the state and is now in the process of being acquired by a family firm. The Volkswagen group (VW), currently Europe’s largest car manufacturer, was founded in 1937 by the German state.28 Adolf Hitler, chancellor (and dictator) of Germany at that time, asked Ferdinand Porsche to design a car that is suitable for the working man. The result was a 26 27
28
See www.ebay.com (accessed September 12th, 2008). Miller et al. (2008) critisize this approach and argue that a founder-owned firm is conceptually different from a family-owned firm. It is argued that founder-owned firms do not suffer from nepotism and inter-generational squabbles and should thus perform better than family-owned firms. See http://www.volkswagen.com (accessed September 12th, 2008).
10
2 Family firms
car later known as “VW Käfer”. Until today, the German state owns a large fraction of shares of the Volkwagen Group and aims to control the firm.29 However, in October 2005, the ancestors of Ferdinand Porsche and owners of the Porsche Group30, a premium car manufacturer, acquired a large equity stake in the firm and attempted a takeover. Ferdinand Piech, the grandson of Ferdinand Porsche and an influential owner of the Porsche group, heads the supervisory board of Volkswagen.31 Finally, as a third example, consider the case of the Bauer AG. The firm was founded by Sebastian Bauer in the year 1790 and is currently owned and managed by the 7th family generation.32 The firm has never been owned or managed by individuals who are not members of the founding family. The three cases show the broad spectrum of what has been considered as a family firm in the literature. The cases have in common that the founder or the ancestors of the founder own a large share of equity. Yet, they differ in the role of the family towards management: The Volkswagen and the Porsche Group are managed by non-family managers33, whereas the Bauer AG is managed by a member of the owning family. Table 2-1 below summarizes the main definitions that have been used in the literature. It is surprising that only few studies use a definition that distinguishes between founder-owned and family-owned firms.
29
30 31 32
33
The federal state Niedersachsen (Lower Saxony) passed the so-called ‘Volkswagen Law’, under which no shareholder of VW can exercise more than 20% of the firm’s voting rights, regardless of their share of voting stock. Note that the European Court of Justice has indicated that the law is against EU rules. See http://www.porsche.com/usa/aboutporsche (accessed February 4th, 2009). Ferdinand Piech is also a former CEO of Volkswagen. See www.bauer.de (accessed September 12th, 2008). The firm went public on July 4th, 2006. The Bauer family still owns more than 50% of stock. Thomas Bauer, a member of the founding family, is CEO of the firm. Both Wendelin Wiedeking (currently CEO of the Porsche Group) and Martin Winterkorn (currently CEO of the Volkswagen Group) are not members of the Porsche or the Piech family.
2 Family firms
11
Table 2-1: Family firm definitions used in the literature
Definition
Study
Region
Stock market listed/nonlisted firms
Family controls > 20% of shares and is the largest shareholder by a factor of 2
La Porta et al. (1999)
World
Listed
Ownership of founding family > 5% or member of founding family is on board
Anderson and Reeb (2003) BusinessWeek (2003)
US
Listed
Top executive is family member or descendant of founder
McConaughy (1994)
US
Listed
Family is the largest voteholder
Villalonga and Amit (2006)
US
Listed
Family holds some ownership, has more than one management position, and is at least in the 2nd generation
Shanker and Astrachan (1996)
US
Both
Family controls > 50% of the votes
Gallo and Sveen (1991)
Spain
Non-listed
Judgment of the person answering the ques- Gallo et al. (2004) tionnaire
Spain
Both
Family controls > 20% of the votes
Faccio and Lang (2002)
Western Europe
Listed
Family controls > 10% of equity
Blondel et al. (2002)
France
Listed
Family controls > 50% of equity
Ehrhardt and Nowak (2003a)
Germany
Listed
% of family ownership + % of family members in supervisory board + % of family members in management > 100 (if family owns at least some shares)
Klein (2000) Achleitner et al. (2007)
Germany
Both
2 individuals (or their family members) hold IfM Bonn (2007) at least 50% of voting shares and are repre- Stiftung Familienuntersented in the management of the firm nehmen (2007)
Germany
Both
For-profit enterprise with family kinships among owners34
Germany
Both
Gabler Wirtschaftslexikon (1992, p. 1099)
The large variety of definitions for family firms causes some problems. Comparisons between studies become difficult, if not impossible. For example, a large number of studies have analyzed whether family firms perform better financially than non-family firms. In their frequently cited paper, Anderson and Reeb (2003) concluded that family firms perform better, or at least not worse, than non-family firms – a conclusion that conflicts with the findings from other studies (e.g., Bennedsen et al., 2007; Holderness and Sheehan, 1988; Perez-Gonzales, 2006). Later, Villalonga and Amit (2006) show that Anderson and Reeb’s results are not robust to alternative definitions of family firms. In a similar study, Miller et al. (2007) show that only family firms
34
The statement is translated from German by the author.
12
2 Family firms
with a lone founder outperform non-family firms.35 This illustrates that, since there is no widelyaccepted, transparent, and easy-to-use definition of the term ‘family firm’, there is a risk of making statements that do not apply to all family firms but only to a small subsample. Furthermore, it is difficult to conduct a quantitative meta-analysis in the spirit of Hunter and Schmidt (2004), which combines the results of several studies that address related research hypotheses. The only way to circumvent this problem is to use several definitions when analyzing the data and determine whether the same conclusions are reached. Another problem with multiple definitions is that it is difficult to test theories. A number of theories predict differences between family and non-family firms36, but it is difficult to test these theories without a clear definition or a commonly-agreed upon set of distinguishing characteristics for family firms. Again, to avoid this problem, it is necessary to compare the results of an empirical analysis using several family firm definitions and to accept only those results that do not change when an alternative definition is used. Another solution is to develop more fine-grained theories that distinguish between different types of family firms, e.g., family-managed versus non-familymanaged, first-generation versus multiple-generation, or small versus large. In this case, the object of investigation would not be a dichotomy between family and non-family firms; rather, particular subgroups of family firms would be compared to other subgroups of family firms or to the entire population of firms. To some degree, this problem parallels those found in early entrepreneurship research, which at first drew similarly stark distinctions between entrepreneurs and non-entrepreneurs (e.g., Carland et al., 2006). Later, it was understood that different types of entrepreneurs exist, e.g., necessity versus opportunity entrepreneurs (e.g., Reynolds et al., 2002)37 or high-tech versus non-high-tech entrepreneurs (e.g., Song et al., 2008).38 A similarly fine-grained approach would benefit research on family firms in two ways: First, the results would become more robust, and replication and metaanalyses would become possible. Second, there would be an opportunity to make contributions in the more general context of management theory, instead of simply testing theories that try to explain the differences between family and non-family firms.
35 36 37 38
Lone-founder firms are firms in which no relatives of a founder are involved. See Section 2.3. See also Block and Wagner (2007), Block and Sandner (2008) or Block and Köllinger (2008). For a similar comment, see Wiklund (2006).
2 Family firms
13
According to the overview in Table 2-1 above, the main features that define a family firm are:
Ownership. How many shares or how much equity does the family own? Is the family the largest shareholder?
Control. How much equity does the family control? Does the family control the majority of the votes? Is there a powerful supervisory or advisory board39 controlling the firm?
Management. What is the influence of the family on the management of the firm? Does the CEO or the chairman come from the family? What percentage of top management is composed of family members?
Family generation. Is the firm a single-generation or a multiple-generation family firm? How many generations are simultaneously involved as owners or managers of the firm?
The exact definition of a family firm should depend upon the precise research context and research questions. The main reasons for making a distinction are:
Size of the firm. The size of the firm determines how large the family’s holdings need to be in order to be classified as a family firm. In terms of veto power, 10% ownership in a large firm is different from 10% ownership in a start-up or a medium-sized firm. The reason is the low voting rate of shareholders in large firms, which effectively may give a 20% ownership stake a much-amplified voting power of 50% and more.40
Listed versus non-listed firm. A firm which is not listed on the stock market does not need to follow the laws and rules of financial market regulatory authorities such as the US Securities and Exchange Commission (SEC) and does not need to follow listing rules of stock exchanges such as the NYSE or the NASDAQ.41 This exemption affects issues such as insider trading, accounting standards, and the level of reporting requirements. Generally, most of these regulations help to strengthen the position of investors with a minority
39
40
41
Advisory boards (in German: Beirat) play an important role in the governance of e.g. German family firms. Often, they have not only an advisory but also a controlling function. For a summary of the literature on advisory boards, see Henseler, 2006; for an empirical study about the role of advisory boards in German family firms, see May et al. (2002). For example, the voting rate in the 2008 annual shareholders meeting of the Siemens AG was 48.5% (Allianz AG: 40.2%). See http://w1.siemens.com/investor/pool/en/investor_relations/events/annual_shareholders_meeting/2008/voting_res ults.pdf or http://www.allianz.com/en/alianz_group/investor_relations/general_meetings/presence_und_voting_results/agm0 8_presence.pdf (both were accessed January 27th, 2009). See Claessens and Tzioumis (2006) for a comparison of listed and non-listed firms.
14
2 Family firms
stake; correspondingly, a 10% ownership stake in a listed firm gives more power than a 10% ownership stake in a non-listed firm. Thus, most researchers agree that the ownership stake to make a firm a family firm should be lower in a listed than in a non-listed firm (Jaskiewicz, 2006). A similar argument can be made about the management aspects of the family firm definition. Ceteris paribus, the position and influence of management in a nonlisted firm is stronger than in a listed firm. The management does not need to explain its management decisions to the capital market. Correspondingly, for a non-listed firm to be classified as a family firm, members of the family should either be members of the firm’s top management team or balance this lack of power by other means such as a high ownership stake or control through a powerful supervisory board.
Ownership structure. A 10% equity stake in a firm with a widely-dispersed ownership gives more power than a 10% equity stake in a firm with concentrated ownership. Furthermore, it makes sense to differentiate between the type of investors owning the firm, e.g., active versus passive investors, or institutional42 versus non-institutional investors. Power dynamics differ depending on whether or not the family’s ownership stake is equaled by the stake held by an active investor, such as a private equity firm.
Corporate governance system. The family firm definition depends also on the firm’s system of corporate governance. As Section 2.1.2 below discusses, corporate governance differs dramatically between nations. The definition of a family firm should differ depending on whether a one-tier or a two-tier board system prevails. In a two-tier board system, the family often uses a strong supervisory board to exert control, whereas a one-tier board system places great importance on the percentage of family members on the board of directors or the important committees (e.g., executive or audit committee).43
The field of study. The research question and the purpose of the study may also affect the family firm definition that is used. If the research focuses on listed firms, as is often the case in studies in the field of finance, then the family firms within the sample will also be restricted to listed firms. Accordingly, the family firm definition should include low ownership thresholds, in line with the preceding arguments about listed versus non-listed firms.
42
43
Institutional investors pool large sums of money and invest those sums in firms. They include banks, insurance companies, retirement or pension funds, hedge funds, and mutual funds. In this thesis, the terms ‘institutional investor’ and ‘financial investor’ are used as synonyms. For a discussion of one-tier and two-tier board systems, see Maassen and Van den Bosch (1999). For a more general survey of corporate governance systems, see Shleifer and Vishny (1997) or Weimer and Pape (1999).
2 Family firms
15
Although the family itself is an integral part of any family firm definition, the term ‘family’ is rarely defined (Astrachan et al., 2002). Marriage, child-adoption, and divorce make this a nontrivial issue. In addition, the definition of a family may differ across cultures and time. For example, the concept of family may mean something different in Western societies than it does in Asian or Latin-American societies. The concept of family may also differ across disciplines. An economist studying fertility and ‘the demand for children’ or the division of labor in households may use a different concept of family than a psychologist concerned with family therapy, or a sociologist who wants to learn about the role of family ties in social behavior. Making matters even more difficult, research into family firms is interdisciplinary and involves economists, sociologists, management scholars, and social psychologists.44 Similar to the previous situations, the only fully inclusive method is to use several definitions of family and to compare the results. 2.1.2
Definitions used in this thesis This thesis analyzes the performance and long-term orientation of large US family and non-
family firms. Since my sample consists exclusively of US firms, the family firm definitions used should be adapted to the US context. Hence, before describing the exact definitions used, I first summarize some relevant and important characteristics of the US system of corporate governance. Of particular importance are the specifics of board structure, the position and power of the CEO, the status of investors, and the way corporate control is exerted.45 Board structure. US firms typically have a one-tier board system.46 The board of directors is responsible for both management and control of the firm. Typically, the CEO also holds the position of chairman of the board. In countries with a two-tier board structure, such as Germany or the Netherlands, this dual role would be impossible. There is a formal division of power between the management and the supervisory board. The chairman heads the supervisory board but not the management board, and cannot simultaneously be the CEO. In the one-tier board system, an important distinction is drawn between inside directors (current employees of the firm such as the CEO), affiliated outside directors (directors with substantial business relations with the firm such as a consultant or an investment banker), and independent outside directors (directors who are neither insiders nor affiliated outsiders such as a retired executive from an non-affiliated firm). The average US 44
45
46
For a discussion of the interdisciplinary nature of the field of family business research, see Habbershon and Williams (1999) or Wortman (1994). For a more detailed description of the particularities of the US and other systems of corporate governance, I refer to La Porta et al. (2000), Mayer (1997), or Roe (1993). Tirole (2006, pp. 29-35) provides an introduction to the one-tier board system. See also Maassen and Van den Bosch (1999).
16
2 Family firms
firm has a nine-person board, of which 40% are inside directors (including the CEO), 20% are affiliated outside directors, and 40% are independent directors (Denis and Sarin, 1999). What does this imply for a US family firm definition? Clearly, it makes little sense to include membership in a supervisory or an advisory board as a criterion in the definition. Instead, what matters is the presence of family members or affiliated individuals in the board of directors and the relevant committees. It is also of crucial importance whether a member of the family is CEO or chairman, or both. Investor protection. Legal protection for investors forms an integral part of any system of corporate governance. Through regulations and laws, investors obtain certain rights or powers that protect them from insider expropriation. These rights or powers are very broad, encompassing rules for disclosure or accounting rules, dividend rights, voting rights, the right to participate in shareholder’s meetings, and the right to sue the management of the firm. These rules may come from various legal sources such as company, security, bankruptcy, takeover, or competition laws. Accounting standards and stock exchange regulations may also play an important role. La Porta et al. (1988, 2000) discuss the key rules of investor protection and document their prevalence in 49 countries around the world. They show that the prevalence of such protection differs strongly across countries and their respective legal systems. Strong differences exist between countries relying on a common law system (e.g., the US, UK, Canada, and Australia) and countries relying on a civil law system (e.g., Germany and France).47 Generally, investor protection seems to be stronger in common law countries than in civil law countries. For example, common law countries are more likely to have oppressed-minority mechanisms in place than civil law countries (94% vs. 29%48). How does the strength of investor protection affect the definition of family firm? When investor protection is strong, the family needs less voting power to control the firm than when investor protection is weak. Another argument is that, with strong investor protection, the family’s reputation is not needed as a signaling device to attract outside investors; the entrepreneur or her family is not forced to hold large ownership shares (La Porta et al., 2000). With weak investor protection, the situation is different: Outside investors may demand a commitment of the family towards the firm and force the family to hold large ownership shares. Both arguments suggest that the definition of family firm for the US should require a relatively low ownership stake.
47
48
Common law systems emphasize judiciary independence, respect for precedents, and limited codification of laws. The law is developed mainly through decision of courts (case law), rather than through legislative statutes. Civil law systems, in contrast, stress codification and are more centralized (Tirole, 2006, p. 54). This percentage refers to countries with a French civil law system. 52% of countries with a German civil law system have oppressed-minority mechanisms (La Porta et al., 2000).
2 Family firms
17
Position of the CEO. In the US, the position of the CEO is strong. The CEO often has direct control over the composition of the board. Although directors should be voted into office by shareholders and have a fiduciary duty to protect shareholder’s interests, the actual nominating process may differ in practice. As an example consider the case of Ross Johnson, the former CEO of RJR Nabisco who is widely considered to be the pre-eminent example of executive greed.49 Ross Johnson hand picked most of RJR Nabisco’s directors. In an empirical study, Shivdasani and Yermack (1999) show that when the CEO serves on the nominating committee or no nominating committee exists, firms appoint fewer independent outside directors and more directors with conflicts of interest.50 They also find that stock price reactions to independent outsider nominations are lower when the CEO is involved in the nominating process. Consistent with this finding, Anderson and Reeb (2004) show that those family firms in which family influence is balanced by independent directors have a higher market-to-book value than family firms with relatively few independent directors. Hence, in a US context, whether the CEO is a family member or not is important, and this should be reflected in the family firm definition. When the CEO belongs to the family, the family can exert a strong influence on the firm, even with a low ownership share. A similar argument can be made about the position of chairman of the board of directors. Insider- versus outsider-controlled system. The corporate governance literature differentiates between insider-controlled and outsider-controlled corporate governance systems (Franks and Mayer, 1994; Hackethal et al., 2005). The US corporate governance system is an example of an outsider-controlled system; the German corporate governance system is an example of an insidercontrolled system. In an insider-controlled system, control is exerted through internal mechanisms and institutions such as the supervisory board. Seats on the supervisory board are held by key stakeholder groups, including bankers, employee or union representatives, and blockholders.51 An insider control system functions on the basis of internal information; little information is made public. In an outsider-controlled system, control is exerted by a public takeover market (the market for corporate control).52 A company with poor performance either disappears or is taken over by an-
49
50 51
52
See the book Barbarians at the Gate: The fall of RJR Nabisco by Bryan Burrough and John Helyar (Burrough and Helyar, 2003). The book also appeared as a television movie. For further anecdotal evidence on this issue, refer to Mace (1971) or Lorsch and MacIver (1989). For more information about German supervisory boards, see Kaplan (1994) or Tirole (2006, p. 29). For the role of bankers in German supervisory boards, see Dittmann et al. (2009); for the role of unions in supervisory boards and codetermination, see Gorton and Schmid (2004). For more on the idea of a market for corporate control, see Holmström and Kaplan (2001) or Tirole (2006, pp. 43-51).
18
2 Family firms
other firm. Bankers and employee representatives do not engage in corporate control and are not members of controlling institutions. How is the definition of a family firm affected by the difference between insider-controlled and outsider-controlled systems of corporate governance? In an insider-controlled system, internal institutions such as the supervisory board are very important. To exercise control, the family as a shareholder needs to be represented in these institutions. In contrast, in an outsider-controlled system, the family can rely on public information and the market for corporate control to pursue its interests. Being a member of internal institutions is less important. To summarize, a family firm definition adapted to the US context should include the following criteria: a low family ownership threshold (e.g., 5% or 10%), whether or not a family member is CEO or chairman, and the percentage of family members on the board. Due to the one-tier board structure and the fact that the US corporate governance system is an outsider-controlled system, it makes little sense to include family presence in a supervisory board as a criterion (such as this is necessary in a European context). In this thesis, five family firm definitions are used, as summarized in Table 2-2. Table 2-2: Family firm definitions used in this thesis
broad
narrow
#
Definitions
Variable name
1
Combined stock ownership of the family > 5% or member of the family is either CEO or chairman
Family firm1
2
Family owns more stock than all institutional investors together do (minimum of 5%) or member of the family is either CEO or chairman
Family firm2
3
Combined stock ownership of the family > 5% and member of the family is either CEO or chairman
Family firm3
4
Family owns more stock than all institutional investors together do (minimum of 5%) and member of the family is either CEO or chairman
Family firm4
5a
Family owns more stock than all institutional investors together do (minimum of 5%) and member of the family is either CEO or chairman and the firm is younger than 51 years
Family firm5a
5b
Family owns more stock than all institutional investors together do (minimum of 5%) and member of the family is either CEO or chairman and the firm is older or equal than 51 years
Family firm5b
The family firm definitions present a wide spectrum. The definition family firm1 is a broad definition and considers a large number of firms to be family firms. To be classified as a family firm, the combined stock ownership of the family needs to be above 5% or a member of the family needs to be either CEO or chairman. The definition is similar to the definition used by Anderson
2 Family firms
19
and Reeb (2003) and BusinessWeek (2003). The definition family firm5a is at the other end of the spectrum. To be classified as a family firm, (1) the family should own at least 5% of the stock, (2) the combined ownership share of the family should be higher than the combined ownership share of institutional investors, (3) a member of the family should be either CEO or chairman, and (4) the firm should not exist for more than 50 years.53 As already described in Section 2.1.1, the concept of ‘family’ is part of any family firm definition. I define the family in a broad way and include the founders, the spouses of the founders, the descendents of the founders, and the spouses of the descendents of the founders, as well as adopted children. This differs from narrower family definitions, e.g., lone-founder firms are regarded as family firms. An actual or intended transfer of ownership is not included as a criterion in the definition, which is in line with the definition of IfM Bonn (2007) or Stiftung Familienunternehmen (2007).54 Furthermore, it is not required that the founder or the descendent of the founder is married or engaged in a relationship, i.e., that she has already started a family. To reveal the differences between young and old family firms, I created definitions 5a and 5b. Definition 5a refers to family firms younger than 50 years, whereas definition 5b refers to family firms older than 50 years. Despite the broad spectrum of family firm definitions used in this thesis, some shortcomings remain. Chief among these shortcomings is the fact that the presence of the family in the board of directors is only measured for the CEO and the chairman, because data on other members of the board of directors were not available. Another important concern involves the distinction between voting and non-voting stock. Due to data restrictions, I can only make a statement about the ownership share of the family (i.e., the share of common stock), but not the actual voting power. DeAngelo and DeAngelo (1985) and Villalonga and Amit (2008) show that family firms often use dual classes of stock to increase their voting power; this is another reason to use a low ownership threshold in a US family firm definition. The data collection efforts and the steps to apply the family firm definitions are described in Section 4.1.
2.2
Family influence Measures of family influence are an alternative way to overcome the family firm definition di-
lemma. Family business researchers generally agree that family involvement is what makes family 53
54
Note that Anderson and Reeb (2003) also used 50 years as a criterion to differ between young and old family firms. In this context, Villalonga and Amit (2008) differ between founder-controlled and family-controlled firms. For a discussion about generational transfer as a criterion in the family firm definition, see Miller et al. (2007) and Ward (1987).
20
2 Family firms
firms different from other firms (Chua et al., 1999). Most researchers interpret this involvement as either ownership or management (Handler, 1989). Recently, with the ‘familiness’ construct and the F-PEC scale of family influence55, more elaborate concepts of family involvement have been advanced in the literature. This section discusses these concepts and introduces the measures of family influence used in this thesis. Habbershon and Williams (1999) and Habbershon et al. (2003) use the resource-based view (Barney, 1991; Wernerfelt, 1984) and link the interactions between family, business, and individuals with the resources of the firm. They describe the bundle of resources resulting from family influence as ‘familiness’. Their basic idea is that familiness is a unique resource a “particular firm has because of the systems interaction between the family, its individual members, and the business” (Habbershon and Williams, 1999, p. 11). This resource is not reflected by other firm resources such as physical or human capital. They propose that scholars should analyze the competitive implications resulting from ‘familiness’, rather than assessing whether or not family firms have a competitive advantage. This approach offers two advantages: First, a (dichotomous) family firm definition is not needed; the family firm definition dilemma (Klein et al., 2005) is circumvented. Second, by using a continuous variable, it becomes possible to investigate the entire continuum of family firms. The collapse of an interval variable such as level of family ownership into a dichotomous variable (family versus non-family firm) can lead to a loss of important information. Astrachan et al. (2002) and Klein et al. (2005) follow this idea and propose a scale that assesses the extent and quality of family influence. They argue that there are three important dimensions of family influence: Power, Experience, and Culture (F-PEC). Each dimension is determined on a different scale. The power subscale is measured by the proportion of shares held by the family, the percentage of family members in top management, and the percentage of family members on the board. The experience subscale relates to the experience the family provides to the business, and is measured by how many generations have been in charge of management and ownership (the more generations, the more experience). The culture subscale is the overlap of family and business values. The more the values overlap, the stronger the family influence on the business. The culture subscale is measured by the family business commitment questionnaire (Carlock and Ward, 2001). In this thesis, family influence is determined in many ways. Most of these measures refer to the power subscale of the F-PEC. The experience subscale, which deals with the family generation in control, is proxied by the age of the firm. The culture subscale is not included. The reason is that the firms in my sample are among the world’s largest firms, which makes it difficult to obtain ques-
55
PEC refers to Power, Experience, and Culture.
2 Family firms
21
tionnaire data. Moreover, using the culture subscale as a measure of family influence might be problematic, since long-term orientation, the main subject of analysis in this thesis, is part of the culture of an organization (Denison and Mishra, 1995; Denison et al., 2004; Kotter and Heskett, 1992; Marginson and McAulay, 2008). The concrete measures used in this thesis are summarized in Table 2-3 below. Table 2-3: Dimensions of family influence used in this thesis Dimension
Description
Management CEO is member of the family. Chairman of the board of directors is member of the family. CEO or chairman are members of the family. CEO and chairman are members of the family. Ownership Percentage of common stock owned by the family Family owns more stock than financial investors. Family owns more than 5% (20%, 30%) of common stock. Experience Age of the family firm
The data collection efforts, and the concrete steps to collect the family firm measurements, are described in Section 4.1.
2.3
Theories about family firms In this section, I describe the agency and the stewardship theory, which are the two most
widely used theoretical frameworks in research on family businesses. 2.3.1
Agency theory Agency theory is used in this thesis in two ways. First, it is used in the empirical sections to
motivate hypotheses about differences between family and non-family firms or within the group of family firms (see Chapters 5 to 8). Second, agency theory is used to calculate the optimal compensation contract of a non-family manager working in a family firm (see Chapter 9). This section proceeds as follows. After a short introduction to agency theory, four different types of agency costs relevant in the context of family firms are described: (1) agency costs from the separation of ownership and management, (2) agency costs from altruism, (3) agency costs from conflicts between owners and lenders, and (4) agency costs from conflicts between dominant and minority shareholders. The section concludes with a comparison of agency costs in family and non-family firms.
22
2 Family firms
Introduction Generally, agency theory is concerned with resolving problems that arise in agency relationships.56 An agency relationship is described as a situation in which one party (the principal) delegates work to another party (the agent). This principal-agent relationship exists between employers and employees, lawyers and clients, or buyers and suppliers. Agency theory attempts to solve two problems. The first one is the agency problem that arises when the goals of the principal and the agent diverge, and it is difficult for the principal to verify the agent’s actions. The second one is the problem of risk sharing which arises when the principal and the agent have different attitudes towards risk. Agency theory has developed along two lines: a positivist stream and a principal-agent stream (Jensen, 1983). Both streams share a common unit of analysis (the contract between the principal and the agent) and common assumptions about people, organizations, and information (bounded rationality, self-interest, etc.). The two streams, however, differ in their mathematical rigor and their dependent variables. The principal-agent focuses on quantitative economic models, whereas the positivist stream is more descriptive. Another difference is that the principal-agent stream is concerned with the general theory of the principal-agent relationship, whereas the positivist stream focuses almost exclusively on the relationship between the CEO and the owner(s) of a firm. This thesis uses the principal-agent stream, in particular the model from Holmstrom and Milgrom (1991)57, to derive the optimal compensation contract of a non-family manager working in a family firm. As there is little connection to the empirical sections, I summarize the relevant literature needed to develop the model in the model section itself (Chapter 9). The review in this section focuses exclusively on the positivist stream. The positivist literature has focused on agency problems that arise between owners and managers of large firms (Berle and Means, 1932). Most studies in this field deal with governance mechanisms needed to solve the agency problem. Jensen and Meckling (1976) discuss how equity ownership by managers aligns the interests of managers and owners. Fama (1980) discusses how efficient capital and labor markets can control the self-serving behavior of management. Fama and Jensen (1983) refer to the board of directors (in particular the non-inside directors) as an information gathering institution that owners of large firms can use to monitor the management.
56
57
For a detailed review of agency theory and its link to organization and management research, see Eisenhardt (1989). The model describes the optimal incentive structure in a multi-task environment.
2 Family firms
23
The next section describes in detail the model by Jensen and Meckling (1976), a good starting point to motivate differences between family and non-family firms. Most publications in the family business field refer to this model as a starting-point from which more specific models are derived. Agency costs from the separation of ownership and management58 Consider an owner-manager59 who faces a decision between maximizing firm value, V, and consuming private benefits, F (such as office space or a corporate jet) (Figure 2-1). The firm value is at an optimum when its marginal benefits from consumption of private benefits are equal to its marginal costs, point V*. If the consumption level increases beyond this optimum, the firm value is reduced, line V*F.60 V* may not be the best solution for the manager. The manager’s taste for firm value and private benefits is represented by her utility function, giving indifference curves such as U1. The owner-manager’s objective is to maximize her utility (described by U1) subject to a budget constraint with a slope of -1 (described by line V*F). Her optimal level of private benefit consumption and the resultant firm value are determined by the point where the budget constraint is tangential to the indifference curve, point A. In this case, the level of consumption of private benefits will be F1, and the resulting firm value will be V1.
58
59
60
This section is partially based on Block and Jaskiewicz (2007). The paper was awarded the 2nd place at WHU’s Best Paper Award on Family Enterprises (2007). An owner-manager owns 100% of the equity of the firm she manages. The firm value is part of her personal wealth. If the owner-manager consumes more private benefits than what is optimal from a firm’s perspective, costs arise. Jensen and Meckling (1976) define F as the cost of providing these extra-benefits.
24
2 Family firms
Figure 2-1: Trade-off between firm value and private benefits for an owner-manager
V*
owner-manager‘s tastes for wealth and private benefits A
V1
firm value
U1
owner-manager‘s budget constraint
slope = -1
F1
F
market value of the stream of manager‘s expenditures on private benefits
2 Family firms
25
Assume now that the owner-manager sells a fraction 1- of her equity at the valuation V1 of the firm, i.e. with the implicit assumption that she will stick to F1 (Figure 2-2). If the equity market believes the owner-manager to consume the same level of private benefits as she did when she was sole owner, the ownership stake of 1- is worth (1-)V1. This partial sale however leads to a new situation for the manager. She no longer bears the full cost of her decision to consume private benefits. The budget constraint rotates outward61, ceteris paribus, resulting in a new optimum level of consumption, F2 (point B). Figure 2-2: Owner-manager sells a fraction of her equity
V*
U1 A
V1
U2
B
firm value
V2
slope = -
slope = -1
F1
F2
F
market value of the stream of manager‘s expenditures on private benefits
61
The slope of the budget constraint is now -.
26
2 Family firms
However, if the equity market is characterized by rational expectations, this scenario is unlikely to hold (Figure 2-3). The market anticipates the manager’s new situation, in particular her new level of consumption of private benefits, and pay her only (1-)V3 for a fraction 1- of her equity (point C). The value of the firm has dropped from V1 to V3.62 Figure 2-3: Rational expectations of the equity market
V* V1
U1 A
B
V2 firm value
U2
C
V3
U3
slope = -
slope = -1
F1
F2, 3
F
market value of the stream of manager‘s expenditures on private benefits
62
Note that for the equity market the budget constraint with slope=-1 applies.
2 Family firms
27
Yet, this situation is suboptimal from the manager’s perspective. By reducing her consumption of private benefits, she could sell her 1- share of equity for a higher price, and thereby reach a higher level of utility (Figure 2-4). F4 is the only level of consumption where both the equity market and the manager are satisfied. Graphically, this corresponds to the situation where the manager’s indifference curve is tangent to her budget constraint with slope -, but also lies on line V*F (point D). Line V*F corresponds to the budget constraint that applies to the equity market.
agency costs of equity
Figure 2-4: Agency costs of equity
V* V1 V2 V4
D
B
U4 C
V3 firm value
U3
slope = - slope = - slope = -1
F1
F4
F2, 3
F
market value of the stream of manager‘s expenditures on private benefits
The value of the firm is now V4, and the agency costs of equity are described by V1-V4. The decline in firm value is caused entirely by the executive’s behavior.
28
2 Family firms
Next, Jensen and Meckling (1976) point out the role of monitoring (e.g., through an auditing firm or a supervisory board) or bonding activities (e.g., incentive compensation) to influence the executive’s behavior. These actions are aimed at reducing the agency costs of equity. Figure 2-5 shows the situation when a monitoring mechanism is installed.
agency costs of equity
Figure 2-5: Role of monitoring activities
V1
G
V5 V4
F E
monitoring costs slope = -
firm value
D
slope = -1
F1
F5 F4
F
market value of the stream of manager‘s expenditures on private benefits
The curve DEG depicts a budget constraint given the possibility of monitoring activities. The points on curve DEG describe the firm value for varying levels of monitoring and at a given level of , namely V=V*-F(M,)-M. The shape of the curve is inverse U-shaped because it is assumed that monitoring activities (M) reduce the level of private benefits (F) at a decreasing rate, i.e., F/M<0 and ²F/²M>0. At point D, no monitoring activities are undertaken (the result from Figure 2-4), whereas at point E the optimum level of monitoring expenditures is reached, as determined by V/M=0 and ²V/²M<0. The corresponding firm value is now V5, monitoring costs are F-E, and the agency costs of equity are V1-V5. Monitoring activities have led to an increase in firm value or to a reduction in the agency costs of equity. Jensen and Meckling (1976) show that both the manager and the outside owners are now better off as compared to a situation without monitoring activities.
2 Family firms
29
What does the Jensen and Meckling (1976) model imply for family firms? The model postulates that owner-managed firms have zero agency costs (see also Ang et al., 2000; Fama and Jensen, 1983). The firm’s benefit equals the owner’s benefit. In contrast to a hired manager, the owner-manager has no incentive to consume more private benefits than what is necessary for the firm’s benefit. The firm’s benefit equals the owner’s benefit. The idea of zero or little agency costs has been extended to family firms because kinship ties are expected to result in altruistic behavior among family members (Chrisman et al., 2004). Borrowing arguments from anthropology63, Stewart (2003) argues that in most cultures, kinship ties result in a moral obligation to help one another. Furthermore, Becker’s famous rotten-kid theorem (1974, 1976) states that altruism by one family member will lead other selfish family members to act efficiently from the perspective of the family. More concretely, in the presence of parental transfers, the child will choose actions that maximize the family’s income, despite being guided by her own self-interest.64 If the arguments by Stewart (2003) and Becker (1974, 1976) are true, family firms can be treated as a special case of an owner-managed firm, in which little or no agency costs prevail. This leads to the argument that family firms will see better financial performance than non-family firms (see also Anderson and Reeb, 2003; Daily and Dollinger, 1992). Regarding the above discussed family firm definition problem and the broad spectrum of family firms that exist (see Section 2.1), a number of other applications of the Jensen and Meckling (1976) model can be considered:
Family versus non-family manager. According to the Jensen and Meckling (1976) model, family-owned firms managed by a family member should be different from those managed by a non-family member. In the latter case, the argument that the CEO acts altruistically towards members of her own family does not apply. As a result, agency costs resulting from over-consumption of private benefits can arise. In this aspect, family-owned firms managed by a non-family member resemble ‘pure’ non-family firms; monitoring or bonding mechanisms need to be in place.
Single-owner versus multiple-owner family firm. The model of Jensen and Meckling (1976) posits a difference between a single- and a multiple-owner family firm. The more owners there are, the lower the equity share of the manager. Since the manager does not bear the full costs of her decision to consume private benefits, she has an incentive to con-
63
64
Anthropologists study human diversity around the world by looking at e.g. cross-cultural differences in social institutions, cultural beliefs, and communication styles. See http://www.brockhausenzyklopaedie.de/be21_article.php (accessed September 12th, 2008). The theorem is discussed in more detail in the next section.
30
2 Family firms
sume more private benefits than what is optimal from the perspective of the firm. An agency problem is present; bonding or monitoring mechanisms should be in place. Combining the model of Jensen and Meckling (1976) with the literature about moral hazard in teams (Alchian and Demsetz, 1972; Holmstrom, 1982a) raises another problem. Multipleowner family firms (perhaps consisting of several clans) might run into a team production problem leading to a situation of free-riding. A family member that engages in monitoring has to share the benefits of monitoring with the other family members and therefore may decide not to engage in monitoring (a public good problem or social dilemma situation). If this is the case, the multiple-owner family firm actually resembles a non-family firm with a widely dispersed ownership structure (Fama, 1980; Maug, 1998).
Listed versus non-listed family firm. Jensen and Meckling (1976) gauge the agency costs of separating management and ownership by the value reduction of the firm in the equity market. Of course, this measurement cannot be applied to a non-listed family firm. In fact, owners of private family firms might not recognize the value reduction from opportunistic management behavior and are thus less likely to install monitoring or bonding mechanisms than listed family firms.
As described above, the mechanisms for curbing agency costs – monitoring and bonding – should differ across the various types of family firms. Ceteris paribus, those family firms with a greater agency problem (family firms with multiple owners and family firms managed by a non-family member) should have more bonding and monitoring mechanisms in place than other family firms. This relates directly to CEO compensation, which is analyzed in Chapters 8 and 9.
2 Family firms
31
Agency costs from altruism This section introduces the concept of altruism in the family and discusses how it affects the behavior of individual family members. Then, I extend the concept of altruism to family firms and discuss its impact on the agency costs of a family firm. Theory of altruism and the rotten-kid theorem Generally, economists regard selfishness as the main driver in a market economy. Yet, at the same time, altruism is regarded as important in a family context. Adam Smith (1853, cited after Becker, 1981) noted “every man feels his own pleasures and his own pains more sensibly than those of other people. […] After himself, the members of his own family, those who usually live in the same house with him, his parents, his children, his brothers and sisters, are naturally the object upon whose happiness or misery his conduct must have the greatest influence.”65 Becker (1976) describes an altruist as a person willing to reduce her own consumption to increase the consumption of others. In a family context, this means that the utility of an altruistic family member depends positively on the utility of other family members. Becker’s rotten-kid theorem (Becker, 1974, 1976) suggests that selfish family members will act to help one another if their incentives are properly linked. To show that this is true, Becker creates a hypothetical situation in which two children receive gifts of money from an altruistic parent in order to make them happy (Becker, 1981). One of the two kids is an egoistic, a ‘rotten’ kid who takes pleasure in harming her sibling. Consider the situation of the rotten kid harming her sibling, thereby increasing her utility by 1 and decreasing the utility of her sister by 1.5. The overall utility (or happiness) of the two children is reduced by 0.5. The altruistic parent, who gains utility from the well-being of her children, feels less pleasure from giving to her children.66 She will reduce her gifts to her children and instead increase her own consumption. In such a situation, the rotten kid has an incentive to avoid hurting her sibling. Although she cares only about her own consumption, she considers the effects of her behavior on her sibling. She might in fact behave in such a way as to increase the happiness of her sibling. Thus, without using any formal incentives, the altruistic parents can induce the rotten kid to behave benevolently by making her welfare contingent upon the welfare of her sibling.
65 66
See also the discussion of these (and other) statements from Smith by Coase (1976). The altruistic parent does not know that the reduction in the children’s overall utility is due to the action of the rotten kid.
32
2 Family firms
This family-inherent altruism leads to the argument that family members follow the same utility function, so that a family firm can be treated as a special case of an owner-manager firm (as described in the model of Jensen and Meckling, 1976, above). Yet, although altruism mitigates the agency costs related to the separation of ownership and management, it could also lead to new types of agency costs from free riding by family members, as in the Samaritan’s dilemma (Bruce and Waldman, 1990; Buchanan, 1975). Free riding by family members and the Samaritan’s dilemma Bruce and Waldman (1990) have proposed an extension of Becker’s one-period model (Becker 1974, 1976) to two periods and show that parental altruism may lead to an inefficiency described in other contexts as the Samaritan’s dilemma. Generally, the Samaritan’s dilemma refers to a situation in which an individual receiving transfers (e.g. from government) has an incentive to behave in a manner that increases her probability of receiving transfers.67 In the case of a parentchild relationship, this means that the child has an incentive to overstate her own needs and remain dependent upon her parents. In a two-period setting, a child has an incentive to save too little and consume too much in the first period relative to the efficient level from the family’s perspective. She has an incentive to behave in such a way because placing herself in a more impoverished condition leads to increased transfers from her parents in the second period. More broadly, the Samaritan’s dilemma belongs to a class of agency problems associated with the lack of self-control by the principal (Schulze et al., 2003). Such problems arise whenever a party to a contract has both an incentive and the ability to harm itself and others (Jensen, 1994; Thaler and Shefrin, 1981). Thaler and Shefrin (1981) propose two techniques to deal with self-control problems. The first technique is to alter incentives. Parents might offer their children contingent contracts (Lindbeck and Weibull, 1988; O’Donoghue and Rabin, 2000) such as paying the child’s college education only if the child does well in high school (Schulze et al., 2003). The second method is to introduce rules that eliminate discretion over decisions for which the self-control problem is particularly acute. For example, parents may prohibit their children from spending money on expensive consumer goods, or they may persuade their children to prepare and implement a savings plan. Schulze et al. (2001, 2003) propose a theory of agency and altruism in family firms. They argue that agency problems with altruism and self-control are exacerbated when a CEO owns a significant share of the company. The CEO has a greater capacity to make altruistic transfers that benefit the members of the owning family. Other family members have an incentive to overstate
67
For a further discussion of the Samaritan’s dilemma, see Buchanan (1975) or Kotlikoff (1987).
2 Family firms
33
their needs in order to receive higher transfers, which may spark a Samaritan’s dilemma as described above. Schulze et al. (2001, 2003) also argue that altruism reduces the CEO’s ability to effectively monitor and discipline the members of her family, as the CEO’s perceptions are biased. Furthermore, the CEO’s ability to monitor and discipline other family members might be compromised due to the negative effects this would have on familial relationships. Schulze et al. (2001, 2003) argue that family-managed firms use CEO pay incentives to solve this self-control problem. Their data show that this practice is beneficial for the firm’s performance and find that this effect is moderated by the parents’ estate and share transfer intentions (i.e., the degree of parental altruism).68 Agency costs from conflicts between owners and lenders Conflicts between owners and lenders can lead to agency costs of debt. The following simplified example illustrates the problem: Consider a firm that is financed by 50% debt and 50% equity. The interest rate on the debt is 5%. The owner-manager of the firm can either pursue project A or project B. Project A gives a return of 10% on the total capital with a probability of 80% and a return of zero otherwise; project B generates a return of 20% on the total capital with a probability of 50% and a return of zero otherwise. If the owner-manager pursues project A, her expected return is 8% (0.8·10%); if the owner-manager pursues project B, her expected return is 10% (0.5·20%). Accordingly, the owner-manager will pursue project B. Yet, this way, she has transferred wealth from the lender to herself, a process that is called asset substitution. The lender would be better off if the manager had pursued project A. With project A, the expected return of the lender is 4% (0.8·5%), whereas project B carries an expected return of only 2.5% (0.5·5%).69 Lenders anticipate such incentives and compensate this by demanding higher rents, resulting in higher costs of debt. Alternatively, they may limit the amount of debt which they provide (credit rationing).70 Anderson et al. (2003) argue that the presence of family owners mitigates these agency costs of debt. Family owners have a desire to pass the firm to the next generation and have strong concerns over family and firm reputation. They put a high value on firm survival and do not aim at short-term wealth maximization. They are less likely to invest in risky projects. Their interests are more closely aligned with the interests of capital lenders than is the case with atomistic sharehold68
69
70
See Lubatkin et al. (2005, 2007a, 2007b) and Schulze et al. (2002) for further articles on the relationship between altruism, self-control, and agency costs in family firms. For this calculation, it is assumed that if the project has a return of zero, the firm cannot pay her interest; the principal is paid back. If the principal cannot paid back, the expected return of the lender decreases further. For moral-hazard models explaining agency costs of debt, see Bester and Hellwig (1987), Biais and Casamatta (1999), Stiglitz and Weiss (1981), or Tirole (2006, pp. 115-116).
34
2 Family firms
ers. They are more likely to engage in relationship banking71 – a form of debt finance that emphasizes a long-term relationship between borrower and lender. All these points imply lower agency costs of debt, and, ceteris paribus, should lead to lower overall costs of debt. Agency costs from conflicts between dominant and minority shareholders A number of authors suggest that family firms may suffer from agency costs due to a conflict of interest between dominant and minority shareholders (e.g., Claessens et al., 2000, 2002; GomezMejia et al., 2001; Morck and Yeung, 2003). They argue that family management may lead to problems of managerial entrenchment.72 Managerial entrenchment refers to a situation where a manager has so much power that she is able to use the firm to further her own interests rather than the interests of shareholders (Weisbach, 1988). In particular, it is argued that family managers act solely for one shareholder – the family – and neglect the interests of minority shareholders. As an illustration, consider a family firm organized in a pyramidal structure (see Figure 2-6 below).73 In such a pyramidal structure, minority shareholders can be expropriated in (at least) two ways: First, firms which are low in the pyramidal structure have the same agency problems from the separation of ownership and management as do firms with a widely dispersed shareholder structure. In the example in Figure 2-6, the family controls 51 percent of the shares of firm C. However, the family’s stake of profits is only about 13% (51%·51%·51%). The family manager of firm C thus has an incentive to consume more private benefits than is optimal from firm C’s perspective (the same problem as described in the Jensen and Meckling (1976) model above).74 The family, as a controlling shareholder, is unlikely to intervene because it bears only 13 percent of the costs. Moreover, the controlling family may decide to become part of the game and force the family manager to share her private benefits with the remainder of the family. Ultimately, it is the minority shareholders who pay the bill.
71
72 73
74
Relationship banking is defined as the “provision of financial services by a financial intermediary that invests in obtaining customer-specific information, often proprietary in nature; and evaluates the profitability of these investments through multiple interactions with the same customer over time and/or across products” (Boot, 2000). Relationship banking is argued to decrease the agency costs of debt related to moral hazard and information asymmetries (Boot; 2000; Boot and Thakor, 2000; Tirole, 2006, pp. 369-379). The literature on managerial entrenchment goes back to Morck et al. (1988) and Stulz (1988). Note that the corporate tax system in the US, in particular the double taxation system, discourages pyramidal holding structures (for a discussion, see Morck, 2003). Villalonga and Amit (2008) can show that in the US, pyramidal structures are not widely used. Yet, founding families have found other ways such as dual stock or disproportional board representation to have control rights that exceed cash-flow rights. There is a large finance literature on private benefits of control, e.g., Dyck and Zingales (2004), Ehrhardt and Nowak (2003b), and La Porta et al. (2000). See also Section 3.3.6.
2 Family firms
35
Second, the family may use tunneling or self-dealing to expropriate minority shareholders (Johnson et al., 2000). Firms controlled by the same family often deal with each other, e.g., firm B obtains goods from firm C. By charging artificially high prices to firms which are low in the pyramidal structure, the family can transfer wealth up the pyramid (e.g., to firm A). The family has a strong incentive to do so because its profit share is larger in firms high in the pyramidal structure. Again, it is the minority shareholders of firms low in the pyramidal structure paying the bill (e.g., firm C). Figure 2-6: Family business group organized in a pyramidal structure
Family
owns 51%
Firm A
owns 51%
Firm B
owns 51%
Firm C own 49%
own 49%
Public shareholders
Source: Morck and Yeung (2003)
own 49%
36
2 Family firms
Agency costs in family and non-family firms: a summary Following Chrisman et al. (2004), I integrate the several types of agency costs discussed above into an additive model. The resulting inequality allows a direct comparison of the agency costs in family (subscript Fam) and non-family firms (subscript Non-fam): ALT Fam + OM Fam + OL Fam + DM Fam ALT Non-fam + OM Non-fam + OL Non-fam + DM Non-fam
(2-1)
Where: ALT =
Agency costs from ALTruism
OM =
Agency costs from separation of Ownership and Management
OL =
Agency costs arising from conflicts between Owners and Lenders
DM =
Agency costs arising from conflicts between Dominant and Minority shareholders
Family firms have higher agency costs if the sum on the left side of the inequality is greater than the sum on the right side of the inequality, and vice versa. This inequality illustrates that it is difficult to make a general statement about whether or not family firms have higher agency costs than non-family firms. Not all family firms have unity of ownership and management. The proportion of debt and the equity share of the family may also vary strongly within the group of family firms. Finally, it is difficult to quantify the degree to which the family firm suffers from agency costs due to altruism or self-control. After having reviewed agency theory and its extension for family business research, I now describe stewardship theory, which has become a prominent alternative for explaining differences between family and non-family firms or different types of family firms. 2.3.2 Stewardship theory Recently, agency theory in general, and the Jensen and Meckling (1976) model in particular, have been strongly criticized by management scholars (e.g., Lubatkin et al., 2007a, b; Lubatkin, 2007; Zahra, 2007). The criticism mainly refers to the reductionist approach of the model. It is noted that, to enhance the model’s precision and its scholarly appeal, a number of simplifying assumptions have been made, thereby reducing the model’s relevance. It has been argued that the model is incompatible with management theory and is unable to explain the complexities of realworld organizations or phenomena.75 The two main criticisms are:
75
In fact, the original Jensen and Meckling (1976) model was not intended to be applied to settings beyond largefor-profit organizations that operate in developed markets with widely-diversified shareholders (Lubatkin, 2007; Ulhoi, 2007).
2 Family firms
37
Narrow context. The narrow context of the Jensen and Meckling (1976) model has been criticized. The model describes the situation of public large for-profit firms that operate in developed markets with widely diversified shareholders (e.g., pension funds) (Gomez-Mejia and Wiseman, 2007; Lubatkin, 2007). Yet, as Lubatkin (2007) points out, this context is more the exception than the rule.76 Model assumptions. The second main criticism refers to the assumptions made regarding the model of man that underlies agency theory. The principal-agent model holds that managers are opportunistic and want to enrich themselves at the expense of their shareholders. Critics argue that this assumption does not explain the complexity of human action (e.g., Doucouliagos, 1994). For example, assuming managerial opportunism, agency theory states that managers do not identify with the organization which they serve. Yet, contrary to this managerial opportunism assumption, research from psychology and organizational behavior shows that some managers do indeed identify with the organization and attribute organizational success to themselves (Salancik and Meindl, 1984; Staw et al., 1983). Stewardship theory has been proposed as an alternative theory of corporate governance (Davis et al., 1997; Donaldson and Davis, 1991). It is less reductionist and relies on less-simplified assumptions than agency theory. Moreover, it includes concepts and theories from other disciplines, in particular sociology and psychology. However, as I explain below, I do not see stewardship and agency theory as two mutually exclusive theories or concepts. Rather, I see stewardship theory more as an extension or complement of agency theory. Stewardship theory illustrates the limitations of agency theory and integrates the views of neighboring disciplines such as psychology and sociology. However, whether stewardship theory does a better job at explaining real-world phenomena remains debatable.
76
Note that owner-managed firms, non-listed firms and firms in less developed countries (with less developed financial markets) are not included.
38
2 Family firms
Description of the theory Stewardship theory is rooted in psychology and sociology and describes situations where executives acting as stewards are motivated to behave in the best interests of a firm’s owners (Davis et al., 1997; Donaldson and Davis, 1991). The model of man that underlies stewardship theory is fundamentally different from the model of man that underlies agency theory. Stewardship theory assumes that pro-organizational, collectivist behavior gives a higher utility than individualistic, selfserving behavior. Table 2-4 below summarizes the main differences between agency and stewardship theory. Table 2-4: Differences between agency and stewardship theory Agency theory
Stewardship theory
Model of man
Economic man
Self-actualizing man
Behavior
Self-serving
Collective serving
Lower order/ economic needs (e.g., physiological security)
Higher order needs (e.g., growth, achievement, self-actualization)
Extrinsic
Intrinsic
Other managers Low value commitment Institutional (legitimate, coercive, reward)
Principal High value commitment Personal (expert, referent)
Control oriented Control mechanisms Short term Cost control
Involvement oriented Trust Long term Performance enhancement
Individualism High power distance
Collectivism Low power distance
Psychological Mechanisms Motivation
Social comparison Identification Power
Situational Mechanisms Management philosophy Risk orientation Time frame Objective Cultural differences
Source: Davis et al. (1997)
Davis et al. (1997) suggest a number of factors that lead individuals to act as proorganizational, collectivist stewards rather than as individualistic, self-serving agents. They differentiate between psychological and situational factors. The psychological factors that lead an individual to behave as a steward rather than as an agent are related to motivation, identification, and use of power. Here’s how each of these factors impacts stewardship behavior:
2 Family firms
39
Motivation. Stewardship theory proposes that individuals motivated by higher-order needs are more likely to behave as stewards than individuals who are motivated by lowerorder needs. This proposition refers directly to Maslow’s hierarchy of needs (Maslow, 1970). In particular, stewardship theorists claim that the needs for self-actualization and growth (Alderfer, 1972) as well as for achievement and affiliation (McClelland, 1975; McGregor, 1966) lead individuals to behave as stewards. In addition to that, people who are motivated by intrinsic factors are more likely to behave as stewards than people who are motivated by extrinsic factors.77 This way, stewardship behavior is linked to job characteristics and job design.
Identification. Stewardship theorists propose that people with high identification and high value commitment are more likely to become stewards than other people. Identification means that individuals define themselves in terms of organizational membership by accepting the organization’s mission and goals (Mael and Ashforth, 1992). Those individuals often commit themselves to the goals of the organization (Porter et al., 1974). Individuals who identify with the organization and commit themselves to the goals of the organization interpret comments about the organization as referring to themselves. As a result, the utility function of the individual and the organization converge.
Use of power. Stewardship theory proposes that individuals who are more likely to use personal power to influence others are more likely to behave as stewards than individuals who use institutional power. Institutional power is defined by the individual’s position in the organization, whereas personal power is an inherent part of the individual and is not affected by position (Gibson et al., 1991). The situational factors which influence the probability of acting as a steward rather than as an
agent are management philosophy, collectivism of culture, and power distance. Each of these factors is explained in detail below.
Management philosophy. Advocates of stewardship theory argue that management philosophy has an impact on whether a stewardship or an agency relationship emerges. In particular, it is argued that a control-oriented philosophy leads to an agency relationship, whereas an involvement-oriented philosophy leads to a stewardship relationship. In a control-oriented approach, the thinking and controlling part of work is separated from the do-
77
For theories of intrinsic motivation, refer to Hackman and Oldham (1976) or Manz (1986).
40
2 Family firms
ing; in an involvement-oriented approach, self-control and self-management are emphasized (Lawler, 1986).
Collectivism of culture. In his works on cultural differences, Hofstede (1991) differentiates between individualism and collectivism. Individualism means that personal goals are emphasized over group goals. Collectivism means the opposite. Stewardship theorists argue that collectivist cultures are more likely to develop stewardship relationships, whereas individualistic cultures lead to agency relationships.
Power distance. A second dimension developed by Hofstede (1991) is the concept of power distance. It is defined as the extent to which less powerful members of an organization accept an unequal power distribution. Stewardship theory argues that individuals in a low power distance culture are more likely to behave as stewards than people in a high power distance culture. To conclude the description of stewardship theory, note that a stewardship relationship is not
necessarily stable and may change over time. For example, situational factors such as management philosophy or collectivism may change after a merger or other M&A activities (Corbetta and Salvato, 2004). I do not believe that either stewardship theory or agency theory can be accepted as definitively superior to the other. In my opinion, there are situations or contexts in which stewardship theory is more appropriate than agency theory, and vice versa. Stewardship theory and its relation to family firms In family firms, the owning family can influence the psychological and situational antecedents of organizational behavior. More concretely, the family has a crucial impact on the ‘model of man’ that prevails in their firm, that is, either the self-serving, economically rational man postulated by agency theory or the self-actualizing, collective serving man suggested by stewardship theory (see Table 2-4 above) of any other model. Corbetta and Salvato (2004) suggest four ways in which the family shapes the psychological and situational factors that determine whether an agency or a stewardship culture prevails: (1) non-financial goals, (2) degree of psychological ownership, (3) degree of trust, and (4) relational contracts.
Non-financial goals. It is generally accepted that the owning family pursues both financial and non-financial goals (Sharma et al., 1997; Tagiuri and Davis, 1992, 1996). The more important financial goals are, the more likely is the motivation of family members based on lower order needs and extrinsic motivation. This favors agency relationships. In turn,
2 Family firms
41
the more non-financial goals prevail, the more likely a stewardship relationship will prevail.
Psychological ownership. Psychological ownership is a state of mind in which an individual feels as though the target of ownership (whether material or immaterial) is part of their identity. Individuals have the feeling of possessiveness and of being psychologically tied to an object (Pierce et al., 2001). Family members who are psychologically linked to their firm are likely to identify with their firm and are likely to engage in high value commitment towards their firm, thus favoring the emergence of stewardship relationships.
Degree of trust. Family firms are often described as organizations in which trust plays an important role (Jones, 1983). When trust is based on affections or emotions, family firms are governed by an involvement-oriented management philosophy, i.e. by informal agreements based on a high level of faith. If trust is of a ‘calculative’ kind, a controloriented management is likely to evolve. An involvement-oriented approach favors a stewardship environment; a control-oriented approach favors agency relationships.
Relational contracts. Relational contracts are “informal relationships sustained by the value of future relationships” (Baker et al., 2002, p. 39). The more family bonds are embedded in such relational contracts, the more collectivist is the firm’s culture, which then leads to a stewardship environment. To summarize, there exist specific characteristics of family firms that can lead to a steward-
ship environment (e.g., importance of non-financial goals). Yet, it should be noted that there exist also specific characteristics of family firms that can lead to an agency environment. As an example consider family firms with sibling rivalries or family firms with conflicts between two family generations (Eddleston and Kellermanns, 2007; Harvey and Evans, 1994; Hennerkes, 2004, pp. 58-66; Levinson, 1971).
3 Long-term orientation of firms This chapter is about long-term orientation of firms. First, I discuss key terms related to longterm orientation and describe my understanding of the concept (Section 3.1). Next, I describe the various dimensions of long-term orientation (Section 3.2). Then, I introduce the central explanations for long-term orientation and discuss whether and under which conditions family firms should be more long-term oriented than non-family firms (Section 3.3). Building upon this discussion, Section 3.4 presents the specific research questions that will be analyzed in the following chapters.
3.1
Key terms
3.1.1
Intertemporal choice The idea of long-term orientation is highly intertwined with the concept of intertemporal
choice, an extensively studied notion in both economics and psychology.78 Generally, problems of intertemporal choice occur when costs and benefits of a particular decision fall into different time periods. Top-level management decisions, such as technology investments, development of a new product, or entrance into a new market, often involve intertemporal choice. According to Laverty (1996), an intertemporal-choice problem in management decisions exists because occasionally “the course of action that is best in the short term is not the same course of action that is best over the long run” (Laverty, 1996, p. 828). Generally, problems of intertemporal choice involve a determination of the proper balance between the long and the short term. In a normative economic approach, this trade-off is solved by discounting future cash flows (e.g., Fisher, 1930; von Neumann and Morgenstern, 1953). In this way, the problem is reduced to determining the discount rate that should be applied. Ceteris paribus, the use of a lower discount rate leads to a more long-term behavior and vice versa. The situation becomes more complicated when risk and uncertainty are included in the decision-making process. Risk exists when experience from the past allows attaching a probability to each possible outcome of an uncertain event, whereas uncertainty exists when no probability can be attached (Knight, 1921). Von Neumann and Morgenstern (1953) introduced the notion of expected utility to analyze decisions under risk. The procedure is as follows: (1) A decision is divided into its alternatives, its possible outcomes (in monetary units) and the probabilities associated with 78
For a summary, see Loewenstein and Thaler (1989).
3 Long-term orientation of firms
43
these outcomes. (2) Next, a risk utility function of the decision-maker is determined or assumed.79 (3) Using this risk utility function, each possible outcome is transformed into a utility value. (4) The utility values are then weighted with its respective probabilities of occurrence, leading to the expected utility. (5) Finally, the expected utility of each decision alternative is determined; an alternative with a higher expected utility is preferred to an alternative with a lower expected utility. Via the risk utility function, the expected utility can be transformed back into monetary terms, which gives the certainty equivalent. Generally, risk-averse decision makers have a certainty equivalent that is less than the weighted outcome; risk-seeking decision makers have a certainty equivalent that is greater than the weighted outcome.80 In real-world decisions however, managers rarely know the exact probabilities of each outcome of her course of action. Usually, decisions are taken under uncertainty. Please refer to Schoemaker (1982) for a summary of research on the expected utility model. So how do long-term orientation and attitudes towards uncertainty interact? A cash flow in the future should be discounted for both time and uncertainty (if the cash flow is uncertain). Similarly, Prelec and Loewenstein (1991, p. 784, cited by Laverty, 1996) argue that “time and uncertainty are typically correlated with one another in the real world” and that “anything that is delayed is almost by definition uncertain.” The concepts of long-term orientation and uncertainty attitude are almost inseparable in real-world managerial decisions.81 For example, a manager who decides to invest resources in an R&D project can be described as both long-term oriented and embracing uncertainty. 3.1.2
Short-termism, myopia, and long-term orientation The three terms ‘short-termism’, ‘myopia’, and ‘long-term orientation’ are often used in an
imprecise way. All three deal with the issue of balancing the needs of the long and the short term. The differences between these three terms become clear when considering the issue of optimality. Prior research has referred to short-termism as a suboptimal behavior focusing on the short-term, to the detriment of the long-term (e.g., Laverty, 1996; Marginson and McAulay, 2008). However, there are also situations in which a manager’s focus on the short-term extrapolates to an optimal long-term outcome. Such behavior is not classified as short-termism. The term long-term orienta79
80
81
For example, a risk utility function could be determined by calculating the probabilities of occurrence at which a decision maker is indifferent between a lottery with a certain outcome and a lottery with two uncertain outcomes. The difference between the weighted outcome and the certainty equivalent is the risk premium. Risk-averse decision makers have a positive risk premium; risk-seeking decision-makers have a negative risk premium. For further analysis of the relationship between risk and time horizon, see Samuelson (1989, 1994) or Zellweger (2007).
44
3 Long-term orientation of firms
tion is, in most cases, used without reference to the issue of optimality. The term myopia is used as an equivalent to short-termism. In this thesis, I do not make any conclusions about the issue of optimality and aim to find out whether – and under what conditions – family firms invest more resources in long-term projects, regardless of whether that is the best decision from a financial perspective. I refer to such behavior as long-term orientation and refer to the opposite as short-term behavior. Table 3-1 below summarizes the concept of long-term orientation, as it is understood in this thesis. Table 3-1: Concept of long-term orientation in this thesis
Consider a situation in which costs and benefits of a decision fall into different time periods. In such a situation, all other things being equal, a more long-term-oriented firm uses lower discount rates than other firms to value future payoffs. Long-term orientation does not imply any position on optimality. Sustainability is another popular term describing the need to balance the long and the short term and refers to the survival chances of a particular system such as a company. Most often, however, the term is used expressing concerns about the stability of an ecological system (Brown et al., 1987; Shearman, 1990), which is why this term is not used in this thesis.
3.2
Dimensions of long-term orientation Long-term orientation has several dimensions. Miller and Le Breton-Miller (2005, 2006), re-
ferring explicitly to family firms, differentiate between the following dimensions: brand building, innovation, craftsmanship, operations, and deal making.
Brand building. A ‘brand builder’ aims to create a brand with a strong reputation, for which customers are willing to pay a premium. Long-term orientation is required for brand building because it takes a long time to develop and establish a brand. Farsighted promotion and distribution initiatives – with late paybacks – may help to build a strong brand. A way to operationalize this dimension would be to look at fluctuations in spending on advertising. A company that invests constantly in its brand(s) can be described as long-term oriented.
Innovation. An innovator aims to create innovative products or processes. Such a strategy usually requires constant investment in R&D, a creative and motivated workforce, and consideration of customers’ requests, feedback, and demands. The payoffs from innova-
3 Long-term orientation of firms
45
tion are uncertain and usually do not occur in the near term. This dimension could be operationalized by looking at R&D expenditures or personnel fluctuation in R&D departments.82 Alternatively, one could try to capture the output side of the innovation process by looking at the number of new products introduced or the number of patents obtained in a particular time period.83
Craftsmanship. A ‘craftsman’ aims to manufacture high quality products. In dynamic industries, quality requires a strong emphasis on constantly improving the product. However, investments in product and process improvements may take a long time to pay off (if they pay off at all). That is, such investments may be both long-term and uncertain. This dimension could be operationalized by considering quality awards or (costly) industry certifications (e.g., ISO 9002).
Operations. The superior operations strategy (as described by Miller and Le-Breton Miller (2005, 2006)) refers to Porter’s (1980) cost-leadership strategy and aims to provide the most economically beneficial offer. Often, this strategy requires a superior plant or equipment. These capital-intensive investments are generally long-term oriented, paying off only after a certain number of units have been produced and sold. A way to operationalize this dimension would be to look at investments in property, physical plant, and equipment (PPE).
Deal-making. A ‘deal-maker’ aims to grow the business through (friendly or unfriendly) acquisitions. Deal-making through friendly or unfriendly acquisition relies on personal networking outside the business (e.g., with potential targets, finance providers, or other firms that want to join a bid). Networking and maintaining good relationships require long-term commitment and can be seen as a long-term investment, especially since shortterm, opportunistic behavior can do serious harm to personal relations. This dimension could be operationalized by looking at the composition of and fluctuation in supervisory or advisory boards.84 In addition, one might look at cross-shareholdings, interlocking directorships or the stability of joint ventures.
82 83
84
Chapter 6 analyses R&D expenditures in family and non-family firms. Intellectual property generated in the R&D process (e.g., patents) is almost by definition long-term-oriented. Yet, using patents as a measure of long-term orientation is not without problems. For example, an increase in the filing of patents could be due to changes in the legal environment (Hall and Ziedonis, 2001; Reitzig et al., 2007) or due to short-term behavior in ‘patent wars’ (Jell, 2008; Guellec and van Pottelsberghe de la Potterie, p. 81). Often, long-term partners have a seat in the supervisory board.
46
3 Long-term orientation of firms
An aspect of long-term orientation that has not been covered by Miller and Le Breton-Miller (2005, 2006) concerns the relationship of a firm with its suppliers. A number of studies in the field of marketing have analyzed this relationship (e.g., Cambra-Fierro and Polo-Redondo, 2006, 2008; Heide and Johnson, 1990; Johnson, 1999). A long-term firm-supplier relationship seeks to create a high level of trust among the partners involved, and is usually not ended abruptly.
3.3
Long-term orientation in family firms: a discussion This section introduces the main explanations for long-term-oriented behavior and discusses
whether the explanations hold for family firms referring to family business theory (summarized in Section 2.3 above). However, as there is only little extant work on long-term orientation, I refer to the broader literature on myopia (the contrary of long-term orientation). Generally, the explanations that have been advanced in the literature for myopic behavior can be classified into six categories: (1) management practices, (2) managerial opportunism, (3) executive compensation schemes, (4) signaling, (5) fluid capital, and (6) stock market myopia.85 3.3.1
Management practices One explanation for managerial myopia that has been advanced is the existence of flawed
management practices.86 This explanation refers to, among other things, the overuse of standardized investment evaluation techniques (e.g., the discounted cash flow technique), which neglect intangible and hard-to-quantify payoffs (Hayes and Abernathy, 1980; Hayes and Garvin, 1982).87 Do family and non-family firms differ in their management practices? An empirical study by Bloom and Van Reenen (2007) evaluated management practices in family and non-family firms in the US, France, Germany, and the UK. The management practices were evaluated on a scale from one (worst) to five (best). They were grouped into four areas: operations (three practices), monitoring (five practices), targets (five practices), and incentives (five practices).88 Family firms, in par-
85 86
87
88
For a similar classification, see Laverty (1996). There exists a large literature on the diffusion of management practices. For example, the management fashion literature analyzes how new management practices emerge through the interplay between managers and ‘fashion setters’ such as consultants (Abrahamson, 1991, 1996). Note that Logue (1985) presents counter-evidence against the claim that discounting cash flow techniques have led to a decline in long-term investments. The 18 practices were modern manufacturing – introduction, modern manufacturing – rationale, process documentation, performance tracking, performance review, performance dialogue, consequence management, target breadth, target interconnection, target time horizon, targets are stretching, performance clarity and comparability, managing human capital, rewarding high performance, removing poor performers, promoting high performers, attracting human capital, and retaining human capital.
3 Long-term orientation of firms
47
ticular those where CEO succession is passed over to the oldest son (the rule of primo geniture), have substantially lower evaluation scores. Based on this result, Bloom and Van Reenen (2007) concluded that family firms are managed more poorly than non-family firms are. However, the authors present only an aggregated score and do not report the concrete management practices in cases where family firms have lower scores than non-family firms. An alternative explanation for their findings could be that family firms are managed in a different, less-formalized manner than nonfamily firms. It should be noted that what was considered as good practices was determined by consultants from McKinsey & Company. Whether a particular practice is considered as ‘good’ or ‘bad’ remains, to some degree, a subjective evaluation. Nonetheless, there is evidence that family and non-family firms differ with regard to the type and extent of management practices used. Further research might look deeper into this issue and analyze the concrete management practices in which family and non-family firms differ. At this point, it is impossible to make a statement about whether the use of management practices leads to differences in the long-term orientation of family and nonfamily firms. 3.3.2
Managerial opportunism Managerial opportunism is a prominent explanation for short-term managerial behavior. It is
argued that under certain circumstances short-term behavior is an optimal choice from a manager’s perspective. In moral hazard models, managers either prefer making short-term investments that pay off quickly, in order to enhance personal reputation (Narayanan, 1985), or are preoccupied with job safety and therefore favor short-term payoffs (Hirshleifer and Thakor, 1992). Arguments exist from both agency and stewardship theory that managerial opportunism is less of a problem in family firms relative to non-family firms. Both of these theories have been treated extensively in the previous chapter.89 First, I summarize the relevant arguments from agency theory. From an agency theorist’s perspective, the relationship between management and ownership is fundamentally different in a situation where the firm is governed by a family manager, compared to a situation where a firm is governed by a non-family manager. In the first case, managerial opportunism should be less of a problem because the utility functions of the main shareholders and management coincide to a strong degree. This argument is even stronger given the fact that most family managers own a substantial share of the companies they manage. In this case, they resemble ownermanagers for whom no agency conflicts exist between management and ownership.90 Another ar-
89 90
For a description of agency theory, see Section 2.3.1; for a description of stewardship theory, see Section 2.3.2. See the discussion of the model of Jensen and Meckling (1976) in Section 2.3.1.
48
3 Long-term orientation of firms
gument against managerial opportunism from family managers is that family managers have rather safe jobs (e.g., Allen and Panian, 1982; Le Breton-Miller and Miller, 2006), since it can be difficult to lay off family members. Consequently, family managers do not need to increase their reputation in the market for executives by attaining strong short-term results at the expense of the long term. Stewardship theory describes situations where individuals generate a higher utility from proorganizational, collectivist behavior than from individualistic, self-serving behavior (Davis et al., 1997). In situations in which managers behave as stewards, the managerial-opportunism explanation of short-term behavior is meaningless. Davis et al. (1997) propose psychological and situational factors that lead individuals to act as pro-organizational, collectivist stewards rather than as individualistic, self-serving agents. I argue that some of these factors are more likely to be found in family than in non-family firms. Davis et al. (1997) suggest that people who identify strongly with the organization or are high in value commitment are more likely to act as stewards rather than as agents. For family managers, the argument is simple. Because the firm is part of the family’s identity, family managers often identify strongly with the firm. Additionally, family managers are also likely to believe in (and accept) the firms’ goals, which are determined to a large extent by the owning families; value commitment should be higher. Regarding situational factors that lead to a stewardship behavior, Davis et al. (1997) propose that people who are in an involvement-oriented- rather than a control-oriented-situation are more likely to behave as stewards. In family-owned firms that are governed by family managers, the management philosophy is more likely to be involvementoriented. Two reasons for this exist: (1) Shareholders and managers are members of the same family and are likely to trust one another, reducing the need for control, and (2) a sophisticated monitoring system would only be of marginal use. Few possible sanctions would be appropriate, since it may be difficult to punish a member of one’s own family. 3.3.3 Executive compensation Under some circumstances, defectively designed incentive compensation schemes may lead to short-term-oriented managerial behavior. Reichelstein (2000) showed that accounting-based incentive schemes can cause such problems. For many privately held firms, the share price is unobservable and, hence, compensation schemes must be based on accounting or other indicator variables.91 In an experimental study, Schotter and Weigelt (1992) found that the discounting of future outcomes may depend upon bonus payments. In case the bonus reflects only current performance, individuals tended to undervalue future payoffs. 91
See Bolton et al. (2005) or Dutta and Reichelstein (2003) for more papers on the relationship between incentive compensation and managerial myopia.
3 Long-term orientation of firms
49
Recently however, in particular stock option payment has been criticized for its adverse effects with regard to managerial behavior (Bebchuk and Fried, 2003; Frey and Osterloh, 2005). Among other arguments, it is claimed that managers take short-term actions to drive up the firm’s share price and thereby, the value of the stock options they own. On the other hand, it could be argued that stock and stock option pay is a remedy for, and not a cause of, managerial myopia (Dechow and Sloan, 1991; Hall, 2003). Unlike accounting profits that measure the past performance, stock prices that underlie stock and stock option payment reflect expectations about the future. In contrast to simple stock payments, stock options tend to increase in value with higher firm risk. With stock option compensation, managers are given an incentive to pursue risky investment projects that may only pay off in the long run. In addition to controversy about the effects of equity-based pay on managerial behavior, there are reasons to believe that equity-based compensation is less frequently observed in family than in non-family firms. First, since family owners often know their business well and have large equity positions, they should be good monitors. Since the manager’s behavior can be observed directly, an outcome-based contract would be suboptimal because it needlessly transfers risk to the agent (Eisenhardt, 1989). Monitoring and incentive compensation are, to some degree, appropriate substitutes (Holmstrom, 1979); this is why equity compensation should be lower in family firms than in nonfamily firms. Second, to keep control of the firm in the family’s hands, family owners are often more reluctant to compensate non-family managers with equity. Third, the goals of family owners and family managers are sometimes non-financial in nature (Donckels and Frohlich, 1991; Harris and Martinez, 1994; Tagiuri and Davis, 1992). Incentive compensation, such as equity-based compensation, could be detrimental to non-financial goals. Finally, many family firms are private, i.e., not listed on the stock market; equity-based compensation is thereby not possible due to the lack of a benchmark. 3.3.4
Information asymmetry and signaling Signaling is another explanation for managerial myopia. According to this explanation, man-
agers know more about the firm than investors do, and they need to use strong short-term results as a signaling device to show the quality of their management (Thakor, 1990). The situation is different with family owners. Often, family owners have grown up with the business or have even created the business on their own. Hence, they are familiar with the firm’s business model, its management team, and the industry’s characteristics. The information asymmetry between owners and management in family firms should be lower than in non-family firms. This is even more likely to be the case if owners and managers are related. Consequently, the manager is not forced to use strong
50
3 Long-term orientation of firms
short-term results to build up a reputation as a good manager. In addition to that, since it is their money (and legacy) which is at stake, family owners are more likely to engage in monitoring. This further decreases information asymmetry and leads to more long-term investment decisions (von Thadden, 1995). 3.3.5 Impatient capital, stock market myopia Porter (1992) argues that underinvestment may be the result of the short-term relationship between US firms and the capital market. In contrast to Germany or Japan, where a large portion of equity is held by banks or (non-financial) firms, funds supplied in the US often come from external capital providers (e.g., pension funds or other professional investment firms). Porter (1992) argues that such external capital providers can move their funds more quickly than other investors and therefore do not understand – or care about – the respective companies’ long-run prospects. Stock market myopia is a related explanation for short-term managerial behavior. The argument is that the stock market undervalues long-term-oriented investment behavior, and accordingly managers are forced to think in the short term to avoid the risk of a takeover (Stein, 1988). Johnson and Kaplan (1987) argue that investors employed by professional financial investment firms focus too strongly on short-term figures in quarterly or annual reports. This was strongly corroborated by the so-called accrual anomaly, first strikingly documented by Sloan (1996).92 This extensivelyanalyzed phenomenon theorizes that capital markets overvalue firms that increase their reported earnings by pure earnings-management activities.93 Jacobs (1991) makes a similar argument. If stock is traded as a commodity (i.e. like crude oil or any other raw material), many stockholders will not have much interest in waiting for long-term projects to pay off. For firms that have a family as their main shareholder, the explanations for managerial myopia involving stock market myopia or fluid and impatient capital are of less relevance for two reasons. First, from the perspective of a family owner, the firm is not just an asset that might be sold for profit, since the firm symbolizes the heritage and tradition of the family and is therefore part of the family’s identity. Consequently, family owners frequently intend to pass the firm on to the next family generation (e.g., Casson, 1999; Guzzo and Abbot, 1990; James, 1999; Tagiuri and Davis, 1992). 92
93
The strength of this anomaly depends on the accounting standards used (e.g., German- vs. US-GAAP) and the strength of the corporate governance system. See Kaserer and Klingler (2008). Earnings management (or creative accounting) refers to a strategy used by a firm’s managers to deliberately manipulate the firm’s reported earnings so that the figures match a pre-determined target. Rather than having years of exceptionally good or bad earnings, earnings-management activities attempt to maintain relative stability (using so-called income smoothing). Concrete examples include the use of inappropriate estimates of liabilities or generous reserve accounting.
3 Long-term orientation of firms
51
Second, the public reputation of the family and its members is linked strongly to the wellbeing of the firm (Dyer and Whetten, 2006). This becomes even more evident when the firm bears the family’s name. Hence, family owners should be unlikely to move their funds around quickly, and they should also be unlikely to evaluate their investments only in terms of short-term returns. The theory of psychological ownership might also be used to explain differences between family and non-family owners. Pierce et al. (2001) define psychological ownership as “the feeling of possessiveness and of being psychologically tied to an object” (Pierce et al., 2001, p. 299). One root of psychological ownership lies in the control of the particular object that one owns. I argue that this applies more to owners of family firms than to owners of non-family firms. In contrast to owners of non-family firms, owners of family firms often have a close link to management (often by kinship ties). This close link allows them to explore and alter the firm and its environment. As a result, they have a greater feeling of psychological attachment to the firm as compared to owners of non-family firms.
3.3.6
Further explanations from family business theory Some explanations for managerial myopia that follow from the distinct characteristics of fam-
ily firms have not been considered in general discussions about managerial myopia. These explanations are summarized in the following. Greater visibility of family owners Family owners are more easily identifiable than other owners, such as institutional investors. In conjunction with agency theory, this gives an argument for the stronger long-term orientation of family firms. Agency theory may explain the relationship between family owners and society. Family owners correspond to agents, being monitored and sanctioned by society (being the principal). In line with Wiklund (2006), one may argue that family owners can be more easily monitored and sanctioned by society than other types of owners. Unlike institutional investors, family owners often have their wealth tied to a particular firm and are more easily identifiable, as they are often well known and bear the same name as the firm. Thus, compared to other types of owners, family owners should be more likely to be concerned with their reputation as socially responsible, and should be more likely to be reliable long-term investors. This greater concern for reputation makes such owner-managers more fearful of the negative image associated with opportunistic short-term behavior.
52
3 Long-term orientation of firms
Private benefits of control Private benefits of control are an argument against a stronger long-term orientation of family firms, relative to non-family firms.94 The separation of ownership and control is generally more pronounced in family than in non-family firms (Claessens et al., 2000; La Porta et al., 1999). The reason is that, through pyramid structures or dual class stock, the family’s voting rights may exceed their formal cash-flow rights. Moreover, the family may use their (formal or informal) influence to appoint directors who are friends or family members. Shleifer and Vishny (1997) argue that such entrenched shareholdership may not be in the firm’s best interests, as the firm might be used as a vehicle to generate private benefits of control not shared by other (non-family) shareholders (for this argument, see Claessens et al., 2002; McConnell and Servaes, 1990; Stulz, 1988). The family could also use their influence to divert resources from projects with long-term value for the firm to projects that mainly benefit the family. Then, the firm pursues a rather short-term business policy. To give an example, the family CEO might prefer to invest in a business that is traditionally connected to the family but that is not the most promising business from the firm’s perspective. Altruism and self-control problems An argument can be made that altruism and problems of self-control lead family managers to act short-term. The theory of agency and altruism from Schulze et al. (2001, 2003)95 argues that firms with a family CEO suffer agency costs due to altruism and problems of self-control. The family CEO is forced from family pressure to make transfers that benefit the members of the owning family, who, in turn, have an incentive to place themselves in an ‘impoverished’ condition. Additionally, the family CEO cannot effectively monitor and discipline members of the family, as the CEO’s perceptions will be biased and such discipline could harm familial relationships. How does this relate to long-term orientation? If the CEO wants to pursue a long-term strategy that mainly benefits the firm, she will have to fend off the demands of the other family members. Every dollar that is taken out of the firm by the family is a dollar that cannot be invested in long-run projects of the firm. In might be the case that due to excessively high dividends the firm can no longer finance the investments needed to survive in the long term.
94
95
For a deeper discussion of private benefits of control, see the paragraph on the agency costs from conflicts between dominant and minority shareholders in Section 2.3.1. See Section 2.3.1.
3 Long-term orientation of firms
53
Conflicts within the family Family firms are often plagued by substantial conflicts arising from sibling rivalries, childrens’ desires to distance themselves from their parents, marital discord, and ownership dispersion among family members (Dyer, 1994; Kellermanns and Eddleston, 2004; Levinson, 1971; Schulze et al., 2001). The family and the business become thoroughly intertwined, and the potential for conflict grows (Lee and Rogoff, 1996). Once a conflict emerges, it is difficult to resolve the conflict by pushing a family member out of the firm, since this would also have negative effects on familial relationships. The stewardship theory suggests that conflicts within an organization can lead to short-term thinking and short-term behavior.96 There is also a strong negative effect on non-family employees, who may get the impression that due to inner-family conflicts the family members are not qualified to manage the firm. They may also get the impression that the family members are managing the business mainly for their own benefit and that other stakeholders of the firm such as non-family employees are not important. Extended utility horizons James (1999) compares the investment rule of an owner-manager without children to the investment rule of an owner-manager with children. He argues that the owner-manager without children may choose a suboptimal investment level, since she may have a preference for consumption today relative to consumption in the future. Suboptimal investment means that the owner-manager invests less in the firm than what is optimal from the market investment rule, i.e., the marginal return of an investment in the firm is higher than the marginal return of an investment in the market.97 The situation differs when the owner-manager has children. Assuming that the owner-manager receives positive utility when her child’s utility increases, the owner-manager is less likely to invest at a suboptimal level. Her preference for immediate consumption is balanced by the positive utility she gains from her children’s future consumption. An extended utility horizon thus leads to more optimal investment levels. 3.3.7
Summary of the discussion The preceding discussion demonstrates both the arguments in favor of a stronger long-term
orientation of family firms and the arguments against a stronger long-term orientation of family
96 97
For a description of stewardship theory, see Section 2.3.2. Note that in perfect capital markets, such a situation would not arise. The owner could borrow at the market rate to finance her consumption in period one and pay off the debt in period two with the returns from the investment in the firm (Fisher, 1930; Hirshleifer, 1958).
54
3 Long-term orientation of firms
firms. Table 3-2 summarizes these arguments. The arguments are grouped according to whether they refer to the management or ownership dimension of family firms. Table 3-2: Arguments regarding long-term orientation of family firms Pro: Stronger long-term orientation in family firms
Contra: Stronger long-term orientation in family firms
Ownership dimension
Ownership dimension
- More effective monitoring possible due to greater visibility of family owners in the society
- Private benefits of control to the detriment of minority shareholders and the firm as a whole
- Firm is part of the family identity and is intended to be transferred to the next family generation
- Altruism towards family members; self-control problems of family members
- Family owners are more psychologically tied to their firm than other owners
- Conflicts within the family induce short-term thinking
- Lower information asymmetry between owners and management Management dimension - Family CEO is in a more powerful position than a non-family CEO - Lower share of equity-based compensation in family firms - Extended utility horizon due to children - Long tenures of family CEOs
3.4
Research questions The question of long-term orientation of a firm is of great relevance for the firm’s stake-
holders (e.g., employees, customers, suppliers) and society in general. Furthermore, a firm’s longterm orientation is often referred to as a driver of superior financial performance, which is what most investors are ultimately looking for. A large number of firms around the world are family-owned and/or family-managed. This is true for small and large as well as for listed and non-listed firms.98 Family firms themselves often claim to behave more long-term than non-family firms and use this to market themselves as reliable partners.99 Theory however is unclear as to whether family firms are more long-term-oriented than other firms. Drawing on agency and stewardship theory, arguments both supporting and speaking 98 99
See the overview of studies about the importance of family firms in the introductory chapter. See the interviews and press statements in the introductory chapter.
3 Long-term orientation of firms
55
against a stronger long-term orientation of family firms relative to other firms can be indentified.100 From a theory perspective, the question of a stronger long-term orientation of family firms relative to other firms is an open question. The existing empirical literature on long-term orientation in family firms is so far mostly interview-based qualitative research. Large-scale quantitative studies comparing long-term orientation of family and non-family firms are missing. This thesis aims to close this research gap. Long-term orientation of a firm is a concept that has several dimensions and involves many aspects such as brand building, innovation, and human resource management.101 Reflecting this multidimensionality, this thesis analyzes long-term orientation of family firms in several ways.
1. R&D spending. Spending on R&D is unlikely to pay off in the immediate future – it is inherently of long-term nature. Thus, R&D spending can be used as a proxy for long-term investing more generally. This leads to the following set of questions:
Do family firms, ceteris paribus, exhibit a higher level of R&D spending than non-family firms? Does R&D spending depend on the involvement of the family into the management of the firm? Can we observe a difference between young and old family firms, i.e. first-generation versus later-generation family firms? To what extent do potential differences in R&D-spending contribute to the observed differences between family- and non-family firms in terms of financial performance (measured both by operating and stock market performance)?
2. Downsizing. If family firms are more long-term-oriented, then they can be expected to put a high weight on some aspects of corporate social responsibility (CSR). Responsibility towards employees is certainly part of a long-term CSR policy. Besides just having a positive image with employees, firms may have a direct interest in keeping good relations with their employees. Investments in employees are mostly long-term and most of these investments are lost when employees leave the firm. Thus, downsizing, i.e. laying off employees on a large scale, is a rather short-term strategy.102 This leads to the following set of questions:
100 101 102
See Section 3.3 and Table 3-2. See Section 3.2. The only exception might be the case of impending bankruptcy.
56
3 Long-term orientation of firms
Are family less likely to downsize than non-family firms? Is there a difference in the degree of downsizing between family and non-family firms? What is the role of family management and family ownership – the two distinct characteristics of family firms – with regard to downsizing?
3. Executive pay (empirical analysis). Compensation schemes are an important means to motivate managers to act in the owner’s interest. Yet, executive compensation is also referred to as a reason of managerial myopia.103 Via stock option, stock or bonus pay, the firm’s owners can set strong short-term incentives for the managers.104 The structure of executive compensation reflects to some degree the goals of the firm’s owners and is thus an indicator of the owner’s interest in long-term goals. If family owners are more long-term-oriented relative to other owners, the structure of executive compensation should be different in family versus non-family firms. This leads to the following set of research questions:
What is the role of incentive pay in family firms (in particular stock options)? Are non-family managers in family firms paid in a different way than family managers are? What is the role of incentive payment in this case?
4. Executive pay (theoretical analysis). The final set of research questions is of a normative nature. Using a principal-agent model, this thesis aims to answer the question how a non-family manager in a family firm should be compensated given that the owning family pursues the long-term goal of preserving the existence of the firm. The case is extremely relevant for family firms with succession problems, where due to lack of a suitable family candidate a non-family manager is often the only alternative.
Given that the family wants to preserve the long-term existence of the firm, what is the optimal structure of executive pay? How can the family assure that the non-family manager pursues this goal rather than pursuing more short-term goals? What is the role of incentive pay in this context? Might it even be harmful in that the manager is distracted from pursuing more long-term goals?
103 104
See Section 3.2. As an illustration, consider the case of stock option pay. Executive stock options usually have a maturity of between one and five years. The CEO is required to hold the option for a pre-specified time period (vesting period). It also makes a difference whether European or American options are issued. European options can only be excised at a particular date, whereas American options can be exercised at any date.
3 Long-term orientation of firms
57
These questions constitute the basis of this thesis. The next chapter (Chapter 4) describes the dataset and the methods used to analyze the (empirical) questions. To make the connection between longterm orientation of the firm and financial performance, Chapter 5 analyzes the financial performance of family and non-family firms. Chapter 6 then compares the R&D spending of family relative to non-family firms; Chapter 7 deals with downsizing, and Chapter 8 uses executive compensation data to analyze the incentives given in family and non-family firms. Finally, Chapter 9 answers the normative question of how to reward a non-family manager working in a family firm so that the firm is managed in a long-term way. Chapter 10 summarizes and concludes.
4 Data and method The chapter describes the construction of the dataset (Section 4.1). Furthermore, descriptive statistics are used to compare family and non-family firms in terms of firm characteristics, governance and CEO characteristics, as well as industry membership (Section 4.2). The chapter closes with a description of Bayesian analysis, a statistical method that is used in addition to classical methods in this thesis (Section 4.3).
4.1
Construction of the dataset As already mentioned in the introduction of this thesis, the empirical part of this thesis is
based on US data. The Standard & Poor’s 500 firms (as of July 31, 2003) are the starting point for constructing the sample. The particular date chosen corresponds to an issue of BusinessWeek in which family firms in the S&P 500 were indicated (BusinessWeek, 2003).105, 106 Moreover, the publication gives qualitative information on the ownership structures and management compositions of the 177 family firms covered. The S&P 500 is a stock market index containing the 500 largest publicly listed firms in the US. The index is maintained by Standard & Poor’s and forms part of broader stock market indices such as the S&P 1500 and the S&P Global 1200. The S&P 500 and related indices, such as the Fortune 500, are used widely in the literature to compare the performance of family and non-family firms (e.g., Anderson and Reeb, 2003; Miller et al., 2007; Villalonga and Amit, 2006).
105
106
BusinessWeek (2003) defines any “company where founders or descendants continue to hold positions in top management, on the board, or among the company’s largest stockholders” (p. 111) as a family firm. The definition is very similar to Anderson and Reeb (2003). Table A-1 in the Appendix shows the firms that were included in the S&P 500 index at this date. I am grateful to Andreas Riemann for providing me with the exact composition of the S&P 500 index on July 31, 2003. Family firms (as identified by the BusinessWeek, 2003) are marked with an asterisk.
4 Data and method
59
I collected detailed data about the firms’ ownership structures and management compositions from corporate proxy statements submitted to the US Securities and Exchange Commission (SEC) in the years 1992-2003.107, 108 I then checked and expanded my data with information from Hoover’s Handbook of American Business109, Gale Business Resources110, the Twentieth-Century American Business Leaders Database at Harvard Business School111, Forbes Lists of the 400 Richest Americans112, Marquis Who’s Who in America113, and information available on the firms’ websites. In a final step, the databases Compustat North America and Compustat ExecuComp were used to get additional firm, manager, and market data.114 Compustat North America delivers financial, statistical, and market data on more than 10,000 active and more than 8,000 inactive US and Canadian publicly quoted firms.115 The ExecuComp database provides annual data on executive compensation. It also provides manager-specific information such as age, gender, and employment history.116 All variables used in this thesis are described in Table A-2 in the Appendix. The final sample is an unbalanced panel data set with 4,856 observations from 499 firms.117 The reduction in the number of observations per firm is due to the fact that some firms were not listed on the stock market over the entire period from 1992-2003.
107
108
This information was mostly found in the definitive proxy statement (DEF 14A). The Securities Exchange Act of 1934 requires officers, directors, and five-percent owners to disclose their holdings. Proxy statements are the most accurate source of information about ownership structures. In particular, they are more accurate than databases such as Compact Disclosure, which are found to have many mistakes and biases (Anderson and Lee, 1997; Dlugosz et al., 2006). I am grateful to Thomas Daffner, Gaurav Rishi, Frank Spiegel, and Marc Weiglein for excellent research assistance in collecting the data from the corporate proxy statements.
109
I used the hardcopy version. For further information, see http://www.hoovers.com (accessed September 13th, 2008).
110
See http://www.gale.cengage.com/BusinessRC (accessed September 13th, 2008).
111
See http://www.hbs.edu/leadership/database (accessed September 13th, 2008).
112
See http://www.forbes.com/lists (accessed September 13th, 2008).
113
114
115 116
117
I used the hardcopy version. For information on the product, see http://www.marquiswhoswho.com (accessed September 13th, 2008). I am grateful to the Deutsche Forschungsgemeinschaft (DFG) and the SFB 649 Economic Risk for providing access to the databases. In particular, I would like to thank Andreas Hey of the Financial and Economic Data Center (FEDC) for his support with the handling of the databases. See https://www.compustatresources.com (accessed September 14th, 2008). A good description of the ExecuComp database and its variables can be found at http://umi.compustat.com/docsmi/help/execdefs.htm (accessed September 13th, 2008). The firm AT&T is listed twice in the S&P 500 index in July, 2003, and I decided to include it only once in the sample.
60
4 Data and method
4.2
Descriptive analysis of the dataset
4.2.1
Types of family firms and dimensions of family influence I use the family firm definitions described in Table 2-2 to divide the entire sample into several
groups of family and non-family firm observations. Five subgroups arise: non-family firm observations (3,110 obs.), family firm1 (1,746 obs.), family firm2 (1,665 obs.), family firm3 (585 obs.), family firm4 (428 obs.).118 The results show that the size of the groups of family firms depends strongly on the specific definition applied. Using a definition that requires the family to be active as both owner and manager of the firm reduces the number of family firms by more than 60% (1,746 obs. with family firm1 vs. 585 obs. with family firm3). Figure 4-1 displays the size of the different groups graphically. Figure 4-1: Number of family firm observations in the sample
Total sample (4,856 obs.=100%)
Family firm1 (1,746 obs.=36%)
Family firm3 (585 obs.=12.0%)
Notes: Family firm1 is defined as “family stock ownership > 5% or a member of the family is CEO or chairman.” Family firm3 is defined as “family stock ownership > 5% and a member of the family is CEO or chairman.”
118
Note that the family firm definitions are not mutually exclusive. See also Figure 4-1.
4 Data and method
61
Table 4-1: Changes between the different firm categories Non-family firm Non-family firm Family firm1 Family firm2 Family firm3 Family firm4
30/60 25/50 0/0 0/0
Family firm1
Family firm2
Family firm3
Family firm4
30/60
25/50 15/23
0/0 68/35 68/29
0/0 73/50 73/48 25/36
15/23 68/35 73/50
68/29 73/48
25/36
Example how to read the table: There are 30 changes from the category non-family firm (x-axis) to the category family firm1 (y-axis) and 60 changes the other way round. The number of observations that was considered was N=4,856.
Table 4-1 above shows how often a firm changes its category. A non-family firm becomes a family firm (family firm1) in 30 cases (0.6% of all observations), whereas a family firm (family firm1) becomes a non-family firm in 60 cases (1.2% of all observations). Table 4-2 below presents the channels through which the family influences the business. The group of family firms is defined using the broadest definition (family firm1). The following results regarding firm management stand out: In 83% of the 1,746 observations, a member of the founding family is chairman of the board of directors, while a member of the founding family is CEO in 69% of all cases. Family influence on the firm’s management is high. This is different when the ownership dimension is considered. The mean percentage of common stock owned by the family is 12.6%. Yet, the variation is large: The median is 5.8%, and the standard deviation is 17.3%. Overall, the family’s ownership shares seem to be low, particularly when compared to family firms in Continental Europe or Asia (Andres, 2008; Becht and Roell, 1999; Claessens et al., 2000; La Porta et al., 1999).119 In only 17% of the cases, the family owns more than 30% of common stock, and in only 4% of the cases does the family own more than 40% of common stock. In 40% of the cases, the family owns more stock than financial investors do. These numbers fit well given the particularities of the US system of corporate governance. Due to strong investor protection and a widely dispersed ownership structure of US publicly listed firms in general, the family does not need to own large portions of equity to further its interests.120
119
120
In Europe, the median size of the largest voting block of listed industrial companies ranges from 9.9% (UK) to 55% (Italy) (Becht and Roell, 1999). In Hong-Kong, about 67% of all listed firms have a family shareholder with voting rights of more than 20% (Indonesia: 69%, Thailand: 56%, Japan: 13%) (Claessens et al., 2000). See also Section 2.1.2.
62
4 Data and method
Table 4-2: Channels through which the family influences the business Dimensions of family influence Management CEO is member of the family Chairman of the board of directors is member of the family CEO or chairman is member of the family CEO and chairman are members of the family Ownership Percentage of common stock owned by the family Family owns more than 5% of common stock Family owns more than 20% of common stock Family owns more than 30% of common stock Family owns more than 50% of common stock Family owns more stock than financial investors
69% 83% 85% 62% Mean: 12.6% Median: 5.8% Std. dev.: 17.3% 56% 25% 17% 4% 40%
Experience Age of the family firm
Mean: 49.5 yrs Median: 35 yrs Std. dev.: 39.9 yrs
Notes: The definition family firm1 is used to define the group of family firms. N=1,746
4.2.2
Characteristics of firms Table 4-3 displays characteristics of the firms in the sample grouped by family and non-family
firm observations using the five family firm definitions listed in Table 2-2. Striking differences emerge121: Family firms are smaller than non-family firms, as can be seen from the firm’s assets (median 3.5 bn $ vs. 9.4) and sales (median 2.9 bn $ vs. 5.9) as well as the number of employees (median 13,030 vs. 22,090). In addition to that, family firms are younger than non-family firms. Non-family firms have a median age of 89 yrs, whereas the median age of family firms is less than 40 yrs. An interesting result stands out with regard to cash flow: Due to their smaller size, family firms report a lower cash flow than non-family firms. Yet, when correcting for firm size, it turns out that family firms have more internal finances available than non-family firms. Median cash flow in percent of assets is about 9% in the group of non-family firms, and with family firms this number is between 10.8% (family firm3) and 11.6% (family firm5a). Finally, note that the differences between family and non-family firms seem to be larger than the differences within the group of family firms. For example, consider the small difference in median assets between the group of family firms defined by a broad definition (e.g., family firm1) and the group of family firms defined by a narrow 121
Again, the definition family firm1 is used to define the group of family firms.
4 Data and method
63
definition (e.g., family firm3). The last two columns allow a differentiation between young and old family firms (family firm5a vs. family firm5b). Young family firms have less employees than old family firms (median 12,700 vs. 14,800, p<0.01) and a higher ratio of cash flow to assets (median 11.6% vs. 10.2%, p=0.023). No significant differences exist with regard to the other variables. Table 4-3: Firm characteristics by family firm definition Non-family firm (3,110 obs.)
Family firm1 (1,746 obs.)
Family firm2 (1,665 obs.)
Family firm3 (585 obs.)
Family firm4 (428 obs.)
Family firm5a (257 obs.)
Family firm5b (171 obs.)
Wilcoxon rank-sum test 1
Median
Median
Median
Median
Median
Median
Median
Assets (in bn $)
9.4
3.5
3.6
3.1
3.1
3.3
2.9
p<0.001
Employees (in 1,000)
22.1
13.0
13.0
13.8
14.0
12.7
14.8
p<0.001
Firm age (in yrs)
89
36
35
37
40
25
75
p<0.001
Sales (in bn $)
5.9
2.9
2.9
3.0
2.9
3.0
2.7
p<0.001
Cash flow (in bn $)
0.6
0.3
0.3
0.3
0.3
0.4
0.3
p<0.001
Cash flow/ assets (in %)
9.0
10.9
10.8
10.8
11.4
11.6
10.2
p<0.001
Notes: See Table 2-2 for details regarding the family firm definitions. Note that the family firm definitions are not mutually exclusive. 1 Family firm1 versus non-family firm; two-sided test is used; the Wilcoxon rank-sum test analyzes whether the two samples are from different distributions (sample1: family firm1, sample 2: non-family firms). N=4,856 obs. from 499 firms
64
4 Data and method
4.2.3 Governance and CEO characteristics Table 4-4 gives governance and CEO characteristics of family and non-family firms (by definition family firm1) in the sample. Some differences appear.122 CEOs in family firms are slightly older on average (median 59.7 yrs vs. 58.6), serve a longer tenure (median 9.4 yrs vs. 5.3), and are less likely to serve as both CEO and chairman of the board of directors (CEO duality) (72% vs. 88%). They receive lower total pay (median $3.3 mn vs. $4.5), but tend to have a higher share of stock options in their pay package (median 43% vs. 39%). The board of directors tends to be smaller in family than in non-family firms (median 19 members vs. 22). Family firms are financed to a lower degree by debt (median 21% vs. 26%). In addition, they prefer long-term debt to short-term debt, as can be seen by the low ratio of short-term to long-term debt. Finally, the average percentage of stock owned by financial investors is lower in family than in non-family firms (median 11% vs. 15%). Interestingly, the definition of family firm used has an impact on the size of the differences between family and non-family firms. When a narrower definition is used, the differences between family and non-family firms become more pronounced (with regard to, e.g., CEO duality, ownership share of financial investors, and total debt/ assets). For example, the ownership share of financial investors drops from 11.3% to 9.1% (family firm1 vs. family firm3) or the ratio of total debt to assets drops from 21.3% to 17.8%. The last two columns reveal also differences between young and old family firms (family firm5a vs. family firm5b). Young family firms have a smaller board (mean 18.2 vs. 21.6, p<0.01) and a lower share of short-term debt (mean 21.4% vs. 27.3%, p=0.04). Relative to CEOs of old family firms, CEOs of young family firms are on average younger (mean 58.1 yrs vs. 62, p<0.01), receive a higher salary (median $3.1 mn vs. $1.8, p<0.01), have a higher share of stock options in total pay (mean 39.5% vs. 29.3%, p<0.01), and are less likely to serve as both CEO and chairman at the same time (61.0% vs. 71.6%, p=0.03).
122
I compare non-family firm observations with family firm observations as defined by family firm1.
4 Data and method
65
Table 4-4: Governance and CEO characteristics by family firm definition Non-family firm (3,110 obs.)
Family firm1 (1,746 obs.)
Family firm2 (1,665 obs.)
Family firm3 (585 obs.)
Family firm4 (428 obs.)
Family firm5a (257 obs.)
Family firm5b (171 obs.)
Test for the equality of means/ proportions 1
CEO duality (share in %)
87.9
71.8
71.0
65.8
65.1
61.0
71.6
p<0.001
CEO tenure (mean in yrs)
5.3
9.4
9.6
9.7
10.2
9.5
11.2
p<0.001
CEO age (mean in yrs)
58.6
59.7
59.7
59.3
59.6
58.1
62.0
p<0.001
CEO is male (share in %)
99.3
98.8
98.7
98.6
98.8
98.0
100
p=0.075
Total debt/ assets (mean in %)
26.1
21.3
21.4
17.8
16.9
16.1
18.0
p<0.001
Short-term debt as % of longterm debt (mean in %)
27.4
24.1
23.9
25.3
24.0
21.4
27.3
p<0.001
Ownership by financial investors (mean in %)
15.0
11.3
10.8
9.1
4.4
4.8
3.9
p<0.001
CEO pay (median in mn $)
4.5
3.3
3.3
2.7
2.3
3.1
1.8
p<0.001
Share of stock option pay (mean in %)
39.4
43.1
43.1
38.9
35.4
39.5
29.3
p<0.001
Board size (mean no. of members)
21.9
19.2
19.1
18.9
19.6
18.2
21.6
p<0.001
Notes: See Table 2-2 for details regarding the family firm definitions. Note that the family firm definitions are not mutually exclusive. 1 Family firm1 versus non-family firm; t-test for equality of means, Chi²-test for equality of proportions, or Wilcoxon rank-sum test is used. N=4,856 obs. from 499 firms
66
4.2.4
4 Data and method
Industry classification Table 4-5 shows family and non-family firm observations grouped by industry classification.
Again, the definition family firm1 is used to distinguish between family and non-family firm observations. Overall, the share of family firm observations is 36%, which is comparable to other studies based on US data (e.g., Anderson and Reeb, 2003; Miller et al., 2007; Villalonga and Amit, 2006). Table 4-5 suggests that family firms are present in 85% of the SIC codes. Yet, the share of family firm observations varies strongly across industries. Family firms are overrepresented123 in the industries food products (SIC 20), apparel and other textile products (SIC 23), printing and publishing (SIC 27), electronic and other electrical equipment (SIC 36), general merchandise stores (SIC 53), apparel and other accessory stores (SIC 56), eating and drinking places (SIC 58), miscellaneous retail (SIC 59), and business services (SIC 73). Family firms are underrepresented124 in the industries paper and allied products (SIC 26), petroleum and coal products (SIC 29), fabricated metal products (SIC 34), transportation equipment (SIC 37), railroad transportation (SIC 40), as well as electric, gas, and sanitary services (SIC 49). Table 4-5 also gives the capital intensity in the respective SIC industry. Capital intensity is measured as the firm’s share of property, plant, and equipment (PPE) in assets. The share of family firms tends to be high in industries with low capital intensity, and vice versa. Yet, this tendency is not strong. The Pearson correlation coefficient between the capital intensity of the industry and the share of family firms in the industry is only r=-0.15 (p=0.30).125
123
124
125
Overrepresented means that more than 50% of the observations belong to family firms and that the industry has at least 40 observations. Underrepresented means that less than 20% of the observations belong to family firms and that the industry has at least 40 observations. Financial services industries (SIC 60-63 and SIC 65-67) are excluded, since capital intensity in these industries is almost zero.
4 Data and method
67
Table 4-5: Family and non-family firm observations by industry SIC Code 10 13 14 15 16 20 21 23 24 25 26 27 28 29 30 33 34 35 36 37 38 39 40 42 44 45 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 67 70 72 73 75 78 79 80 82 87 99
Industry description Metal mining Oil and gas extraction Mining nonmetalics, except fuels General building contractors Heavy construction, except buildings Food and kindred products Tobacco products Apparel and other textile products Lumber and wood products Furniture and fixtures Paper and allied products Printing and publishing Chemical and allied products Petroleum and coal products Rubber and miscellaneous plastic products Primary metal industries Fabricated metal products Industrial machinery and equipment Electronic and other electrical equipment Transportation equipment Instruments and related products Miscellaneous manufacturing products Railroad transportation Trucking and warehousing Water transportation Transportation by air Communications Electric, gas, and sanitary services Wholesale trade—durable goods Wholesale trade—nondurable goods Building materials and gardening General merchandise stores Food stores Auto dealers and service stations Apparel and accessory stores Furniture and home furnishings Eating and drinking places Miscellaneous retail Depository institutions Nondepository institutions Security and commodity brokers Insurance carriers Insurance agents, brokers, services Real estate Holding and other investment offices Hotels and other lodging places Personal services Business services Auto repair, services, and parking Motion pictures Amusement and recreation services Health services Educational services Engineering and management services Nonclassifiable establishments Total
Non-family firm obs. 3 109 10 17 17 66 29 8 12 12 81 51 302 57 43 47 68 186 164 134 144 23 44 7 0 10 68 314 8 40 13 57 33 7 8 20 14 33 277 55 51 192 11 0 3 0 0 161 11 3 12 33 0 6 33 3,110 obs.
Note: Capital intensity is calculated as the median of PPE/ assets.
Family firm obs. (family firm1) 10 45 0 11 0 113 0 31 26 8 19 60 85 12 26 20 12 88 186 31 80 12 0 0 10 16 66 31 14 13 8 58 22 10 36 19 31 42 77 18 36 72 12 8 22 31 7 171 0 3 0 13 8 17 0 1,746 obs.
Family firm obs. (in %) 77 29 0 39 0 63 0 80 68 40 19 54 22 17 38 30 15 32 53 19 36 34 0 0 100 62 49 9 64 25 38 50 40 59 82 49 69 56 22 25 41 27 52 100 88 100 100 52 0 50 0 28 100 74 0 36%
Capital intensity (in %) 70 69 57 7 26 31 21 18 56 32 47 25 29 65 28 43 26 17 21 24 18 12 83 55 80 68 47 67 20 20 58 42 46 45 39 23 72 24 1 0 0 0 9 n/a 0 45 7 10 67 12 67 50 14 14 12 25%
68
4 Data and method
Method: Bayesian analysis126
4.3
This thesis uses Bayesian methods in addition to classical null hypothesis testing (hereafter classical analysis) to perform the statistical analyses. As Bayesian analysis is less well known than classical analysis, this section introduces the reader to Bayesian methods. After that, I compare Bayesian to classical analysis (Section 4.3.2), give a motivation of its use (Section 4.3.3) and describe the specific model estimated (Section 4.3.4). 4.3.1 Introduction to Bayesian analysis Bayesian methods rely on Bayes’ theorem of probability theory (Bayes, 1763). This theorem is given by Pr(T | y )
Pr( y | T ) Pr(T ) , Pr( y )
(4-1)
where T represents the set of unknown parameters and y represents the data. Pr(T ) is the prior
distribution of the parameter set T 127, which may be derived from theory, expert opinion, or other external resources. Pr( y | T ) is the likelihood function, which is the probability of the data y given the unknown parameter set T . Pr(y) is the marginal distribution of the data y, and finally, Pr(T | y ) represents the posterior distribution, which is the probability of the parameter set T given the data
y. Equation (4-1) may also be written as Pr(T | y ) v Pr( y | T ) Pr(T ) ,
(4-2)
where v means ‘proportional to’. The posterior distribution is proportional to the likelihood function multiplied by the prior distribution. In Bayesian analysis, inference comes from the posterior distribution, which states the likelihood of a particular parameter value. This way, Bayesian methods are useful for testing theory. When testing a hypothesized relationship between two variables, Bayesian analysis proceeds in the following steps. First, a priori beliefs (from theory or an interview) about the relationship of interest are formulated (the prior distribution, Pr(T ) ). Next, a probability of occurrence of the data given these a priori beliefs is assumed (the likelihood function, Pr( y | T ) ). In a third step, data are
126 127
This section is based on Block and Thams (2007a, 2007b). This distribution is often referred to as prior.
4 Data and method
69
used to update these beliefs. The result is the posterior distribution, Pr(T | y ) . This posterior distribution gives a density function of the parameter of interest (i.e., the coefficient that describes the relationship between the two variables). The posterior distribution allows for statements in terms of likely and unlikely parameter values. 4.3.2
Differences between Bayesian and classical analysis
Most researchers in economics and management are trained to use classical econometric techniques. In most cases, classical methods rely on null hypothesis significance testing.128 Null hypothesis testing has recently been criticized for various reasons (e.g., Cohen, 1994; Schmidt, 1996; Starbuck, 2006).129 One criticism is that scientific journals almost never publish negative (not statistically significant) results and thereby present a biased picture of the reality. Furthermore, the statistical significance required for publication (in most cases 5%) is arbitrary and has no mathematical basis, but it is based on long tradition. Another, more pedagogical criticism concerns the fact that overstressing statistical significance draws attention away from the size of the effect (economic significance). Note that, in practice, a statistically significant result can almost always be found if a large enough sample is analyzed.130 At the 2008 Academy of Management Meeting in Anaheim, Bill Starbuck, a former president of the Academy and former editor of various prestigious journals such as Administrative Science Quarterly (ASQ), organized a professional development workshop making the case against null hypothesis testing. The purpose of the workshop was to highlight the inherent problems with null hypothesis significance testing and to introduce alternative techniques that have become standard in other disciplines, such as psychology (e.g., estimating confidence intervals for effect sizes). Bayesian methods, used widely in the biological (e.g., Woodworth, 2004) and medical sciences (e.g., Goodman, 2005), are another alternative to null hypothesis testing. To understand the differences and similarities between the two methods, it is important to know about the basics of null hypothesis testing.131
128 129
130
131
Estimating confidence intervals for effect sizes is an alternative (Cumming and Finch, 2001; Thompson, 2007). The criticism has been most intense in the field of psychology. The American Psychological Association reacted and enhanced their publication guidelines, reducing the relative importance of such testing. The reporting of confidence intervals for effect sizes has become standard. By definition of statistical significance, one in 10 studies will produce a significant result (p<10%, two-sided test). If this study is then published and the other studies not, the uninformed reader will conclude that there exists a significant relationship in the population. The following two paragraphs are based on Lancaster (2005, pp. 359-364).
70
4 Data and method
The classical null hypothesis testing approach defines a population (e.g., all students in Germany) and draws a sample from this population (e.g., all students at Technische Universität München) to learn about the value of a particular parameter in the population (e.g., mean age). The view is that a parameter varies over the population. A sample taken from the population will yield a particular parameter value, and another sample will yield another value – which is referred to as sampling variation. The statistician’s task is to arrive at the ‘true’ parameters (in the population) using the evidence provided by the sample. To achieve this, a sample estimator and an accompanying test statistic are selected. The sample estimator would be for example “the average student age in the sample.” An appropriate test statistic would then be t ( x P 0 ) /(
s n
)
(independent one sample
t-test), where x is the sample estimator, 0 is a specific parameter in the population (e.g., 25 yrs), s is the standard deviation, and n is the sample size. The test statistic is known to follow a particular distribution (in the case above: the student’s t distribution). A comparison of the value of the test statistic and its distribution then leads to a p-value, which is the probability that the hypothesis “the true parameter in the population equals 0” can be rejected.132 To summarize, the classical approach is tied to the notion of a sample and a population. It uses sample estimators and test statistics to learn something about the ‘true’ parameters in the population. The Bayesian approach is different. The main differences are described below.
Posterior distribution. As explained above, the result of Bayesian analysis is a posterior
distribution of the parameter of interest. This differs from classical econometrics in that the outcome of the estimation is not a point estimate – which is either statistically significant or not – but an entire distribution function. This way, Bayesian analysis allows for statements such as “the probability of a positive effect of A on B is 70%”, which is not possible with classical analysis. The classical approach only permits a statement such as “the effect of A on B is positive. The probability of making an error with this statement is below 5% (10%).”
Notion of sample and population. Bayesian analysis is not tied to the notion of a sample
and a population. The results of Bayesian analysis is a statement about the particular data that is used in the analysis. There is no statement about a ‘true’ parameter in an underlying distribution. When Bayesians refer to their data as a sample, this is just due to convention.
132
Depending on the discipline and the size of the sample, p-values less than 5% (1%) are generally considered satisfactory. The corresponding estimators would be statistically significant.
4 Data and method
71
Prior and likelihood function. A Bayesian133 must attach probabilities to the values of
the data observed. As explained above, this is typically done in three stages. First, the Bayesian assumes (based on theory or other sources) a prior distribution Pr(T ) . Then she assumes a likelihood function Pr( y | T ) , the probability of the data given the prior beliefs. These two distributions are then used to calculate the posterior distribution Pr(T | y ) , the result of Bayesian analysis. The Bayesian approach is thus more demanding than the classical approach in the pre-estimation phase, as specific assumptions with regard to the likelihood function and the prior distribution are required. Yet, this investment pays off since the posterior distribution permits inferences that are ‘exact’. Contrary to classical econometrics, Bayesian analysis requires no assumptions regarding asymptotic properties of estimators or test statistics. The Gauss-Markov theorem134, the foundation of the classical approach, is irrelevant. Consequently, any violations of the theorem (e.g., problems of multicollinearity or heteroscedasticity) are also less problematic than in classical analysis (Leamer, 1973). Table 4-6 summarizes the main differences between Bayesian and classical analysis.
133
134
People who follow Bayesian methods are often called Bayesians. People who follow classical methods are referred to as ‘frequentists’. The theorem states that the least squares estimator is the best linear unbiased estimator (BLUE) if the errors have an expectation of zero, are uncorrelated, and have equal variances.
72
4 Data and method
Table 4-6: Differences between classical and Bayesian analysis Classical analysis
Bayesian analysis
A sample is taken from a population to learn about The notion of a sample and a population does not parameters in the population. exist. The results of the estimation apply only to the specific data used.
4.3.3
The result of the estimation is a point estimate, which is then either statistically significant or not.
The result of the estimation is an entire distribution function of the parameter of interest.
Large-sample approximations may not hold for small or skewed samples.
Distributional results are exact under skewed and small samples. No inference to a pre-defined population is made.
Violations of the Gauss-Markov theorem (e.g., multicollinearity or heteroscedasticity) may lead to biased estimates.
Gauss-Markov theorem is irrelevant. Violations do not lead to biased estimates.
A prior assumption about the significance level is required, typically 10%, 5%, or 1%.
No assumption with regard to significance level is required.
No prior assumptions about the functional form of the parameters and their likelihood are needed.
Prior assumptions about the functional form of parameters and their likelihood are necessary.
Classical analysis does not allow a researcher to measure evidence that there is no relation between two variables (the null hypothesis of ß=0 cannot be accepted).
Through the posterior distribution, Bayesian analysis can provide evidence that there is no relation between two variables (Fryar et al., 1988).
Motivation to use Bayesian methods
Bayesian methods have become increasingly prevalent in econometric analysis in recent years.135 First applied in macroeconomics and decision theory, Bayesian methods have found their way to other social sciences, particularly with applications in marketing research.136 But only few Bayesian studies exist in management research. A rare example is Hansen et al. (2004), who used a Bayesian approach to operationalize the resource-based view.137 In this thesis, Bayesian methods are used in addition to classical methods for the reasons summarized below. First, Bayesian methods offer interpretations which are more intuitive and consistent with theory. Contrary to classical methods, which assume that there are some ‘true’ and non-stochastic coefficients, Bayesian methods give a region that contains the corresponding coefficient with a certain probability. This is useful because models in social sciences are usually not as omnipotent as those in natural sciences. A specific theory may not be a valid description of all data observations, and 135
136 137
See The Economist (2000, 2006a) for a discussion of Bayesian methods. I thank Tobias Krebs for providing me with these references. Rossi and Allenby (2003) discuss the potentials of Bayesian statistics in marketing science. Hahn and Doh (2006) discuss the potentials for Bayesian methods in strategy research.
4 Data and method
73
Bayesian analysis states the probability that the particular theory describes the data. In this thesis, theory leads to competing hypotheses.138 A Bayesian approach allows to state which theory is more likely, instead of rejecting one (or both) of the two hypotheses as being irrelevant. Second, Bayesian methods have strong small sample properties and are robust to multicollinearity problems (Hahn and Doh, 2006; Leamer, 1973). The reason is that Bayesian methods do not rely on asymptotic theory and use prior knowledge about the coefficients in the distribution of the prior. Thus, if the sample is small (which is the case if I restrict my analysis to a particular group of family firms only) and very little is known beforehand about the data-generating process behind it, this uncertainty may be expressed via a distribution of the prior with a large variance. The result of the Bayesian estimation, the posterior distribution, reveals which parameter values are more likely.139 In my dataset, the variable family firm is strongly related to other variables, such as ownership share of financial investors, firm age, leverage, and firm size (see Table A-3 in the Appendix). With classical econometrics, I might run into problems of identifying the effect attributed to the variable family firm. Problems of multicollinearity might occur. Third, Bayesian analysis has the ability to determine that no relationship between two variables exists (ß=0), e.g., the situation that family firms are neither investing more nor investing fewer resources in R&D projects than non-family firms. Classical analysis does not allow such a statement. Two scenarios can occur when evaluating a null hypothesis of ß=0: (1) The data may support the alternative hypothesis, i.e. the null hypothesis is rejected (e.g., ß=0.2, p<0.05 or ß=-0.2, p<0.05), (2) the data may support neither the null hypothesis nor the alternative hypothesis (e.g., ß=-0.2, p>0.1 or ß=0.2, p>0.1). Note that (2) does not imply that the null hypothesis is supported. Rather, the data may be inconclusive and support neither the null nor the alternative hypothesis (see Fryar et al., 1998). Finally, Bayesian methods serve as a useful robustness check. A comparison of the results of Bayesian and classical approaches allows me to be very cautious about the results and to provide additional evidence. Only the findings confirmed by both types of models are considered robust. This is particularly helpful in cases where the dataset is small and it is difficult to obtain precise estimates (Hahn and Doh, 2006).
138 139
See the theory section in Chapter 6. These advantages are not without a cost, however. The results of a Bayesian analysis relies on “subjective” a priori beliefs expressed in the prior distribution and the shape of the likelihood function (Rossi and Allenby, 2003). Therefore, where these beliefs come from and how they influence the results must be made clear. The usual way is estimating several models with different assumptions about prior distributions and likelihood functions and comparing the results.
74
4.3.4
4 Data and method
Description of the Bayesian model estimated
For the prior distribution of the individual effects, I choose either a uniform distribution or a normal distribution. A uniform or flat distribution implies that the regression model does not include an individual specific error term, which corresponds to a fixed-effects model specification in classical econometrics. A normal distribution implies that the regression model includes a normally distributed, time-independent error term, which corresponds to a random-effects model specification in classical econometrics. For the coefficients, I assume a normally distributed prior with a mean of zero for all coefficients. Such a prior specification implies that the model has no explanatory power at all, which fits well with the fact that in this thesis theory leads to competing hypotheses.140 This ensures that any evidence for the validity of one particular theory (against the other theory) is not induced by the specification of the prior. Furthermore, this particular prior specification ensures that the posterior distributions functions are identifiable, i.e., that a flat or uniform posterior distribution does not result (Koop, 2003, p. 291).141 Finally, the prior of the variance is assumed to follow a 2 distribution, which is consistent with the assumption of normally distributed error terms. As a robustness check, I also estimate the models with different prior specifications.142 I allow for different means and variances in the normal distribution and for different classes of distributions, such as a uniform distribution. The results of these alternative specifications are reported in the notes accompanying the tables. In most cases, the posterior distribution Pr(T | y ) is multidimensional, i.e. it encompasses several parameters. However, the object of interest is often to learn about the specific distribution of one particular parameter. Usually, these specific distributions cannot be deduced analytically by using e.g. methods of numerical integration. To this end, a simulation approach is often used. A computer does the work and provides simulated samples from the required distribution. With such a sample from the posterior, it is possible to find any feature of the joint distribution. The higher the sample size, the more accurate the estimates become. For the estimation and the simulation, I use a MatlabTM code, which takes 10,000 draws from the (joint) posterior distribution.143 Taking 10,000 draws has emerged as a standard in the Bayesian literature (Lancaster, 2005). I apply Markov Chain
140 141 142
143
See the theory section in Chapter 6. A flat posterior distribution implies that every parameter value is equally likely. Usually, I assume the opposite of what theory would lead me to expect. If the Bayesian estimation then updates these beliefs in line with theory, the result is considered robust. One might also use the software package WinBUGS, which is freely available from the internet website of the Medical Research Council at University of Cambridge. See http://www.mrc-bsu.cam.ac.uk/bugs (accessed September 13th, 2008). Lancaster (2005) provides a number of useful examples using WinBUGS.
4 Data and method
75
Monte Carlo Techniques (MCMC) and a Gibbs Sampler to arrive at the corresponding univariate distributions of the coefficients.144 As usual, the first 1,000 draws are discarded.145 The MatlabTM code is included in the Appendix (see Figure A-1). After having described Bayesian analysis in detail, I apply the method in the next chapter analyzing the performance of family firms.
144
145
See Lancaster (2005, pp. 183-226) to learn more about Markov Chain Monte Carlo Techniques (MCMC) and the Gibbs Sampler. The quality of the sample derived from a Markov chain improves as a function of the steps, i.e. the draws taken. See Robert and Casella (2004).
5 Performance of family firms This chapter analyzes the performance of family firms, with a focus on return on assets (ROA) and market-to-book value (MTB).146 After a literature review (Section 5.1), I compare the performance of family and non-family firms using univariate and multivariate analysis (Section 5.2). After that, I estimate a multivariate model that determines the effect of the degree of family ownership and family management, two family firm characteristics, on firm performance (Section 5.3). Another model is estimated using the family firm definitions determined in Chapter 2. The multivariate models are estimated both with classical and Bayesian methods. Finally, Section 5.4 provides a summary of the main empirical results and discusses the contribution to the literature.
5.1
Literature review A substantial body of literature in the fields of management, economics, and finance has fo-
cused on the performance differences between family and non-family firms (e.g., Anderson and Reeb, 2003; Andres, 2008; Ang et al., 2000; Bennedsen et al., 2007; Cronqvist and Nilsson, 2003; Holderness and Sheehan, 1988; Jaskiewicz, 2006; Kaserer and Moldenhauer, 2008, Maury, 2006; Miller et al., 2007; Villalonga and Amit, 2006). The evidence is mixed.147 Using data on large US firms, Anderson and Reeb (2003), McConaughy et al. (1998) and Villalonga and Amit (2006) report that the market-to-book value of family firms is higher than that of non-family firms. This finding contradicts evidence from Europe (Barontini and Caprio, 2006; Bennedsen et al., 2007; Cronqvist and Nilsson, 2003; Maury, 2006; Perez-Gonzalez, 2006) and Asia (Claessens et al., 2002). In a more recent study, Miller et al. (2007) find that the superior performance of family firms relative to non-family firms can be attributed to a large degree to lone-founder firms, in which no family of the founder is involved. To summarize, family firms do not per se exhibit superior performance. Rather, the findings are very sensitive to the definition of a family firm (Andres, 2008; Miller et al., 2007). The definition of family firm that is used influences the empirical results.
146
147
The most commonly used performance measures to analyze the relationship between ownership and performance are Tobin’s q (24%), return on assets (19%), return on equity (19%), and return on investment (8%). See Jaskiewicz (2006, p. 45). For a detailed review of studies on the performance of family firms, see Astrachan and Zellweger (2008) and Jaskiewicz (2006, pp. 41-86).
5 Performance of family firms
77
Table 5-1 summarizes the main theoretical arguments in favor of and against a superior performance of family firms relative to other types of firms.148 Table 5-1: Arguments regarding superior performance of family firms Arguments in favor of a superior performance of family firms relative to other firms
5.2
Arguments against a superior performance of family firms relative to other firms
Alignment of interests between family owners and family management, which leads to lower agency costs than in a situation in which ownership and management are separated (e.g., Fama and Jensen, 1983; Jensen and Meckling, 1976)
Limited human resource pool of the owning family (e.g., Bennedsen et al., 2007; Burkart et al., 2003; Perez-Gonzales, 2003)
Conflicts within the family (e.g., Eddleston and Kellermanns, 2007; Harvey and Evans, 1994; Hennerkes, 2004, pp. 58-66)
Effective monitoring due to concentrated ownership (Fama, 1980; Maug, 1998)
Altruism towards members of the owning family (e.g., Schulze et al., 2001, 2003)
Alignment of interests between family owners and lenders, which leads to low agency costs of debt (e.g., Anderson et al., 2003)
Entrenchment of minority shareholders (e.g., Claessens et al., 2002; Gomez-Mejia et al., 2001; Morck and Yeung, 2003)
Strong long-term orientation due to intended transfer of ownership to the next family generation (e.g., Berghoff, 2006; Bertrand and Schoar, 2006; Böttcher and Linnemann, 2008; James, 1999; Miller and Le Breton-Miller, 2005; Simon et al., 2005)
Capital restrictions (e.g., Landes, 1949)
Quick and unbureaucratic decisions (e.g., Schein, 1983)
High degree of trustworthiness (e.g., Dyer and Whetten, 2006)
Family firms versus non-family firms To compare the performance of family and non-family firms, I use the sample as described in
Section 4.1, but, in line with other studies, exclude observations from financial institutions (SIC 6069)149 and public utilities (SIC 46, 48, and 49)150 (see also Anderson and Reeb, 2003; Jaskiewicz 2006, p. 143). To begin with, I compare market-to-book value (MTB)151, return on assets (ROA)152,
148
149
150 151
Some arguments refer to the discussion as to whether family firms have higher or lower agency costs than nonfamily firms. See Section 2.3.1 or Chrisman et al. (2004). It is difficult to calculate the market-to-book value for financial institutions. It requires a lot of adjustments to calculate the book value of financial institutions (Rezaee, 2001). Government regulations may affect the firm performance of public utilities. Market-to-book value is constructed as the sum of market value of equity and book value of debt divided by book value of total assets.
78
5 Performance of family firms
return on equity (ROE)153 and the dividend yield154 of family and non-family firms using descriptive statistics. 5.2.1 Univariate analysis
Table 5-2a compares the performance of family and non-family firms using univariate analysis. The groups of family firms are defined according to the definitions in Table 2-2. The group of non-family firms does not change with the family firm definitions used. The variables are lagged by one year to avoid the widely discussed problem of endogeneity in the relationship between ownership and performance.155 A Wilcoxon rank-sum test is used to find out whether the median performance of non-family firms differs from the median performance of family firms (using the broadest definition family firm1). The results are as follows: Family firms have a higher market-to-book value than non-family firms (2.0 vs. 1.5, p<0.01). The difference increases as the group of family firms is more narrowly defined. A similar pattern can be observed with ROA. The median ROA of family firms is higher than the one of non-family firms (7.4% vs. 5.9%, p<0.01). Again, the performance differences grow when a narrow family firm definition is used. Interestingly, another pattern emerges. With both ROA and the market-to-book value, the variance in the performance is larger with family firms than with non-family firms – even though the number of family firm observations is considerably smaller. The pattern is different concerning ROE. Family firms seem to have a slightly lower ROE than non-family firms (median 15.1% vs. 16.2%, p<0.05), which can be explained by the lower levels of debt in family firms (see Table 4-4). Yet, ROE differences between family and non-family firms are sensitive to the way family firms are defined. When the group of family firms is defined more narrowly (e.g., family firm4), family firms seem to have a slightly larger ROE than non-family firms (median 17.4% vs. 16.2%, p=0.20)156.
152 153 154
155
156
ROA is defined as income before extraordinary items divided by total assets (see Table A-2). ROE is defined as income before extraordinary items divided by common equity (see Table A-2). Dividend yield is constructed as the annual dividend per share divided by the company’s close price for the fiscal year. This figure is then multiplied by 100 (see Table A-2). Demsetz (1983) as well as Demsetz and Lehn (1985) argue that the structure of corporate ownership is determined by corporate performance and not the other way around. They posit that, under the assumption of perfect markets, rational investors, and complete contracts, the ownership structure is irrelevant for corporate performance. I believe this argument to be not important in the case of the specific dataset used it this thesis. Table 4-1 shows that only relatively few firms change their status from family firm to non-family firm or the other way round. Note that a two-sided test is used.
5 Performance of family firms
79
Finally, there are large differences in the dividend yields between family and non-family firms. Family firms have a much lower dividend yield than non-family firms (median 0.3% vs. 1.5%, p<0.01). This reflects the fact that family owners may prefer to reinvest profits in the firm instead of receiving a dividend; it is unlikely to be an indicator of weak performance.157 Supporting this argument, the median dividend yield of young family firms (family firm5a) is 0. Young, founder-managed firms have a particular need for internal financing to sustain their growth (Harhoff, 1998; Himmelberg and Peterson, 1994). Table 5-2a: Financial performance in family and non-family firms (group of non-family firms remains fixed) Non-family Family Family Family Family firm4 Family firm t-1 firm1 t-1 firm2 t-1 firm3 t-1 Firm5a t-1 t-1 (1.931 obs.) (1,288 obs.) (1,219 obs.) (402 obs.) (294 obs.) (182 obs.)
Family WilcoxonFirm5b t-1 rank-sum (112 obs.) test 1
Median (Std. dev.)
Median (Std. dev.)
Median (Std. dev.)
Median (Std. dev.)
Median (Std. dev.)
Median (Std. dev.)
Median (Std. dev.)
Marketto-book value
1.46 (1.98)
1.96 (3.63)
1.95 (3.71)
2.20 (4.88)
2.38 (2.53)
2.85 (2.89)
2.12 (1.58)
p<0.001
ROA (in %)
5.94 (8.37)
7.43 (18.99)
7.33 (19.49)
8.00 (10.82)
8.47 (11.90)
8.47 (14.50)
8.42 (5.58)
p<0.001
ROE (in %)
16.16 (53.06)
15.10 (45.75)
14.98 (46.98)
15.53 (18.87)
17.36 (21.06)
17.64 (24.95)
16.51 (12.45)
p=0.013
Dividend yield (in %)
1.48 (1.92)
0.34 (1.28)
0.29 (1.26)
0.19 (1.19)
0.28 (1.23)
0 (0.97)
1.12 (1.40)
p<0.001
Notes: The family firm definitions are not mutually exclusive. Family firm1 versus non-family firm; two-sided test is used; the Wilcoxon rank-sum test analyzes whether the two samples are from different distributions (sample1: family firm1, sample 2: non-family firms). N=3,219 ROA is calculated using income before extraordinary assets in the numerator. 1
157
The ROA would otherwise be lower, but this is not the case (Table 5-2a). The result that family firms pay less in dividends than non-family firms is also in line with the findings of other studies. See, for example, Gugler (2003), who finds that family-controlled firms choose significantly lower target payout levels.
80
5 Performance of family firms
Table 5-2b compares the financial performance of family and non-family firms. In contrast to Table 5-2a above, the group of non-family firm observations does not remain the same over the different family firm definitions used but is adjusted to the specific family firm definition used (family firm1 to family firm4). As expected from the results above (Table 5-2a), the median marketto-book value and the median ROA of the group of non-family firms increases slightly when a narrower family firm definition is used. Still, there are strong differences between family and nonfamily firms in terms of financial performance. The differences become larger when a narrow family firm definition is used (e.g., family firm3 or family firm4). Table 5-2b: Financial performance in family and non-family firms (group of non-family firms changes with the family firm definitions used) Definition used: family firm1
Definition used: family firm2
Definition used: family firm3
Family Non-family Family Non-family Family Non-family firm1 t-1 firm t-1 firm2 t-1 firm t-1 firm3 t-1 firm t-1 (1.931 obs.) (1,288 obs.) (2,000 obs.) (1,219 obs.) (2,817 obs.) (402 obs.)
Definition used: family firm4 Non-family firm t-1 (2,925 obs.)
Family firm4 t-1 (294 obs.)
Median (Std. dev.)
Median (Std. dev.)
Median (Std. dev.)
Median (Std. dev.)
Median (Std. dev.)
Median (Std. dev.)
Median (Std. dev.)
Median (Std. dev.)
Marketto-book value
1.46 (1.98)
1.96 (3.63)
1.47 (1.97)
1.95 (3.71)
1.58 (2.33)
2.20 (4.88)
1.58 (2.80)
2.38 (2.53)
ROA (in %)
5.94 (8.37)
7.43 (18.99)
6.02 (8.27)
7.33 (19.49)
6.34 (13.99)
8.00 (10.82)
6.33 (13.80)
8.47 (11.90)
ROE (in %)
16.16 (53.06)
15.10 (45.75)
16.18 (52.15)
14.98 (46.98)
15.75 (53.37)
15.53 (18.87)
15.59 (52.41)
17.36 (21.06)
Dividend yield (in %)
1.48 (1.92)
0.34 (1.28)
1.47 (1.90)
0.29 (1.26)
1.19 (1.78)
0.19 (1.19)
1.12 (1.76)
0.28 (1.23)
Notes: The family firm definitions are not mutually exclusive. Except for the case of ROE with definition family firm3, a two-sided Wilcoxon rank-sum test is significant at p<0.05. N=3,219 ROA is calculated using income before extraordinary assets in the numerator.
5 Performance of family firms
81
5.2.2 Multivariate analysis
The multivariate analysis proceeds as follows. Random-effects generalized least squares (GLS) models are estimated to analyze the performance of family and non-family firms. The dependent variables are ROA (Table 5-3) and the market-to-book value (Table 5-4). As independent variables, I use the family firm dummies (family firm1-5) in different models and the usual control variables such as firm size, firm age, leverage, firm risk, CEO-specific variables as well as industry and year dummies (Anderson et al., 2003; Miller et al., 2007; Villalonga and Amit, 2006).158 To determine whether a random-effects GLS model should be preferred to a pooled ordinary least squares (OLS) model, I calculated the Breusch-Pagan Lagrangian Multiplier Test, which examines whether the firm-specific intercepts differ from one another. If the test is significant, a randomeffects GLS model should be preferred to a pooled OLS model (Breusch and Pagan, 1980).159 To avoid endogeneity problems, all independent variables except firm age and the year and industry dummies are lagged by one year. Table A-3 in the Appendix displays a correlation table. Family firms (as defined by family firm1) are younger (r=-0.36, p<0.01), smaller (r=-0.19, p<0.01), have a lower level of debt (r=-0.21, p<0.01), and have a lower share of equity owned by financial investors (r=-0.17, p<0.01). CEOs in family firms are less likely to serve also as chairman of the board of directors (r=-0.22, p<0.01) and are more likely to serve longer tenures (r=0.30, p<0.01). The correlation between family ownership and family management, which are the two distinct characteristics of family firms, is (only) r=0.30 (p<0.01). This indicates the large heterogeneity within the group of family firms. It is not the case that every family-owned firm is also managed by a family member. Return on assets
Table 5-3 presents the results of random-effects GLS regressions on ROA160. I estimate five models, which differ by the family firm definition used. In each of the five models, the BreuschPagan test returns a significant result, indicating that a random-effects model is preferred to a pooled OLS model. As the result of the Wald Chi2-test shows, all models are statistically signifi-
158 159
160
For the exact definition of the variables, refer to Table A-2 in the Appendix. Typically, in a next step, a Hausman (1978) specification test is applied to determine whether a random-effects or a fixed-effects model is more appropriate. However, since the family firm dummy variables (family firm1 - family firm5a) are, to a large degree, time-invariant, and since the industry variables are completely time-invariant, I was not able to estimate a (interpretable) fixed-effects model. The Hausman test is therefore inappropriate. ROA is calculated as income before extraordinary items divided by total assets. Other studies have also used net income or earnings before interest, taxes, depreciation, and amortization (EBITDA).
82
5 Performance of family firms
cant.161 Yet, the R2-values differ strongly across the models. The prediction quality of the model increases strongly when a narrow family firm definition is used (e.g., R² of 11% in model I vs. R² of 42% in model III)162. Obviously, there is large heterogeneity within the group of firms that fall under the broad family firm definition. The effect of the family firm variable differs depending on the family firm definition used. With a broad definition (e.g., family firm1), the effect of the family firm dummy is insignificant (see model I). Yet, with a narrow family firm definition, the effect of the variable is positive and weakly significant (see models III-V). Some of the control variables have a significant impact on the firm’s ROA. In models I and II, the variable ownership share of financial investors has a positive impact (e.g., model I: ß=0.69, p<0.1). Interestingly, this effect becomes weaker and insignificant when a narrow family firm definition is used (see models III-V). The variable firm size has a negative impact on ROA in all five models (e.g., model I: ß=-7.06, p<0.05). Also, the variable CEO duality has a negative effect in all models (e.g., model I: ß=-2.36, p<0.1). Firms in which the CEO serves also as chairman of the board of directors have a two percentage points lower ROA than firms in which the roles are separated. The firm’s level of debt, research intensity and age as well as CEO duration and the structure of CEO pay do not have a significant impact on the firm’s ROA. Industry and time dummies, however, have a significant impact; with both industry and time dummies, an F-test on joint significance returns highly significant p-values. Finally, note that in the models II-V, I decided to exclude those observations that fall under the broad family firm definition (family firm1) but that do not classify as a family firm using the respective narrower definitions. This is why the number of observations differs across the models. This exclusion of observations has a great impact on the results. For example, the family firm dummy in model III (family firm 3) becomes insignificant when including also the observations that fall under the broad family firm definition (family firm1), but do not fall under the narrower definition (family firm 3).
161 162
Hence, at least one independent variable has an effect which is statistically different from zero. These numbers are R²-between values.
5 Performance of family firms
83
Table 5-3: Random-effects GLS regressions of return on assets
Independent variables Family firm1 t-1
Model I
Model II
Model III
Model IV
Model V
Coefficient (SE)
Coefficient (SE)
Coefficient (SE)
Coefficient (SE)
Coefficient (SE)
-3.71 (4.30)
Family firm2 t-1
-3.97 (4.57)
Family firm3 t-1
3.00* (1.80)
Family firm4 t-1
3.88* (2.19)
Family firm5a t-1
5.22* (2.74)
Ownership share of financial investors t-1
6.90* (4.06)
6.92* (4.07)
5.85 (5.03)
5.96 (5.64)
6.38 (5.76)
Firm age
2.36 (3.40)
2.36 (3.38)
-1.37 (0.98)
1.16 (1.04)
-0.83 (1.08)
Firm size t-1
-7.06** (3.36)
-7.20** (3.40)
-2.33*** (0.80)
-2.31*** (0.81)
-2.49*** (0.81)
Leverage t-1
0.05 (0.08)
0.05 (0.08)
-0.03 (0.04)
-0.04 (0.04)
-0.03 (0.04)
CEO t-1
-2.36* (1.32)
-2.63* (1.38)
-2.16*** (0.78)
-2.01*** (0.80)
-2.05** (0.86)
CEO duration t-1
0.08 (0.08)
0.10 (0.08)
0.01 (0.06)
0.03 (0.06)
0.02 (0.07)
Share of option payment t-1
-1.61 (1.40)
1.71 (1.51)
<0.01 (1.34)
0.80 (1.39)
1.05 (1.48)
Share of stock payment t-1
0.42 (1.52)
0.28 (1.63)
-1.11 (1.62)
-0.88 (1.67)
-0.83 (1.71)
R&D/sales t-1
0.50 (0.37)
0.50 (0.38)
-0.18 (0.17)
-0.18 (0.17)
-0.18 (0.17)
Diversified risk
1.83 (1.48)
1.82 (1.49)
0.16 (0.32)
0.18 (0.33)
0.22 (0.34)
Undiversified risk
-0.03 (0.02)
-0.03* (0.02)
-0.03*** (0.01)
-0.03*** (0.01)
-0.03*** (0.01)
Industry dummies (35 categories)
p<0.001
p<0.001
p<0.001
p<0.001
p<0.001
Year dummies (9 categories)
p<0.001
p<0.001
p<0.001
p<0.001
p<0.001
N obs. (firms) 1,586 (243) 1,552 (241) 1,148 (200) 1,091 (193) 1,043 (183) Obs. per group: min./avg./max. 1; 6.5; 10 1; 6.4; 10 1; 5.7; 10 1; 5.7; 10 1; 5.7; 10 R² within, R² between, R² overall 0.18; 0.11; 0.10 0.19; 0.11; 0.10 0.11; 0.42; 0.22 0.11; 0.42; 0.22 0.11; 0.42; 0.22 p-value Chi²-test p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 p-value Breusch-Pagan test p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 0.40 0.40 0.36 0.36 0.37 Rho (fraction of variance due to ui) Notes: SE=robust standard errors; * p0.1 ** p0.05 *** p0.01; two-sided tests are used. ROA is calculated using income before extraordinary assets in the numerator.
84
5 Performance of family firms
Market-to-book value
Table 5-4 gives the results of random-effects GLS regressions on market-to-book value (MTB), which is defined as the sum of market value of equity and book value of debt divided by book value of total assets. Again, in each of the five models, the Breusch-Pagan test returns a significant result, indicating that a random-effects model is preferred to a pooled OLS model. All models are statistically significant. The R2-values are slightly higher than those of the models using ROA as a dependent variable (R² ranges from 29% in model III to 47% in model IV)163. Although the models seem to have a slightly greater prediction power, the effects of the respective variables are similar. The family firm dummy is insignificant in models I and II but has a significant positive effect in the models III-V (model III: ß=1.49, p<0.05). The only family-owned firms that outperform other firms are those in which the family owner is also active in the management. Regarding the control variables, I find that both firm size and firm age have a negative impact. A higher level of debt also has a negative effect: An increase of leverage by 10 percentage points leads to a reduction in market-to-book value of 0.2 (p<0.01, model I). CEO duality (where the CEO serves also as chairman) has a negative impact on the market-to-book value (model I: ß=-1.08, p<0.01). A higher share of option payment leads to an increase in the market-to-book value (model I: ß=0.78, p<0.05). This is surprising because option payment is not found to have an impact on the firm’s ROA (see Table 5-3). Hence, it seems that option payment leads managers to increase the firm’s market value but fails in delivering a higher profitability. Finally, F-tests on joint significance show that industry and time dummy variables have a significant impact on the market-to-book value. In all models except model II, the variable ownership share of financial investors does not have a significant impact on the firm’s market value. CEO duration and research intensity are also not correlated with the firm’s market-to-book value. As with the regression of ROA, note that in the models II-V, I decided to exclude those observations that fall under the broad family firm definition (family firm1) but that do not classify as a family firm using the respective narrower definitions. This exclusion of observations has a great impact on the results. Again, most family firm dummies become insignificant when including all observations.
163
These numbers are R²-between values.
5 Performance of family firms
85
Table 5-4: Random-effects GLS regressions of the firm’s market-to-book value Model I
Model II
Model III
Model IV
Model V
Independent variables
Coefficient (SE)
Coefficient (SE)
Coefficient (SE)
Coefficient (SE)
Coefficient (SE)
Family firm1 t-1
0.32 (0.25)
Family firm2 t-1
0.52 (0.37)
Family firm3 t-1
1.49** (0.75)
Family firm4 t-1
1.10*** (0.43)
Family firm5a t-1
1.43*** (0.56)
Ownership share of financial investors t-1
1.45 (1.18)
1.23* (0.75)
1.47 (1.02)
0.58 (0.73)
0.55 (0.72)
Firm age
-0.57** (0.24)
-0.56* (0.31)
-0.84** (0.39)
-0.62** (0.27)
-0.54** (0.28)
Firm size t-1
-0.68*** (0.19)
-0.95*** (0.18
-0.78*** (0.17)
-0.29*** (0.12)
-0.32*** (0.11)
Leverage t-1
-0.02*** (0.01)
-0.02*** (0.01)
-0.00* (0.01)
-0.02*** (0.01)
-0.02*** (0.01)
CEO duality t-1
-1.08*** (0.50)
-0.51** (0.23)
-0.42 (0.28)
-0.47* (0.27)
-0.48* (0.29)
CEO duration t-1
0.02* (0.01)
0.01 (0.01)
0.00 (0.01)
0.01 (0.01)
0.01 (0.02)
Share of option payment t-1
0.78** (0.35)
0.79** (0.34)
0.52* (0.28)
0.56** (0.27)
0.65** (0.29)
Share of stock payment t-1
0.92** (0.42)
0.59** (0.28)
0.36 (0.27)
0.41 (0.36)
0.45 (0.38)
R&D/sales t-1
0.03 (0.04)
0.01 (0.05)
-0.04 (0.04)
0.03 (0.03)
0.03 (0.03)
Diversified risk
0.14 (0.10)
0.25 (0.17)
0.27 (0.18)
0.03 (0.06)
0.03 (0.06)
<0.01 (<0.01)
<0.01 (<0.01)
<0.01 (<0.01)
<0.01 (<0.01)
<0.01 (<0.01)
Undiversified risk Industry dummies (35 categories)
p<0.001
p<0.001
p<0.001
p<0.001
p<0.001
Year dummies (9 categories)
p<0.001
p<0.001
p<0.001
p<0.001
p<0.001
N obs. (firms) Obs. per group: min./avg./max. R² within, R² between, R² overall p-value Chi²-test p-value Breusch-Pagan test Rho (fraction of variance due to ui)
1,571 (240) 1,537 (238) 1,134 (197) 1,080 (190) 1,032 (180) 1; 6.5; 10 1; 6.5; 10 1; 5.8; 10 1; 5.7; 10 1; 5.7; 10 0.14; 0.45; 0.28 0.15; 0.30; 0.26 0.10; 0.29; 0.23 0.10; 0.47; 0.32 0.10; 0.48; 0.32 p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 p<0.001 0.25 0.70 0.75 0.24 0.22
Notes: SE=robust standard errors; * p0.1 ** p0.05 *** p0.01; two-sided tests are used.
86
5 Performance of family firms
P (ß)
ß Prob. ß>0=83.8%; median ß=1.95; mean ß=1.92
ß
See Figure A-1 in the Appendix for the MatlabTM code used to estimate the models.
Notes: I use normally distributed priors with a mean of zero and a standard deviation of one. Number of draws: 10,000 (the first 1,000 draws are discarded). The estimation is based on a random-effects model with the independent variables displayed in Table 5-4. Using a prior of one leads to the following results: family firm1 (prob. ß>0=68.99%, median ß=0.20); family firm2 (prob. ß>0=96.16%, median ß=0.58) family firm3 (prob. ß>0=83.51%, median ß=1.19); family firm4 (prob. ß>0=83.84%, median ß=1.32); family firm5a (prob. ß>0=85.73%; median ß=1.40)
Prob. ß>0=75.2%; median ß=1.42; mean ß=1.48
P (ß)
Distribution curve of family firm5a (model V)
Figure 5-1: Bayesian regression of the firm’s market-to-book value (continued)
Distribution curve of family firm4 (model IV)
5 Performance of family firms 87
88
5 Performance of family firms
Figure 5-1 displays the results of a Bayesian analysis of the models I-V.164 The models are estimated with the same variables and data as the classical models.165 Unlike the classical approach, the Bayesian approach shows the distribution of the effect of a particular variable. Figure 5-1 displays the distributions of the coefficients of the respective family firm dummy in models I-V. The results are similar to those of the classical models. The effect of the family firm dummy differs strongly across the models. When a family firm is broadly defined, only a small difference is observed between family and non-family firms: The median effect is 0.09, and the likelihood that family firms exhibit a stronger performance than non-family firms is 59%. Yet, the pattern changes with a more narrow definition. Using definition family firm2, the likelihood of a positive effect is 95% (model II), and the median effect is 0.55. The effect is largest when the definition family firm5a is used (median effect ß=1.95, the probability of a positive effect is 84%). Recall that family firm5a corresponds to firms “in which the family owns more stock than financial investors do and the family is active in the management of the firm as either CEO or chairman and the firm is younger than 51 years” (Table 2-2).
164
165
A Bayesian model with ROA as the dependent variable has also been estimated. The results were not robust. That is, they were dependent on the choice of the prior and differed strongly from the results of the classical regressions. Therefore, I only show the model with market-to-book value as dependent variable. The Bayesian approach is explained in detail Section 4.3. The Matlab code used to estimate the model is included in the Appendix (Figure A-1).
5 Performance of family firms
5.3
89
Impact of family management and family ownership The second research question in this chapter concerns the differences in the effects of family
ownership and family management on firm performance. I determine whether family ownership or family management (or both) are drivers of superior performance. Table 5-5 shows the results of a random-effects GLS model estimated to answer this question. The dependent variable is the marketto-book value.166 The two variables of interest are family management167 and ownership share of family (in % of total shares), and the sample and the control variables are the same as in the regressions in Section 5.2. I find that family management has a significant positive effect on the firm’s market-to-book value (ß=0.97, p<0.05), whereas family ownership does not have a significant impact. These results are confirmed by Bayesian analysis. Figure 5-2 displays the distribution functions of the effects of the variables ownership share of family and family management. As with the classical regression, family management has a positive impact on the firm’s market-to-book value: the median effect is 1.82, and the probability of a positive effect is 100%. Contrary to that, the probability that a higher level of family ownership leads to superior performance is only 63%. I also estimated a model including an interaction term family management X ownership share of family (Model II). The interaction term turned out to be insignificant.
166
167
I have also estimated a model that analyzes the effect of family management and family ownership on ROA. Both family management and family ownership were found to have no statistically significant impact on the firm’s ROA (ß family management=-4.82, p=0.33; ß ownership by family=-1.01, p=0.91; two-sided tests are used). The variable is 1 if either the CEO or the Chairman is a member of the founding family and zero otherwise.
90
5 Performance of family firms
Table 5-5: Effect of family firm characteristics on the firm’s market-to-book value
Independent Variables
Model I
Model II
Coefficient (SE)
Coefficient (SE)
Family management t-1
0.97** (0.46)
0.69** (0.33)
Ownership share of family t-1
1.53 (1.97)
-3.84 (3.65)
Family management t-1 X ownership share of family t-1
7.62 (6.86)
Ownership share of financial investors t-1
1.68 (1.34)
1.75 (1.36)
Firm age
-0.46** (0.23)
-0.53** (0.23)
Firm size t-1
-0.66*** (0.18)
-0.64*** (0.16)
Leverage t-1
-0.18*** (0.01)
-0.02*** (0.01)
CEO duality t-1
-0.95*** (0.43)
-0.94** (0.42)
CEO duration t-1
0.01 (0.01)
0.01 (0.01)
Share of option payment t-1
0.83** (0.38)
0.84** (0.39)
Share of stock payment t-1
1.00** (0.45)
1.04** (0.46)
Research/ sales
3.35 (4.26)
3.42 (4.31)
0.12 (0.09)
0.12 (0.09)
<0.01 (<0.01)
<0.01 (<0.01)
t-1
Diversified risk Undiversified risk Industry dummies (35 categories)
p<0.001
p<0.001
Year dummies (9 categories)
p<0.001
p<0.001
N obs. (firms) Obs. per group: min./avg./max. R² within, R² between, R² overall p-value Chi²-test p-value Breusch-Pagan test Rho (fraction of variance due to ui)
1,563 (240) 1; 6.5; 10 0.15; 0.44; 0.28 p<0.001 p<0.001 0.25
1,563 (240) 1; 6.5; 10 0.16; 0.43; 0.28 p<0.001 p<0.001 0.26
Notes: SE=robust standard errors; * p0.1 ** p0.05 *** p0.01; two-sided tests are used.
5 Performance of family firms
91
Figure 5-2: Effect of family firm characteristics on the firm’s market-to-book value (Bayesian analysis) Effect of family management
P (ß)
ß Prob. ß>0=100% ; median ß=1.82 ; mean ß=1.83 Effect of ownership share of family
P (ß)
ß Prob. ß>0=63.21% ; median ß=0.68 ; mean ß=0.71 Notes: I use normally distributed priors with a mean of zero and a standard deviation of one. Number of draws: 10,000 (the first 1,000 draws are discarded); the estimation is based on a random-effects model with the independent variables as displayed in Table 5-5. Using a prior of one leads to the following results: family management (Prob. ß>0=100%, median ß=1.92), ownership share of family (Prob. ß>0=89.84%, median ß=1.98). See Figure A-1 in the Appendix for the MatlabTM code used to estimate the models.
92
5 Performance of family firms
5.4
Summary and conclusions This chapter deals with the following two questions: (1) Do family firms have a better finan-
cial performance than non-family firms? (2) What is the relation of the two dimensions of family management and family ownership with financial performance? Using ROA and market-to-book value as indicators of financial performance, I find that the results concerning financial performance of family and non-family firms are highly sensitive to the way family firms are defined. A broad definition results in no statistically significant difference between family and non-family firms – both with ROA and market-to-book value. However, a narrow definition (in which family firms are defined as firms in which the family is both active as owner and manager of the firm) leads to the opposite result. Using this and other narrow definitions, family firms are shown to outperform non-family firms. Although there are slight differences in the size of the effects, Bayesian and classical analysis produce similar results, which is an indicator of robustness. Turning to the second question, regarding the impact of the two family firm dimensions family management and family ownership, an interesting result emerges: family management has a strong positive impact, whereas the effect of family ownership is neutral. Again, this finding is supported by both Bayesian and classical analysis. How do these findings contribute to the existing literature? As mentioned in the beginning of this chapter, the existing literature is unclear about a superior performance of family firms. The findings in this chapter contribute to the discussion in four ways. First, the findings confirm the difficulty of attributing superior performance to either family or non-family firms. The results differ depending on the exact definition of a family firm and depending on the performance measure used (market-to-book value and ROA). For more on this issue, I refer to Astrachan and Zellweger (2008) and Miller et al. (2007). Second, it is shown that when a narrow family firm definition is used, in which the family is required to be active as both owner and manager of the firm, family firms do indeed perform better than non-family firms. These results are in line with Andres (2008) and Miller et al. (2007). Further, the Bayesian analysis indicates that it is in particular the ‘young’ family firms (defined as firms with age 50) who outperform. This leads to the question of how much of the performance differences between family and non-family firms can be attributed to family firms in the 2nd or later generation and how much can be attributed to founder-managed- and/or founder-owned firms.168 Note that most of the disadvantages of family versus non-family firms be-
168
Due to missing data about the exact number of family members involved, I am not able to analyze this question.
5 Performance of family firms
93
come more pronounced the more family members are involved (e.g., conflicts within the family or altruism towards members of the family, Table 5-1). Third, my results suggest that family management and family ownership, the two distinct characteristics of family firms, differ with regard to their impact on financial performance. A superior performance of family firms is driven by the management dimension. This contributes to the discussion about the reasons for performance differences. Those explanations that are linked to the ownership dimension such as more effective monitoring (Fama, 1980; Maug, 1998) or stronger long-term orientation by the owners (Anderson and Reeb, 2003) are not supported by the data. Finally, the combination of Bayesian and classical methods is new to this area of research. The combination of the two methods led to more robust findings; furthermore, the exact inferences generated from the Bayesian approach allowed a direct comparison of the different groups of family firms, irrespective of whether a result is statistically significant or not. This way, I see a potential of Bayesian methods to make a contribution with regard to the widely discussed family business definition dilemma (see Section 2.1). How do these results relate to the main research question of this thesis whether family firms are more long-term oriented than non-family firms? As noted in the introduction of the chapter and the introduction to the thesis in Chapter 1, long-term orientation is referred to as an explanation why family firm perform better than non-family firms. It is argued that family firms perform better, since family owners provide more ‘patient capital’ than other owners do, which allows the management to pursue profitable projects that pay off in the long run. The findings of this chapter do not support this view. Hence, either family owners do not provide more ‘patient capital’ compared to other owners or the link between long-term orientation and financial performance is weak or outweighed by other, negative effects. The next two chapters about investments in R&D and human capital in family and non-family firms will shed further light on this issue.
6 Family firms and R&D spending169 This chapter explores R&D spending in family firms. R&D spending is used as an indicator for long-term orientation in innovative industries. Long-term orientation refers to the stance of a firm that focuses on benefits accruing over the long term. The following research questions are analyzed: (1) Are family firms spending more on R&D than non-family firms? (2) What are the effects of family ownership, family management, and family firm age on R&D spending? Section 6.1 poses the research questions and introduces R&D spending as a measure of longterm orientation, as well as providing an overview of the empirical work that has been done in this area. Section 6.2 uses current theory to develop hypotheses concerning the behavior of R&D spending in family firms. Section 6.3 describes the data. Section 6.4 tests the hypothesized relationships against the empirical data. Section 6.5 discusses the implications of the findings, both from a research and a management perspective. Section 6.6 summarizes the results and conclusions.
6.1 6.1.1
Introduction R&D spending, innovation, and long-term orientation
Successful innovation usually requires a long-term perspective. It often takes years to discover a new product and fully realize its commercial benefits. Consider, for example, the cases of laser technology and fiber optics. In the case of laser technology, the scientific community had known the principles and ideas required for its invention since the 1920s.170 It was only in 1958, however, that Arthur Schawlow and Charles Townes introduced the concept of a laser (Schawlow and Townes, 1958). Following the work of Schawlow and Townes (1958), the first operable laser was in-
169
170
This chapter is a substantially revised version of Block and Thams (2007b). The paper was presented at European Business School (European Family Business Center), Technische Universität München, Ludwig-MaximiliansUniversität München (TIME Kolloquium, Seminar für Empirische Wirtschaftsforschung), Erasmus University Rotterdam (Organization Seminar), the 2007 and 2008 annual conferences of the International Family Enterprise Research Academy (IFERA), the 2008 Academy of Management Meetings, and the G-Forum 2007 (annual conference of the FGF). It received the Jack A. Dinos/Cox Family Enterprise Center Best Family Business Paper Award at the 2008 Academy of Management Meetings and the Family Business Network (FBN) Best Family Business Research Paper (Policy) Award at the 2008 annual IFERA conference. An abridged version of the paper is also included in the best paper proceedings of the 2008 Academy of Management Meetings (see Block and Thams, 2008). The theoretical foundation for research in this area was formed in 1917 by Albert Einstein in his theory of stimulated emission (Einstein, 1916). See http://www.greatachievements.org/?id=3706 (accessed Sep. 20th, 2008).
6 Family firms and R&D spending
95
vented in 1960, the first medical laser in 1961, and the first commercial semiconductor laser in 1975.171 Today, laser technology is a multi-billion dollar industry with a wide range of practical applications, such as bar-code readers, laser printers, and CD or DVD players. Lasers are also widely used in medicine and manufacturing. This example clearly demonstrates that it can take a long time for scientific knowledge to create innovative products. The fiber-optics industry tells a similar story. In 1966, Charles Kao and George Hockham of Standard Telecommunications Laboratories in England demonstrated that optical fiber transmits laser signals better the purer the glass strands are (Kao and Hockham, 1966).172 Following this discovery, researchers immediately focused on ways to purify glass. In 1970, a team from Corning Glass Works, United States, (hereafter Corning) reported the creation of optical fibers meeting the standards set by Kao and Hockham (1966). Corning is a family firm that, at that time, was in its fifth family generation.173 It was not by chance that Corning’s researchers were the first to develop such demanding fibers: the firm had consistently invested a large share of its profits in R&D. James Houghton, CEO of the company and great-great-grandson of Corning’s founder, Amory Houghton, explains the devotion of Corning to invest in R&D as follows: “When I was growing up I was taught that investing in R&D was like a ‘religion’. […] The most important effect of family owner sentiments is that you’re in it for the long run. You don’t focus on the next quarter. So Corning invested in fiber optics for eighteen years before realizing any returns.” (cited after Miller and Le Breton-Miller, 2005, p. 150). This statement bears directly on the research questions that will be analyzed in this chapter. Are family shareholders really in it for the long-run, and do they therefore invest more in R&D than other shareholders? If so, is this true only for family firms in the first generation or does it also apply to multi-generation family firms? Finally, what happens with R&D spending when a family owner leaves the management and the firm comes under the control of a non-family manager? Before I proceed to develop hypotheses to answer these questions, I will discuss the indicator R&D spending (Section 6.1.2) and summarize the empirical literature that uses R&D spending to measure long-term orientation (Section 6.1.3).
171 172
173
See http://www.greatachievements.org/?id=3706 (accessed Sep. 20th, 2008). See Kaplan (2008) for a detailed study about the optical fiber industry. The study analyses investments in R&D or the patents generated from these investments for 71 firms over a period of 20 years. See http://www.corning.com/index.aspx (accessed Sep. 20th, 2008).
96
6.1.2
6 Family firms and R&D spending
What does R&D spending measure?
R&D spending is used as an indicator for a variety of concepts, and this discussion will focus on two of them: long-term orientation and entrepreneurial orientation.174 Long-term orientation. R&D spending has been used in a number of studies as a proxy for
long-term orientation (e.g., Bushee, 1988; David et al., 2001; Hansen and Hill, 1991; Graves, 1988; Zahra, 1996). As the above example of the laser industry demonstrates, R&D spending involves an intertemporal choice problem: The costs of R&D are incurred in the near term, whereas the payoffs from R&D are likely to occur only over the long term. Contrary to other investments (e.g., capital expenditures), R&D spending is usually expensed rather than depreciated.175 R&D spending is therefore a good proxy for the long-term orientation of a firm in an innovative industry. However, using R&D data to determine long-term orientation is not without its problems. R&D data may also include some short-term projects (Laverty, 1993).176 Furthermore, R&D does not always create economic value (Erickson and Jacobson, 1992; Hall, 1993). This aspect implies that R&D spending may, in fact, measure risk-taking behavior rather than the willingness to invest in long-term projects.177 Finally, high R&D expenses may also indicate that R&D expenses are out of control, i.e. that a firm is managed inefficiently. It also indicates the influence of the scientists in the firm. Entrepreneurial orientation. The concept of entrepreneurial orientation has received a lot of
attention from entrepreneurship, strategy, and family business scholars (e.g., Covin and Slevin, 1989; Naldi et al., 2007; Nordqvist, 2008; Wiklund and Shepherd, 2003).178 Entrepreneurial orientation relates to the processes, practices, and decisions leading to entrepreneurial behavior (e.g., entry into a new market). Its key characteristics are the willingness to act autonomously, the willingness to be innovative and to take risks, proactive behavior towards market opportunities, and aggressive behavior towards competitors (Lumpkin and Dess, 1996). A number of reliable and validated scales
174
175
176
177
178
R&D spending has also been used as a proxy for absorptive capacity (Cohen and Levinthal, 1990) and for risktaking (Zahra, 2005). In other words, the full costs appear in the income statement in the year in which the firm makes the investment; they are not spread over time as it is done with other investments (e.g., capital expenditures). The profit is reduced in the year the firm makes the R&D investments. Note that this is true for US GAAP, but not for all accounting systems. For more on the issue of R&D spending and earnings management, see Baber et al. (1991). R&D spending includes salaries, raw material, equipment for development efforts, travel, and training (Gildersleeve, 1999, p. 112). Note that the figure does not indicate whether the expenses are incurred for the development of new products, or whether they are incurred for the further development of existing products; the latter may not be a long-term strategy. See also the discussion about the relationship between long-term orientation and attitude towards uncertainty in Section 3.1. Note that a number of studies have analyzed corporate entrepreneurship in family firms, which is a closely related concept (e.g., Kellermanns and Eddleston, 2006; Salvato, 2004).
6 Family firms and R&D spending
97
exist to measure entrepreneurial orientation (e.g., Covin and Slevin, 1989; Knight, 1997; Miller and Friesen, 1978).179 In addition, a number of scholars have examined firm resource allocations in order to analyze different dimensions of entrepreneurial orientation. Hitt et al. (1997) used R&D intensity as a proxy for innovation. Some studies have used the ratio of debt to equity as an indicator of the willingness to take risks (e.g., Baird and Thomas, 1985; Gale, 1972).180 I argue that R&D spending correlates strongly with the three entrepreneurial orientation dimensions innovativeness, willingness to take risks, and proactiveness. Innovativeness describes a firm’s capacity to innovate, i.e. to create or adopt innovations and implement them. A high level of R&D spending enables a firm to create innovations and correlates with the degree of innovativeness. Willingness to take risks refers to the extent to which a firm engages in projects with uncertain outcomes. In most cases, the outcome of R&D is uncertain; firms with a high level of R&D spending can therefore be described as risk takers. Finally, proactiveness refers to the degree to which a firm acts in anticipation of future events. A proactive, forward-looking firm in an innovative industry would be expected to engage in a high level of R&D spending in order to remain competitive. To summarize, R&D spending is used as a proxy for a variety of concepts, including longterm orientation and entrepreneurial orientation. In this thesis, I use R&D spending as an indicator of long-term orientation. The next section reviews empirical work in this area. 6.1.3
Studies using R&D expenditures as a proxy to measure long-term orientation
There exist a number of studies who use R&D data to measure long-term orientation. Most of these studies concern the role the role of institutional investors in formulating a firm’s R&D strategy. Their main finding is that, generally, institutional investors encourage managers to invest more in R&D and thereby help to mitigate the problem of managerial myopia (e.g., Baysinger et al., 1991; Bushee, 1998; David et al., 2001; Hansen and Hill, 1991). Similarly, using an outcome-based measure of innovation, Kochhar and David (1996) find that the rate of new product development increases with activism of institutional investors. It should be noted, however, that this positive effect varies by the type of institutional investor and the degree of shareholder activism. Bushee (1998) finds that institutional investors with high portfolio turnover who engage in short-term trading usually pressure managers to reduce R&D spending. David et al. (2001) find that institutional
179 180
Lyon et al. (2000) provide an overview of scales measuring entrepreneurial orientation. Although the reliability of such archival data is generally high, its validity with regard to the construct of entrepreneurial orientation or the specific dimension can be questioned. It must be kept in mind, for example, that R&D spending does not capture the outcomes of innovation. Moreover, it is impossible to distinguish between different types of innovation, e.g., new product innovation, technological innovation, and process innovation (Lyon et al., 2000).
98
6 Family firms and R&D spending
ownership alone is insufficient to have an effect: Investor activism is also required to induce managers to invest more in R&D. In addition to differences between types of institutional investors, there are also differences between countries, which can be explained by the different systems of corporate governance in different countries. Comparing the US and Japan, for example, Lee and O’Neill (2003) find that the level of ownership concentration has a positive impact on R&D spending for US firms but not for Japanese firms.181 Following a similar line of research, Hall et al. (1999) and Mulkay et al. (2001) find that cash flow has a much larger effect on R&D in the US than in France or Japan. To the best of my knowledge, there has been no published study analyzing R&D spending in family firms.
6.2
Theory and hypotheses This section develops hypotheses that will be tested against empirical data in a later part of
this chapter. Most of the arguments concern R&D spending as an indicator of long-term orientation, and hence this section is closely linked to the discussion about long-term orientation in family firms (Section 3.3). 6.2.1
Impact of family ownership on R&D spending
It has been argued that stock market myopia and impatient capital induce myopic decisions.182 According to this view, the stock market undervalues long-term behavior and forces managers to pursue short-term benefits. However, these short-termism arguments based on stock market myopia (e.g., Jacobs, 1991; Johnson and Kaplan, 1987) and impatient capital (e.g., Porter, 1992) are much less relevant for firms that have a family as their main shareholder. First, from the perspective of a family shareholder, the firm is not just an asset that may be sold easily: In general, the firm also symbolizes the heritage and tradition of the family and is part of the family identity. Consequently, family shareholders generally intend not to sell the firm but to pass the firm over to the next family generation (e.g., Casson, 1999; Guzzo and Abbot, 1990; James, 1999; Tagiuri and Davis, 1992). Second, the reputation of the family in the public is strongly linked to the well-being of the firm (Dyer and Whetten, 2006). In fact, family-owned firms often bear family names. As a result, shareholder families are expected to be less likely than other types of shareholders to move their funds around quickly and to evaluate their investments only in terms of short-term results. The theory of
181
182
For further studies comparing the long-term orientation of Japanese firms versus US firms, see Beldona et al. (1998) or Peterson et al. (2002). For more details regarding these two explanations of managerial myopia, refer to Section 3.3.5.
6 Family firms and R&D spending
99
psychological ownership may also be called upon to clarify the differences between family and nonfamily shareholders. Pierce et al. (2001) define psychological ownership as “the feeling of possessiveness and of being psychologically tied to an object” (Pierce et al., 2001, p. 299). A root of psychological ownership is in the control of the particular object that one owns. I argue that this applies more to family than to non-family shareholders. In contrast to non-family shareholders, family shareholders often have a close link to management – frequently through kinship ties. This close link allows them to explore and alter the firm and its environment. As a result, they are likely to experience a greater feeling of being psychologically tied to the firm than non-family shareholders are. Two other arguments suggest a positive relationship between family ownership and R&D spending. First, managerial opportunism should be less of a problem, since the incentives to engage in effective monitoring are greater for family shareholders than for other shareholders. As described above, family owners are strongly linked to their firms by feelings of identity and by concern for its reputation outside the organization. In addition, since their fraction of ownerships are typically higher than with other investors, economic benefits from monitoring are also higher; the free-rider problem associated with firms that have dispersed shareholders is unlikely to exist in a family firm (Fama, 1980; Maug, 1998). With more extensive and more effective monitoring in family firms, managers have less latitude to engage in opportunistic behavior. Second, information asymmetry between owners and management is less of a problem in family firms since business-owning families have usually known their businesses for a long time and have developed a thorough understanding of its underlying processes. As a result of this familiarity of the owner with the workings of a firm, there is less need for a manager to use strong short-term results as a signaling device to indicate her competence to the owner (Thakor, 1990). From this consideration of the properties of family firms, the following hypothesis is proposed:
Hypothesis 6-1a:
The level of family ownership has a positive influence on the level of R&D spending.
This hypothesis should, however, be introduced with reservations, and there are reasons why high shares of family ownership may not result in high R&D investment; in fact, R&D investment may even be lower in family firms. This can be the case if the owning family seeks primarily private benefits of control, i.e. benefits for the family that would not be shared by other (non-family) shareholders. This has an impact of R&D spending if the owning family uses its influence to divert
100
6 Family firms and R&D spending
resources away from projects that provide a long-term benefit to the firm as a whole (such as R&D spending) to projects that generate short-term benefits mainly for the family. In many family firms, the owning family is in a good position to pursue such a strategy. Effective monitoring and, as a result, a strong degree of corporate control gives the owning family a strong hold over the management of the firm. Furthermore, the separation of ownership and control is generally more pronounced in family than in non-family firms (Claessens et al., 2000; La Porta et al., 1999).183 The reason is that through pyramid structures or dual class stock the family’s voting rights may exceed their formal cash-flow rights. Moreover, the family may use their (formal or informal) influence to appoint directors who are friends or family members. In the finance literature, it has been argued that such entrenched shareholders and their desire to gain private benefits of control can be disadvantageous for a firm and its (long-term) performance (Claessens et al., 2002; Dharwadkar et al., 2000; McConnell and Servaes, 1990; Shleifer and Vishny, 1997; Stulz, 1988). If these characteristics of family firms predominate, then the following hypothesis (contrary to 6-1a) will hold:
Hypothesis 6-1b:
6.2.2
The level of family ownership has a negative influence on the level of R&D spending.
Impact of family management on R&D spending
Kets de Vries (1993) compares non-family CEOs with family CEOs and finds that the latter are “not haunted by quarterly results” […] and “are more willing to plow profits back into the business” (Kets de Vries, 1993, p. 62). How can this be explained? Agency theory can be used to explain differences in investment behavior between firms governed by a family manager and firms governed by a non-family manager.184 The arguments from agency theory relate in a direct way to the managerial opportunism explanation and the information asymmetry explanation of managerial myopia.185 Agency theory is widely used to explain the relationship between the management and the owners of a firm (e.g., Fama, 1980; Jensen and Meckling, 1976; for a review, see Eisenhardt, 1989). It describes managers as rational actors, who seek to maximize their individual utilities. From an agency theorist’s perspective, then, the relationship between the management and the owners of a firm differs fundamentally between family-managed firms and non-family-managed firms. 183 184 185
For more on this conflict between dominant and minority shareholders, see Section 2.3.1. Agency theory is discussed in detail in Section 2.3.1. For more on the managerial opportunism explanation of managerial myopia, see Section 3.3.2; for more on the information asymmetry explanation, see Section 3.3.4.
6 Family firms and R&D spending
101
In the first case, managerial opportunism is less of a problem, since no (or only minor) agency conflicts exist. The utility functions of shareholders and management coincide to a large degree, since they are often members of the same family and generally have similar or identical goals. This argument is made even stronger by the fact that most family managers own a substantial share of the company they manage. In this case, they resemble owner-managers where no agency conflict exists between the management and owners of a firm.186 Another argument for lower levels of managerial opportunism in family firms is that family managers have relatively safe jobs (e.g., Allen and Panian, 1982; Le Breton-Miller and Miller, 2006), because it is harder for an owner to lay off a member of her own family. Accordingly, family managers do not need to increase their reputation among corporate executives through strong short-term results. The situation is different with nonfamily managers who can be more easily laid off and who therefore have an incentive to engage in managerial opportunism. In the absence of kinship ties, their utility functions do not necessarily coincide with those of the owners, and they are much less likely to own a significant share of the firms they manage. Finally, due to a lower degree of entrenchment, their jobs should be less safe compared to those of family managers, and hence they have a greater incentive to produce strong short-term results to secure their jobs and maintain good reputations on the external market for executives (Campbell and Marino, 1994; Narayanan, 1985). If these factors predominate, the following hypothesis should hold:
Hypothesis 6-2a:
Family management has a positive influence on the level of R&D spending.
There are, however, factors that can work in the opposite direction. Due to the limited human resource pool of the owning family, non-family managers are more likely to have a greater stock of knowledge than family managers (e.g., Bennedsen et al., 2007; Burkart et al., 2003; PerezGonzalez, 2006). In addition, non-family managers often have more varied career experiences than family managers. Both greater managerial knowledge and greater variation in career experiences positively influence the level of R&D spending (Barker and Mueller, 2002; Cohen and Levinthal, 1990; Kaplan, 2008). These effects may dominate over those due to the agency mismatches described above. If this is the case, then the following hypothesis is expected to hold:
Hypothesis 6-2b:
186
Family management has a negative influence on the level of R&D spending.
See the Jensen and Meckling (1976) model in Section 2.3.1.
102
6 Family firms and R&D spending
Finally, there could be an interaction effect between the degree of family ownership and family management.187 The direction of this effect, however, is unclear. On the one hand, the family may use its seats on the management board to further its interest in generating private benefits of control, in which case the interaction term has a negative sign. On the other hand, a duality of family ownership and management may indicate the family’s strong interest in the firm’s development and its long-term existence, in which case the interaction term has a positive sign. 6.2.3
Impact of family firm age on R&D spending
The next two hypotheses relate to differences within the group of family firms. Evidence suggests that only approximately 30% of family firms are transferred to the second generation and only approximately 13% to the third generation (Ward, 1987; Westhead et al., 2002). Viewing the life cycle of a family firm as a selection process provides an argument for a positive effect of family firm age on long-term orientation. According to this view, the owning families behind multigeneration family firms should have a long-term horizon and be capable of organizing the succession process successfully. Moreover, each succession adds valuable experience to the family and to the firm (Klein et al., 2005), which, in turn, should have a positive influence on the firm’s capacity to absorb new knowledge (Cohen and Levinthal, 1990). Both arguments suggest a positive relationship between the age of a family firm and the degree of R&D spending, which yields the following hypothesis:
Hypothesis 6-3a:
R&D spending in old family firms is higher than R&D spending in young family firms.
Other perspectives argue against this hypothesis. The owning family is likely to grow in size over time, and so family ownership is often more dispersed in old family firms than in young family firms.188 This has two implications. First, due to free-rider problems among the many members of the owning family, monitoring of management may be less effective (Fama, 1980; Maug, 1998); a family member that engages in monitoring has to share the benefits of monitoring with the other family members and therefore may decide not to engage in monitoring. Furthermore, a larger family also increases the potential for sibling rivalries or conflicts between two family generations (Eddle-
187 188
In other words, the effect of family ownership increases or decreases with family management. Consider, for example, the two German family firms Freudenberg Group and Haniel Group. In 2008, the Freudenberg Group is composed of 434 subsidiary firms and 306 family shareholders (Sywottek, 2008); the Haniel Group is owned by more than 550 members of the Haniel family (James, 2006).
6 Family firms and R&D spending
103
ston and Kellermanns, 2007; Harvey and Evans, 1994), which may distract the family from effective monitoring. Second, as the number of family members grows, peer pressure and mutual monitoring are less likely to mitigate the negative effects of the free-rider problem associated with a larger family (Kandel and Lazear, 1992).189 Another argument for the negative impact of family firm age on the level of R&D spending is that, over time, family firms may become hostile to change and adopt conservative strategies: routines used in the past are used again and again with potentially inefficient outcomes. The reasons for such conservatism may include the founder’s reluctance to transfer power or difficulties in integrating competent non-family employees into the firm (Beckhard and Dyer, 1983; Stavrou, 1999; Vago, 2004; Ward, 1987; Zahra et al., 2004). If the negative effects of family size on long-term orientation predominate, the following hypothesis should hold:
Hypothesis 6-3b:
6.3 6.3.1
R&D spending in old family firms is lower than R&D spending in young family firms.
Data Sample
To analyze R&D spending in family firms, I use a subsample of the dataset described in Section 4.1. The subsample contains only 1,291 observations from 197 firms (full dataset: 4,856 obs. from 499 firms). To construct the subsample, the following ‘data cleaning’ steps were undertaken. First, all observations with missing data on R&D spending (3,234 obs.) were excluded. Next, I lagged the independent variables by one year, which then led to a further loss of 262 observations. In a final step, I excluded observations with missing data on one or more of the independent variables used in the regressions below (69 obs.). Table 6-1 provides family and non-family firm-year observations by industry. Most observations fall into the industries SIC 28 (chemical and allied products) (186 obs.), SIC 35 (industrial machinery and equipment) (191 obs.), SIC 36 (electronics and other electrical equipment) (238 obs.), SIC 37 (transportation equipment), SIC 38 (instruments and related products), and SIC 73 (business services) (138 obs.).190 With the broad definition family firm1, approximately 37% of the observations in the sample belong to family firms (479 obs.); with the narrow definition family 189
190
Pieper (2007) shows the ways in which large, multi-generation family firms deal with these problems in order to assure family cohesion and the long-term survival of the firm. Note that this result is similar to those reported in prior studies. Griliches and Mairesse (1984) identify the two digit industries 28, 35, 36, and 38 as being research-driven (see also Bernstein and Nadiri, 1988; Himmelberg and Peterson, 1994).
104
6 Family firms and R&D spending
firm3, only 11% of the observations in the sample belong to family firms (148 obs.).191 The share of family firm observations varies strongly by industry (e.g., 59% in business services vs. 14% in transportation equipment). Table 6-1 also shows the median of R&D/assets per industry. R&D spending differs very much by industry; for example, the median ratio of R&D expenditures to assets is 8.5% in the electronics industry, whereas it is only 2.8% in the transportation equipment sector. The variations in the R&D/assets ratios and the shares of family firm observations by industry indicate the need to control for industry specifics in the multivariate analyses.
191
For the family firm definitions, see Section 2.1.2 and Table 2-2. Using similar data, Anderson and Reeb (2003) classify 35% of firms as family firms. Depending on the definition, Villalonga and Amit (2006) classify between 7% and 37% as family firm observations.
6 Family firms and R&D spending
105
Table 6-1: Family and non-family firm observations by R&D-intensive industry
SIC code 13 20 21 25 26 28 29 30 33 34 35 36 37 38 39 51 73 99
Industry description Oil and gas extraction Food and kindred products Tobacco products Furniture and fixtures * Paper and allied products * Chemical and allied products * Petroleum and coal products Rubber and miscellaneous plastic products Primary metal industries * Fabricated metal products Industrial machinery and equipment * Electronic and other electrical equipment * Transportation equipment * Instruments and related products * Miscellaneous manufacturing products Wholesale trade – nondurable goods Business services * Nonclassifiable establishments Total
Median of R&D/assets1
N Family firms (obs.)
N Non-family firms (obs.)
Family firm obs. in % of all observations
0.14%
0 (0)
2 (13)
0%
1.21%
5 (45)
1 (5)
90%
0.77%
0 (0)
2 (10)
0%
0.57%
1 (7)
2 (7)
50%
0.90%
3 (12)
5 (25)
32%
3.16%
10 (48)
23 (138)
26%
0.12%
0 (0)
3 (11)
0%
1.89%
2 (6)
4 (25)
19%
0.72%
2 (7)
3 (22)
24%
1.95%
1 (9)
6 (38)
19%
3.79%
10 (64)
19 (127)
34%
8.53%
24 (121)
19 (117)
51%
2.84%
3 (14)
13 (86)
14%
5.72%
10 (56)
14 (77)
42%
3.87%
1 (9)
2 (16)
36%
0.37%
0 (0)
3 (23)
0%
7.31%
16 (81)
13 (57)
59%
1.30%
0 (0)
2 (15)
0%
(
88 (479)
136 (812)
37%
Note: The broad definition family firm1 is applied. 27 firms changed from the category family firm1 to the category non-family firm or vice versa. The industries in which such a change occurred are indicated by an asterisk. 1 Based on observations (not firms)
106
6.3.2
6 Family firms and R&D spending
Measures
Dependent variables
The main dependent variable in the regressions is the ratio of R&D expenditures to assets (R&D/assets). In order to test robustness of these results, I also use the ratio of R&D expenditures to sales (R&D/sales) as a dependent variable. Independent variables
To analyze differences between family and non-family firms in terms of R&D spending, I use five family firm dummies constructed from the definitions in Table 2-2 (family firm1-5). To investigate the effect of family firm characteristics on R&D spending, I constructed the variables family management, ownership share of family, and firm age. The variable family management is an indicator variable that equals one if a member of the founding family is either CEO or chairman. The variable ownership share of family gives the percentage of common equity owned by the founding family. Firm age is the number of years since the firm was founded. The following variables are used as controls. The variable ownership share of financial investors controls for the influence of institutional investors and gives the percentage of stock owned by large banks (e.g., Citigroup or JP Morgan), insurance companies (e.g., The Prudential Insurance Company or AXA), mutual funds (e.g., Fidelity Investments), private equity firms (e.g., KKR) and large individual financial investors (e.g., Warren Buffet).192 The variable cash flow/assets accounts for the firm’s state of liquidity.193 The firm’s market-to-book value is used to control for the firm’s market valuation.194 To distinguish between fast- and slow-growing firms, I employ the variable sales growth, which gives the 3-year least squares annual growth rate of sales. The variables firm size and leverage are used to account for firm size and capital structure.195 The variable CEO duration measures the number of years the CEO has occupied the role of chief executive in the firm.196 The variable CEO duality equals one if the CEO is also chairman of the board of directors. Industry 192
193 194
195
196
I am not able to apply a more discriminating measure because it is often difficult to distinguish among the different types of financial investors. Large banks and insurance companies are also sometimes active in the private equity business. Cash flow is the sum of after-tax income, depreciation, and after-tax R&D (Hall, 1992). The variable is determined as the market value of equity at the end of the year plus the book value of debt divided by the book value of total assets. Firm size is the book value of total assets; leverage is determined as the ratio of the book value of debt to the book value of total assets. See Dechow and Sloan (1991) for a paper about the relationship between CEO duration and R&D spending.
6 Family firms and R&D spending
107
dummies are included to control for the 18 different industries in the sample. Finally, time dummies for the years 1994-2003 are used to control for macro-economic effects (e.g., the burst of the Internet bubble in 2001). Since the distributions of the variables firm size and firm age are highly skewed, logarithmic values are taken. Aside from the variables firm age, sales growth, industry, and time dummies, all other covariates are lagged by one year in order to avoid problems of endogeneity. Table A-2 in the Appendix describes the construction of the variables in more detail.
6.4
Results
6.4.1 Univariate analysis
Table 6-2 compares family and non-family firms with respect to R&D spending and firm characteristics.197 It presents means, medians, and interquartile ranges of the variables for the two groups. Most distributions are highly skewed, which is why median values and a Wilcoxon ranksum test are used to compare family and non-family firms. Univariate statistics show family firms to have a higher ratio of R&D expenditures to assets than non-family firms (median 5.8% vs. 3.2%, p<0.01).198 However, this preliminary result should be interpreted with great caution because the proportion of family firms is particularly high in industries with high levels of R&D spending (e.g., business services or electronics and other electrical equipment; Table 6-1). Note that the situation is different with regard to capital expenditures. There are no statistically significant differences between family and non-family firms regarding the variables CAPEX/assets and CAPEX/sales.199 Concerning firm characteristics, the following differences stand out: relative to non-family firms, family firms are on average younger, smaller, faster-growing, and associated with a lower level of debt. They are valued higher by the stock market, as measured by the firm’s market-to-book value (median 3.1 vs. 1.0, p<0.01). The ratio of cash flow to assets is also higher in family firms than in non-family firms (median 13.8% vs. 11.3%, p<0.05). Finally, CEOs in family firms have a longer tenure (median 5.5 yrs vs. 4, p<0.01) and serve less often as chairman (69% vs. 87%, p<0.01). In summary, the univariate analysis reveals that R&D-intensive family and non-family firms have significantly different characteristics, which shows the necessity to control for these differences in the multivariate analysis.
197 198
199
The broad definition family firm1 is used to determine the group of family firms. The differences between family and non-family firms are even bigger when the ratio of R&D expenditures to sales is used (median 9.02% vs. 3.43%, p<0.001). Often, capital expenditures concern long-term projects. E.g., consider the decision to build a new factory.
9.9
0.87
4.5
CEO duality t-1 (yes/no)
CEO duration t-1 (in yrs)
3.2
14.0
1.9
12.6
7.5
3.2
43.6
8.7
3.8
2.5
9.5
5.3
2.6
2.2
1.0
14.0
8.0
2.6
40
10.7
3.6
3.5
1.8
1.7
Lower quartile
4
6.3
1.6
23.5
11.3
5.1
80
15.4
4.8
4.9
3.4
3.2
Median
Non-family firms (N=122)
Std. dev.
5
13.6
2.8
30.1
16.1
11.6
102
20.7
7.0
6.3
10.2
8.3
Upper quartile
8.0
0.69
24.2
4.0
15.9
13.1
5.6
42.1
11.9
6.6
5.0
12.0
7.3
Mean
5.5
6.9
32.2
3.5
12.6
10.6
8.9
39.3
9.5
4.4
3.1
13.7
3
5.4
1.8
3.3
9.2
1.6
14
3.6
3.7
3.1
3.3
3.1
Lower quartile
5.5
15.1
3.1
17.2
13.9
2.8
25
10.6
5.3
4.3
9.0
5.8
Median
Family firms (N=75) Std. dev.
11
30.9
6.1
15.7
18.6
5.7
60
17.7
7.6
5.3
16.3
11.0
Upper quartile
p<0.001
p=0.004
p<0.001
p<0.001
p<0.001
p=0.226
p=0.057
p<0.001
p=0.003
p=0.292
p=0.774
p=0.003
p=0.016
Test for equality of means/ proportions
p<0.001
p<0.001
p<0.001
p<0.001
p=0.035
p<0.001
p<0.001
p<0.001
p=0.468
p=0.158
p<0.001
p=0.002
Wilcoxon rank-sum testa
Notes: Data for the univariate statistics are based on time series averages for each firm, and then averages are taken across firms. A firm is classified as a family firm if it has more family than nonfamily observations and vice versa. The definition family firm1 is used. Some variables are lagged by one year. The reason is that they enter the multivariate analyses as lagged values. The number of observations per firm and the time period covered does not differ between family and non-family firms. a The Wilcoxon rank-sum test analyzes whether the two samples are from different distributions (sample1: non-family firms, sample2: family firms).
2.3
13.0
Firm size t-1 (in bn $)
Sales growth (in %)
75.9
Firm age t-1 (in yrs)
Market-to-book value t-1
16.0
Ownership share of financial investors t-1 (in %)
11.5
5.9
CAPEX/sales (in %)
23.2
5.1
CAPEX/assets (in %)
Leverage t-1 (in %)
7.1
R&D/sales (in %)
(Cash flow/assets) t-1 (in %)
5.4
R&D/assets (in %)
Mean
Table 6-2: Characteristics of (R&D-intensive) family and non-family firms
108 6 Family firms and R&D spending
6 Family firms and R&D spending
109
6.4.2 Multivariate analysis Family versus non-family firms
Do family firms have a higher level of R&D spending relative to non-family firms? To examine this question, I estimate several regression models with R&D/assets as the dependent variable, using independent variables as shown in the tables below. Table 6-4 presents the results of several classical regressions comparing family and non-family firms, and Table 6-5 displays the corresponding Bayesian regressions.200 Figure 6-1 displays the respective distribution functions graphically. The results are surprising. The family firm definition plays a significant role in the level of R&D spending. Using a broad definition, the effect of the family firm dummy is negative. Controlling for the particularities of family firms such as younger firm ages, lower levels of debt and smaller firm sizes, R&D spending is about one percentage point lower in family firms than in nonfamily firms (model I).201 The Bayesian model shows that the situation changes does not change strongly when a narrow definition is used (e.g., family firm3): The effect of being a family firm stays negative, with R&D spending about one percentage point lower than in non-family firms (models III-V).202 Some control variables have a strong effect. Both the Bayesian and the classical analysis indicate a negative effect of firm size and firm age on R&D spending. In the Bayesian model, the degree of ownership by financial investors correlates negatively with R&D spending.203 The same is true for the variable sales growth. Finally, industry and year effects seem to be important, as affirmed by F-tests on joint significance of the industry and time variables in the classical models. CEO characteristics such as tenure and duality do not seem to have a significant effect on the level of R&D spending.
200 201
202
203
See Section 4.3 for a comparison of classical and Bayesian methods. The coefficient of family firm1 is -0.010 (p<0.1) in the classical model. The Bayesian analysis yields a median coefficient of -0.011; the probability that the coefficient is positive is only 2%. In the Bayesian model, the median coefficient of family firm3 is -0.012; the probability of a positive effect is 3.9%. In the classical model, the effect of the variable family firm3 is -0.0111 (p<0.05). In the classical model, the effect of the variable is insignificant.
110
6 Family firms and R&D spending
Table 6-3: Family versus non-family firms – classical RE regressions of R&D/assets Independent variables Family firm1 t-1
Model I Coefficient (SE) -0.0096 * (0.0050)
Model II Coefficient (SE)
Model III Coefficient (SE)
Model IV Coefficient (SE)
-0.0072 (0.0045)
Family firm2 t-1
-0.0111 ** (0.0197)
Family firm3 t-1
-0.0030 (0.0056)
Family firm4 t-1 Family firm5a t-1 Ownership share of financial investors t-1 Firm age Firm size t-1 (Cash flow/assets) t-1 Leverage t-1 Market-to-book value t-1 Sales growth CEO duality t-1 CEO duration t-1 Industry dummies (17 categories) Year dummies (9 categories) N obs. (firms) Obs. per group: min./avg./max. R² within, between, overall p-value Chi²-test p-value Breusch-Pagan test Rho (fraction of variance due to ui)
Model V Coefficient (SE)
-0.0197 (0.0128) -0.0163 *** (0.0043) -0.0102 *** (0.0031) -0.0067 (0.0111) -0.0281 * (0.0151) 0.0001 (0.0004) -0.0126 * (0.0067) 0.0004 (0.0037) -0.0002 (0.0002) p<0.001
-0.0198 (0.0127) -0.0176 *** (0.0043) -0.0101 *** (0.0031) 0.0059 (0.0108) -0.0276 * (0.0150) 0.0001 (0.0003) -0.0138 ** (0.0066) -0.0002 (0.0036) -0.0001 (0.0036) p<0.001
-0.0208 (0.0127) -0.0171 *** (0.0042) -0.0102 *** (0.0031) 0.0061 (0.0109) -0.0272 ** (0.0151) 0.0001 (0.0004) -0.0134 ** (0.0066) 0.0001 (0.0035) -0.0002 (0.0002) p<0.001
p=0.009
p=0.009
p=0.008
p=0.009
p=0.009
1,291 (197)
1,291 (197)
1,291 (197)
1,291 (197)
1,291 (197)
1; 6.6; 10
1; 6.6; 10
1; 6.6; 10
1; 6.6; 10
1; 6.6; 10
0.07; 0.42; 0.36 p<0.001 p<0.001
0.07; 0.42; 0.35 p<0.001 p<0.001
0.07; 0.42; 0.36 p<0.001 p<0.001
0.64
0.64
0.64
Notes: SE=robust standard errors; * p0.1 ** p0.05 *** p0.01; two-sided tests are used. The number of non-family firm observations changes over the five models.
-0.0200 (0.0129) -0.0159 *** (0.0043) -0.0102 *** (0.0031) -0.0068 (0.0111) -0.0274 * (0.0151) 0.0001 (0.0003) -0.0138 ** (0.0066) 0.0006 (0.0038) -0.0002 (0.0002) p<0.001
-0.0029 (0.0073) -0.0198 (0.0130) -0.0160 *** (0.0043) -0.0102 *** (0.0032) -0.0068 0.0111 -0.0275 * (0.0152) 0.0001 (0.0003) -0.0139 ** (0.0065) 0.0006 (0.0038) -0.0002 (0.0002) p<0.001
0.06; 0.42; 0.36 0.06; 0.42; 0.36 p<0.001 p<0.001 p<0.001 p<0.001 0.64
0.64
6 Family firms and R&D spending
111
Table 6-4: Family versus non-family firms – Bayesian RE regressions of R&D/assets Model I
Independent Variables
Median ß
Prob. (ß>0)
Family firm1 t-1
-0.0107 1.9%
Family firm2 t-1
Model II Median ß
Prob. (ß>0)
Model III Median ß
Prob. (ß>0)
Model IV Median ß
Prob. (ß>0)
Model V Median ß
Prob. (ß>0)
-0.0094 2.5%
Family firm3 t-1
-0.0124 3.9%
Family firm4 t-1
-0.0020 38.7%
Family firm5a t-1
0.0035 32.4%
Ownership share of financial investors t-1
-0.0171 5.0%
-0.0186 3.7%
-0.0168 5.3%
-0.0176 4.7%
-0.0178 4.4%
Firm age
-0.0149 11.7%
-0.0145 14.6%
-0.0127 14.4%
-0.0115 16.8%
-0.0115 16.7%
Firm size t-1
-0.0152 0%
-0.0153 0%
-0.0157 0%
-0.0156 0%
-0.0157 0%
(Cash flow/ assets) t-1 Leverage t-1 Market-to-book value t-1
0.0053 81.2%
0.0051 80.1%
0.0060 84.2%
0.0062 84.9%
0.0062 84.8%
-0.0109 15.5%
-0.0112 14.1%
-0.0116 14.1%
-0.0107 16.2%
-0.0109 15.7%
0.0001 57.0%
0.0001 57.3%
0.0001 51.0%
0.0000 54.1%
0.0000 53.5%
Sales growth
-0.0148 0.1%
-0.0140 0.3%
CEO duality t-1
-0.0001 47.3%
-0.0004 44.6%
0.0004 55.6%
0.0006 57.7%
0.0006 57.9%
CEO duration t-1
-0.0001 32.4%
-0.0000 32.3%
-0.0002 19.5%
-0.0002 22.1%
-0.0002 20.9%
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
1,291 (197)
1,291 (197)
1,291 (197)
1,291 (197)
1,291 (197)
1; 6.6; 10
1; 6.6; 10
1; 6.6; 10
1; 6.6; 10
1; 6.6; 10
479
449
148
111
87
812
842
1,143
1,180
1,204
Industry dummies (17 categories) Year dummies (9 categories)
N obs. (firms) Obs. per group min./avg./max. Obs. family Obs. non-family
-0.0136 0.2%
-0.0147 0.1%
-0.0147 0.1%
Notes: I use normally distributed priors with a mean of zero and a standard deviation of one. Number of draws: 10,000 (the first 1,000 draws are discarded). See Figure A-1 in the Appendix for the MatlabTM code used to estimate the models. The number of non-family firm observations changes since the family firm definitions used change over the five models.
112
6 Family firms and R&D spending
P (ß)
Prob. ß>0=32.4% ; median ß=-0.0035 ; mean ß=-0.0035
ß
Notes: I use normally distributed priors with a mean of zero and a standard deviation of one. Number of draws: 10,000 (the first 1,000 draws are discarded). The estimation is based on a random-effect model with the independent variables as displayed in Table 6-4. See Figure A-1 in the Appendix for the MatlabTM code used to estimate the models.
Prob. ß>0=38.7% ; median ß=-0.0020 ; mean ß=-0.0019
ß
P (ß)
Distribution curve of family firm5a (model V)
Figure 6-1: Family versus non-family firms – Bayesian RE regressions of R&D/assets (continued)
Distribution curve of family firm4 (model IV)
6 Family firms and R&D spending 113
114
6 Family firms and R&D spending
Impact of family ownership and family management
Table 6-5 shows a classical regression that analyzes the impacts of family management and family ownership on R&D spending; Table 6-6 estimates the same model in a Bayesian framework. Figures 6-2 and 6-3 display the results of the Bayesian analysis graphically. The results obtained by the two methods are very similar. Hypothesis 6-1a states that family ownership has a positive influence on R&D spending (patient capital hypothesis); hypothesis 6-1b, on the other hand, predicts a negative influence (entrenchment hypothesis). To test which of these two competing hypotheses better describes reality, two differently specified models are estimated. The first model includes a continuous variable ownership share of family; the second model splits the variable ownership share of family into three indicator variables: ownership share of family 5-19.9% (139 obs.), ownership share of family 2029.9% (54 obs.), and ownership share of family 30% (61 obs.). The results are surprising: There is a non-linear relationship between the level of family ownership and the level of R&D spending. Firms with family ownership between 5% and 19.9% spend an average of 0.8 percentage points less on R&D than firms with no family ownership. However, firms with family ownership 30% spend approximately 1.7 percentage points more on R&D than other firms (see Table 6-6). Hypothesis 61a thus seems to be supported for high levels of family ownership, whereas hypothesis 6-1b seems to hold for low levels of family ownership. Hypothesis 6-2a argues that family management has a positive effect on R&D spending; hypothesis 6-2b asserts the opposite. Controlling for the family’s ownership stake, CEO characteristics, and a number of other factors, I find strong support for the idea that firms governed by a family member invest approximately 1.5 percentage points less of their resources in R&D than firms governed by a non-family manager.204 The empirical results support hypothesis 6-2b and not hypothesis 6-2a. I also looked for a possible interaction effect between family management and the degree of family ownership. No evidence was found for the existence of such an effect: The interaction term is insignificant in the classical model, while the probability of a positive coefficient is 44% in the Bayesian model (model II).
204
The median coefficient in the Bayesian model is -0.015, and the probability of a positive effect is 0.4% (Table 66). The result is similar in the classical regression, in which the coefficient of the variable family management is 0.011 (p<0.05).
6 Family firms and R&D spending
115
Table 6-5: Effect of family firm characteristics on R&D/assets (classical RE model)
Independent variables
Model I
Model II
Model III
Coefficient (SE)
Coefficient (SE)
Coefficient (SE)
Family management t-1
-0.0107 ** (0.0048)
-0.0113 *** (0.0049)
Ownership share of family t-1
0.0205 (0.0215)
0.0006 (0.0409)
Ownership share of family 5-19.9% t-1
-0.0100 ** (0.0047)
-0.0107 ** (0.0051) 0.00234 (0.0078) 0.0142 (0.0095)
Ownership share of family 20-29.9% t-1 Ownership share of family 30% t-1 Ownership share of family t-1 X family management t-1
0.0265 (0.0465)
Ownership share of financial investors t-1
-0.0184 * (0.0105)
-0.0183 * (0.0105)
-0.0183 * (0.0105)
Firm age
-0.0172 *** (0.0040)
-0.0173 *** (0.0040)
-0.0183 *** (0.0040)
Firm size t-1
-0.0100 *** (0.0021)
-0.0099 *** (0.0021)
-0.0096 *** (0.0021)
(Cash flow/ assets) t-1
0.0056 (0.0060)
0.0058 (0.0060)
0.0068 (0.0060)
Leverage t-1
-0.0272 *** (0.0106)
-0.0271 *** (0.0106)
-0.0306 *** (0.0106)
Market-to-book value t-1
0.0002 (0.0003)
0.0002 (0.0003)
0.0001 (0.0003)
Sales growth
-0.0138 *** (0.0047)
-0.0143 *** (0.0048)
-0.0157 *** (0.0047)
CEO duality t-1
-0.0010 (0.0030)
-0.0010 (0.0030
-0.0012 (0.0030)
CEO duration t-1
-0.0001 (0.0002)
-0.0001 (0.0002)
-0.0001 (0.0002)
Industry dummies (17 categories)
p=0.033
p=0.039
p=0.045
Year dummies (9 categories)
p=0.046
p=0.051
p=0.031
N obs. (firms) Obs. per group: min./avg./max. R² within, R² between, R² overall p-value Chi²-test p-value Breusch-Pagan test Rho (fraction of variance due to ui)
1,291 (197) 1; 6.6; 10 0.07; 0.41; 0.35 p<0.001 p<0.001 0.65
1,291 (197) 1; 6.6; 10 0.07; 0.42; 0.35 p<0.001 p<0.001 0.65
1,291 (197) 1; 6.6; 10 0.07; 0.42; 0.36 p<0.001 p<0.001 0.64
Notes: SE=robust standard errors; * p0.1 ** p0.05 *** p0.01; two-sided tests are used.
116
6 Family firms and R&D spending
Table 6-6: Effect of family firm characteristics on R&D/assets (Bayesian RE model) Model I Independent Variables Family management t-1 Ownership share of family t-1
Model II
Model III
Median ß
Prob. (ß>0)
Median ß
Prob. (ß>0)
Median ß
Prob. (ß>0)
-0.0146
0.4%
-0.0147
0.5%
-0.0150
0.3%
0.0299
82.1%
0.0327
72.4%
Ownership share of family 5-19.9% t-1
-0.0080
9.3%
Ownership share of family 20-29.9% t-1
0.0015
55.9%
Ownership share of family 30% t-1
0.0171
93.0%
-0.0149
7.7% 4.7%
Ownership share of family t-1 X family management t-1
-0.0089
44.0%
-0.0156
6.0%
Ownership share of financial investors t-1
-0.0156
7.0%
Firm age
-0.0167
7.7%
-0.0184
8.6%
-0.0219
Firm size t-1
-0.0143
0%
-0.0143
0%
-0.0139
0%
0.0046
77.8%
0.0047
78.4%
0.0055
82.5%
-0.0128
11.4%
-0.0128
12.3%
-0.0156
54.8%
0.0001
58.6%
0.0001
58.5%
0.0001
7.8%
(Cash flow/assets) t-1 Leverage t-1 Market-to-book value t-1 Sales growth
-0.0147
0.1%
-0.0143
0.2%
-0.0169
0%
CEO duality t-1
-0.0016
30.6%
-0.0015
30.9%
-0.0019
27.1%
CEO duration t-1
-0.0001
35.7%
-0.0001
36.0%
-0.0001
35.2%
Industry dummies (17 categories) Year dummies (9 categories) N obs. (firms) Obs. per group min./avg./max.
Included Included
Included Included
Included Included
1,291 (197) 1; 6.6; 10
1,291 (197) 1; 6.6; 10
1,291 (197) 1; 6.6; 10
Notes: I use normally distributed priors with a mean of zero and a standard deviation of one. Number of draws: 10,000 (the first 1,000 draws are discarded). See Figure A-1 in the Appendix for the MatlabTM code used to estimate the models.
6 Family firms and R&D spending
117
Figure 6-2: Effect of family management and family ownership on R&D/assets (Bayesian model) Effect of family management P (ß)
ß Prob. ß>0=0.4% ; median ß=-0.0146 ; mean ß=-0.0147 Effect of ownership share of family P (ß)
ß Prob. ß>0=82.1% ; median ß=0.0299 ; mean ß=0.0300 Notes: I use normally distributed priors with a mean of zero and a standard deviation of one. Number of draws: 10,000 (the first 1,000 draws are discarded). The estimation is based on a random-effects model with the independent variables as displayed in Table 6-6 (model I). See Figure A-1 in the Appendix for the MatlabTM code used to estimate the model.
118
6 Family firms and R&D spending
6 Family firms and R&D spending
119
Figure 6-4: Interaction effect Effect of family management X ownership share of family P (ß)
ß Prob. ß>0=44.0%; median ß=-0.0089; mean ß=-0.0087 Notes: I use normally distributed priors with a mean of zero and a standard deviation of one. Number of draws: 10,000 (the first 1,000 draws are discarded). The estimation is based on a randomeffects model with the independent variables as displayed in Table 6-6 (model II). See Figure A-1 in the Appendix for the MatlabTM code used to estimate the model.
Family firm age
Table 6-7 shows a Bayesian regression that includes only family firm observations (479 obs. from 86 firms).205 The model is estimated to analyze the impact of family firm age on R&D spending. Hypothesis 3a states that the age of a family firm and the level of R&D spending are positively related; hypothesis 3b asserts the opposite. I find evidence for hypothesis 3b. The probability of a positive effect of the variable firm age is only 5.1% (model I). To obtain this result, I control for the fact that younger firms tend to grow faster and are often smaller than older firms (variables sales growth and firm size).
205
The definition family firm1 is used to classify family firms.
120
6 Family firms and R&D spending
Table 6-7: Effect of family firm age on R&D/assets (Bayesian RE model) Model I Prob. (ß>0)
Median ß
Prob. (ß>0)
-0.0031
42.3%
0.0033
21.8%
0.0468
92.2%
0.0952
56.6%
-0.0635
91.2% 18.9%
Independent variables Family management t-1 Ownership share of family t-1
Model II
Median ß
Ownership share of family t-1 X family management t-1 Ownership share of financial investors t-1
-0.0156
19.6%
-0.0169
Firm age
-0.0369
5.1%
-0.0450
4.0%
Firm size t-1
-0.0021
33.8%
-0.0025
32.4%
0.0026
57.1%
0.0013
53.5%
(Cash flow/assets) t-1 Leverage t-1
-0.0362
1.5%
-0.0380
1.2%
Market-to-book value t-1
-0.0004
19.1%
-0.0004
15.9%
Sales growth
-0.0076
12.3%
-0.0079
13.9%
CEO duality t-1
-0.0100
4.8%
-0.0098
5.9%
CEO duration t-1
0.0002
68.9%
0.0002
70.5%
Industry dummies (12 categories) Year dummies (9 categories)
Included Included
Included Included
N obs. (firms) Obs. per group min./avg./max.
479 (86) 1; 5.6; 10
479 (86) 1; 5.6; 10
Notes: I use normally distributed priors with a mean of zero and a standard deviation of one. Number of draws: 10,000 (the first 1,000 draws are discarded). See Figure A-1 in the Appendix for the MatlabTM code used to estimate the models.
Is this negative effect of firm age on R&D spending stronger for family than for non-family firms? In other words, do we observe a kind of founder-effect? Figure 6-5 shows the distribution of the regression coefficient firm age in the subsample of family firms and in the subsample of nonfamily firms. There is a clear negative effect of firm age on R&D spending in the subsample of family firms (Prob. ß>0=4%), whereas the effect in the subsample of non-family firms is neutral (Prob. ß>0=49.2%).206 Family firms seem to lose some of their long-term orientation the older they become, whereas this seems to be not the case with non-family firms; there is no difference as regards long-term orientation between young and old non-family firms.
206
A classical regression would lead to the following result: sample of family firms ßfirm age=-0.20 (p<0.05); sample of non-family firms: ßfirm age=-0.15 (p<0.01).
6 Family firms and R&D spending
121
Figure 6-5: Impact of firm age on R&D/assets in family and non-family firms Effect of firm age on R&D/assets in the sample of family firms
P (ß)
Prob. ß>0=4.0% median ß=-0.0450 mean ß=-0.0438
ß Effect of firm age on R&D/assets in the sample of non-family firms
Prob. ß>0=49.2%
P (ß)
median ß=-0.0003 mean ß=-0.0020
ß Notes: I use normally distributed priors with a mean of zero and a standard deviation of one. Number of draws: 10,000 (the first 1,000 draws are discarded). The estimation is based on a random-effects model with the independent variables as displayed in Table 6-5 (model II). See Figure A-1 in the Appendix for the MatlabTM code used to estimate the model.
Robustness checks
To check the robustness of these results, (1) varying priors have been used in the Bayesian models207, and (2) the ratio of R&D/sales (instead of R&D/assets) has been used as the dependent variable. Results from the latter are reported in the figures below.
207
The results are reported in the notes accompanying the tables.
122
6 Family firms and R&D spending
Figure 6-6: Family versus non-family firms – Bayesian RE regressions of R&D/sales Effect of family firm1
Effect of family firm3 P(ß)
P(ß)
ß
ß Prob. ß>0=0%; median ß=-0.0524
Prob. ß>0=29.3%; median ß=-0.0094
Notes: I use normally distributed priors with a mean of zero and a standard deviation of one. Number of draws: 10,000 (the first 1,000 draws are discarded). The estimation is based on a random-effects model with the independent variables and the number of observations as displayed in Table 6-3. See Figure A-1 in the Appendix for the MatlabTM code used to estimate the model.
Overall, the results are very similar to those obtained using R&D/assets as a dependent variable. Family firms, defined in a broad way (e.g., family firm1), spend fewer resources on R&D than non-family firms. Using a narrow definition (e.g., family firm 3) reduces this effect, but does not lead to the opposite effect of family firms having a higher level of R&D spending than non-family firms. Figure 6-7 displays the results of a Bayesian regression analyzing the effects of family management and family ownership on R&D/sales. The results point in the same direction as those of the regression using R&D/assets. Family management has a negative effect, whereas family ownership has a positive effect. The only difference is in the interaction term, which is now positive (Prob. ß>0=97.2%). To summarize, similar with the results of the regressions of R&D/assets, family firms in general seem not to spend more resources on R&D than non-family firms do. This statement seems to be only true for family firms in which the family has a substantial share of ownership (e.g., above 30%). For other types of family firms, this statement seems not to hold. In contrary: there is some evidence that they might actually have a lower level of R&D spending than other firms.
6 Family firms and R&D spending
123
124
6 Family firms and R&D spending
As another robustness check, I estimated a model in which the different family firm definitions exclude each other and are included in only one model. The results are shown in Table 6-8 below and are as follows: family firms that fall under the broad definition family firm1 but do not classify as family firms using the narrower definition family firm3 invest considerably less resources in R&D as compared to non-family firms (about 1.1 percentage points). The same is true for family firms that classify as being family firms using the narrower definition family firm3 (about 2.3 percentage points). The effect of the variable family firm3 only becomes positive when the group is restricted to those firms in which the family has an ownership share of above 20% (model II: about 1.3 percentage points). The results shown in Table 6-8 confirm to some degree the results of the empirical analyses shown above. First, there is no evidence that family firms using the broad definition family firm1 have a higher level of R&D spending than non-family firms. Second, this is also true for family firms that classify as family firms using the narrower definition family firm3. As in with the results shown in Table 6-4 above, the results even suggest that the opposite might be true: family firms have a lower level of R&D spending as compared to non-family firms. Finally, the results confirm the above findings regarding the positive effect of the level of family ownership on the level of R&D spending. Those firms that classify as a family firm using the definition family firm3 and the level of family ownership is above 20% invest more in R&D as compared to other firms.
6 Family firms and R&D spending
125
Table 6-8: Effect of being a family firm on R&D/assets (alternative family firm definitions included in one model) Bayesian RE model
Bayesian RE model
Model I
Model II
Median ß
Prob. (ß>0)
Median ß
Prob. (ß>0)
Family firm1 t-1 excl. obs. that fall under family firm3 (331 obs. from 70 firms)
-0.0108
1.5%
-0.0103
1.7%
Family firm3 t-1 (148 obs. from 28 firms)
-0.0234
0.4% -0.0270
0%
0.0134
83.1%
Independent variables
Family firm 3 t-1 and ownership by family t-1 5-19.9% (77 obs. from 16 firms) Family firm 3 t-1 and ownership by family t-1 >20% (71 obs. from 15 firms) Ownership share of financial investors t-1
-0.0161
6.3%
-0.0157
6.5%
Firm age
-0.0150
0%
-0.0145
0%
Firm size t-1
-0.0188
4.9%
-0.0162
13%
0.0050
80.1%
0.0067
87.9%
(Cash flow/assets) t-1 Leverage t-1
0.0003
53.9%
0.0001
59.8%
Market-to-book value t-1
-0.0136
10.3%
-0.0147
9.1%
Sales growth
-0.0142
0.1%
-0.0195
0%
CEO duality t-1
-0.0001
31.3%
-0.0002
21.9%
CEO duration t-1
-0.0004
44.2%
-0.0005
43.6%
Industry dummies (12 categories) Year dummies (9 categories) N obs. (firms) Obs. per group min./avg./max.
Included Included
Included Included
1,291 (197) 1; 6.6; 10
1,291 (197) 1; 6.6; 10
Notes: I use normally distributed priors with a mean of zero and a standard deviation of one. Number of draws: 10,000 (the first 1,000 draws are discarded). See Figure A-1 in the Appendix for the MatlabTM code used to estimate the models. The comparison group for the family firm dummies is the category non-family firm.
126
6 Family firms and R&D spending
6.5 6.5.1
Discussion Implications for theory
Causes of corporate myopia
The findings in this chapter contribute to the discussion of the causes of myopic corporate behavior (e.g., Bushee, 1998; Campbell and Marino, 1994; David et al., 2001; Hansen and Hill, 1991; Hirshleifer and Thakor, 1992; Laverty, 1996; Narayanan, 1985; Porter, 1990; Thakor, 1990). The main finding is that corporate governance seems to play an important role. Family firms, defined in a narrow way, invest more in R&D than comparable non-family firms. In addition, I find that family ownership and family management have different effects on the level of R&D spending. Family management has a negative effect, whereas family ownership 30% has a positive effect. A high degree of ownership by financial investors leads to a reduction in R&D spending. The positive effect of family ownership is in line with explanations of myopic corporate behavior that invoke information asymmetry (Thakor, 1990) and fluid and impatient capital (Porter, 1990). The first explanation postulates that managers know more about the firm than investors and are thereby forced to use strong short-term results to signal the quality of their management to the owners. According to the latter explanation, underinvestment in R&D may be the result of the short-term relationship between firms and capital providers. Both explanations suggest a positive relationship between the degree of family ownership and long-term orientation, which is supported by my data. The finding that ownership by financial investors has a negative impact on the level of R&D spending is also in line with these explanations. Contrary to family owners, managers of financial institutions do not possess a deep understanding of the firm’s business model. They view the firm more as a vehicle to earn money for their investors than as a long-term investment; in short, they do not feel connected to the firm and do not have as much of a stake in the firm’s long-term existence. The results in this chapter do not support the managerial opportunism explanation of corporate myopia (Hirshleifer and Thakor, 1992; Narayanan, 1985). This postulates that in order to enhance personal reputation on the external job market or to secure one’s position as a CEO, short-term behavior can be the optimal choice from a CEO’s perspective. Given that it is hard for an owner to lay off a member of her own family, family managers should be in safe positions and therefore do not need to show strong short-term performance in order to secure their jobs or increase their reputation on the external job market. If the managerial opportunism explanation were correct, R&D spending would be higher in firms managed by a family member than in firms managed by a non-family
6 Family firms and R&D spending
127
manager. The results of my empirical analyses, however, show the opposite to be true. After controlling for CEO and ownership characteristics, the analysis shows that R&D spending is lower when a firm is managed by a member of the owning family. This finding may be explained by the limited human resource pool of the owning family (e.g., Bennedsen et al., 2007; Burkart et al., 2003; Perez-Gonzalez, 2006), which significantly restricts the set of qualified candidates for management positions. Since the pool of potential candidates is larger, non-family managers should, on average, have a greater stock of knowledge than family managers. Furthermore, non-family managers should have more varied career experiences than family managers. Both more knowledge and more variation in career experiences lead to higher levels of R&D spending (Barker and Mueller, 2002; Cohen and Levinthal, 1990; Kaplan, 2008). Finally, the results suggest that firms with levels of family ownership between 5 and 20% spend less on R&D than other firms. Thus, it seems that the long-term relationship between family shareholders and their firm not only creates positive effects such as a better understanding of the underlying business model or more effective monitoring, but it also opens up the possibility of generating private benefits of control (e.g., Claessens et al., 2000, 2002; Gomez-Mejia et al., 2001; Morck and Yeung, 2003). In this case, family shareholders may not be emotionally connected to their firms and, like institutional owners, view their firms more as vehicles to earn money rather than as institutions that should prosper in the long term. Family business research Definition of family firms. This chapter offers methodological contributions. The findings
indicate that a large heterogeneity exists among family firms. This justifies the use of several definitions of family firms or the use of continuous measures of family influence, such as the F-PEC (Astrachan et al., 2002; Klein et al., 2005) or the familiness construct (Habbershon and Williams, 1999).208 A dichotomous world consisting of only family and non-family firms certainly does not exist. A large group of firms fall between these two categories. For example, in terms of R&D spending, firms with family ownership between 5 and 20% are close to non-family firms and are fundamentally different from firms with family ownership of 30%. When long-term orientation and transfer of wealth to the next family generation is considered to be an integral part of a family firm definition, then such firms should not be classified as family firms. Performance of family firms. Using the broad definition of a family business that is applied
in some finance papers (e.g., Anderson and Reeb, 2003), I find that family firms are less likely than 208
See Section 2.1 and Table 2-2 for the large number of family firm definitions that have been used; see Section 2.2 for measures of family influence.
128
6 Family firms and R&D spending
non-family firms to behave in a way that maximizes long-term benefits, at least when long-term benefits are measured in terms of R&D spending. Hence, the superior performance of family firms cannot be explained by a higher degree of long-term orientation (Anderson and Reeb, 2003; Bertrand and Schoar, 2006; James, 1999; Le Breton-Miller and Miller, 2006). To make this argument, the definition of a family firm must be narrower. The empirical results in this chapter give some suggestions. Comparing different types of family firms, I find long-term orientation to be positively related with family ownership 30% and negatively with family management and family firm age. It seems that the argument for strong long-term orientation in family firms applies primarily to young, founder-owned family firms, and considerably less to other types of family firms. Interestingly, it can be argued that these young, founder-owned enterprises should not be defined as family firms and should be treated as a separate category (e.g., Daily and Thompson, 1994). My result is in line with Miller et al (2007), who find that it is the lone-founder firms209 that show especially strong performance. The effects of family ownership and family management. The results in this chapter have
some implications for the discussion about incentive and entrenchment effects of family ownership and management (Claessens et al., 2002; McConnell and Servaes, 1990; Shleifer and Vishny, 1997; Stulz, 1988). In a nutshell, family ownership and family management are found to differ with regard to their effects on R&D spending. I find that family management has a negative effect on the level of R&D spending, which is contrary to the incentive hypothesis postulated above (hypothesis 6-2a). Family ownership 30%, on the other hand, has a positive impact on R&D spending, while family ownership between 5 and 20% has a negative impact on R&D spending. The relationship between the degree of family ownership and the level of R&D spending is noticeably non-linear, which may be explained by saying that different effects predominate with different degrees of family ownership. As argued above, family owners may use their strong position as entrenched shareholders to generate private benefits of control not shared by other shareholders (hypothesis 6-1b). They may, for example, use this influence to divert resources from projects with long-term value for the firm (such as R&D spending) to projects that benefit mainly the family. The returns on such a strategy would be higher when the family’s share of common equity was lower. In summary, entrenchment does seem to be associated with family ownership, but only noticeably at low levels of equity ownership. The impact of family firm age. The results in this chapter show that young family firms tend
to spend more on R&D than old family firms. Thus, family firm age has a negative impact on the
209
That is, firms in the first family generation, in which only the founder and not her relatives are involved.
6 Family firms and R&D spending
129
level of R&D spending. This finding is in line with those scholars who argue that multi-generation family firms are characterized by sibling rivalries or conflicts between two family generations (Eddleston and Kellermanns, 2007; Harvey and Evans, 1994), which lead to short-term corporate behavior (hypothesis 6-3b). According to Ward (1987, 2004) and LeMar (2001), family firms in the third family generation face a high risk that they will not be continued as a family firm.210 As already discussed above, these results fit well with the observation that much of the superior performance of family firms is the result of lone-founder firms (Miller et al., 2007). Entrepreneurial orientation in family firms. Keeping the entrepreneurial spirit alive is es-
sential to the survival of a family business and must be properly managed in the firm’s intergenerational transition (Habbershon and Pistrui, 2002; Nordqvist, 2008). Some authors argue that the patient capital provided by family owners creates a hospitable environment for entrepreneurial activities (Aldrich and Cliff, 2003; Rogoff and Heck, 2003). Long-term orientation helps the firm to recognize and exploit opportunities, which are the core elements of entrepreneurship (Shane and Venkataraman, 2000). On the other hand, it is argued that, over time, family firms become hostile to change and follow conservative strategies that limit future growth (Beckhard and Dyer, 1983; Stavrou, 1999; Vago, 2004; Ward, 1987; Zahra et al., 2004). The reasons for such a conservatism are manifold, and they include the founder’s reluctance to hand over the business at the right point in time, inadequate attention given to grooming future leaders, or difficulties in integrating competent non-family employees into the firm. In fact, the organization can end up in a state of ‘strategic simplicity’, in which routines used in the past are used again and again (Miller, 1993). Emotional attachment to the firm may further increase the tendency towards conservatism (Miller et al., 2003). As argued in the beginning of this chapter, R&D spending is correlated with three specific dimensions of entrepreneurial orientation: innovativeness, willingness to take risks, and proactiveness (Lumpkin and Dess, 1996). What do the results in this chapter tell us about the entrepreneurial orientation of family firms? First, the result that younger family firms spend more on R&D than older non-family firms is in line with the contention that, over time, family firms become conservative and less entrepreneurial (Miller et al., 2003; Vago, 2004; Ward, 1987). Second, the results show that family ownership 30% has a positive effect on a firm’s entrepreneurial orientation, while low levels of family ownership have a negative effect. Hence, although there is a strong link between family ownership and entrepreneurial orientation, this link is non-linear. Finally, family managers are not more entrepreneurial than non-family managers: After controlling for the level of family ownership, the analysis
210
In other words, the firm either fails or the family sells the business.
130
6 Family firms and R&D spending
shows that family managers do not invest more in R&D than comparable non-family managers. It seems that family ownership, and not family management, is responsible for the strong entrepreneurial orientation of successful family firms like Corning, mentioned at the beginning of this chapter. 6.5.2
Implications for practice
Non-family shareholders
The results in this chapter demonstrate that the level of R&D spending in firms in which the founding family owns more than 30% of shares is higher than R&D spending in comparable firms. This has two main implications for non-family shareholders that invest in family firms, such as pension funds, banks, insurance companies, or other financial investors. First, managerial behavior in these firms seems to be strongly oriented towards long-term goals. This may be good news for (non-family) investors who prefer this type of firm strategy. On the other hand, this may be bad news for investors who prefer a stable stream of dividends, such as mutual funds. The latter type of investor will find it hard to convince the management of such a firm to pay out dividends when this would reduce the firm’s R&D budget. Second, the results suggest that non-family shareholders should be aware of possible (negative) entrenchment effects in family firms, particularly in those firms where the family’s ownership share is between 5 and 20%. Other stakeholders
Other stakeholders like employees, important suppliers, large customers, and the local community are affected by the firm’s strategic choices, but they are not owners of the firm. The result that firms with a large family shareholder invest more in R&D is important news for them. Members of these groups are less likely to lose their relationship-specific investments. For example, in times of financial distress, employees in R&D departments should be less afraid of losing their jobs; similarly, suppliers who engage in collaborative R&D with the firm should be less concerned about the future of their joint enterprises. For these stakeholders, investing in a relationship with such a family firm is a low-risk strategy. Multi-generation family firms
The result that old family firms tend to spend less on R&D than young family firms supports the view that multi-generation family firms suffer from conservatism and resistance to strategic change. Multi-generation family firms should be aware of this problem and take measures against it. They may wish to follow the example of the German multi-generation family firm Freudenberg
6 Family firms and R&D spending
131
Group211, which has implemented a plan in its family constitution to limit the dividends paid out to shareholders, thereby enabling the management to consistently invest in projects that are important to the future of the firm.212
6.6
Summary and conclusions A stronger long-term orientation is often considered a competitive advantage that family firms
have over non-family firms. Taking R&D spending as an indicator of long-term orientation, I use panel data of US family and non-family firms to analyze whether and under what conditions family firms spend more on R&D than non-family firms. The findings are surprising. No evidence is found that family firms in general invest more resources in R&D than non-family firms. As regards family ownership and family management – the two main characteristics of family firms – it is found that family management is negatively correlated with a higher level of R&D spending, whereas a large degree of family ownership seems to have a positive impact. These findings contribute to the literature on managerial myopia and on family businesses (see Section 6.5.1). The findings have important implications for practitioners, in particular non-family shareholders, important stakeholders, and multi-generation family firms (see Section 6.5.2). The empirical analysis is based on a Bayesian approach, which has only rarely been used so far in management or strategy research. I can show that this approach offers some unique advantages when analyzing the robustness of the results with regard to the definition of key variables – in this case, the definition of a family firm. This chapter is one of the few studies to use large-scale quantitative data to analyze business strategies of family and non-family firms. By using the extent of R&D spending as a proxy for the degree of long-term orientation, family firms using a broad definition are not found to be more long-term-oriented than comparable non-family firms. In fact, family firms constitute a very heterogeneous group. I believe that future research on the impact of family ownership and family management on strategic practices would benefit by viewing family firms as a heterogeneous group containing several important subgroups (see also Miller et al., 2007; Westhead and Howorth, 2007). There are at least four opportunities for future research. One avenue would be to look more closely at the different types of family firms that exist. My dataset is confined to large, publicly listed US firms, which is not representative of all firms. One may ask: Is there a difference in R&D spending between private and publicly listed family firms? What happens to R&D spending when there is a conflict within the owning family? Another direction for future research is to focus on 211 212
See http://www.freudenberg.com (accessed November 6th, 2008). The example is explained in more detail in Sywottek (2008). See also Chapter 1.
132
6 Family firms and R&D spending
other aspects of successful innovation; for example, how does investment in long-term relationships with employees and/or clients affect performance? Or what role does the management of intellectual property (e.g., patents or trademarks) play in the success of the firm? A third possibility would be to address the issue of long-term orientation in innovative firms more holistically, focusing on the individual and organizational factors that determine long-term orientation (e.g., the employee’s inclination to invest in firm-specific knowledge) (Laverty, 1996; Marginson and McAulay, 2008). Finally, a fourth possibility would be to relate differences in long-term orientation or R&D spending to quantifiable differences in financial performance, which would be of interest to finance and management scholars alike. With regard to the main research question of this thesis, it is found that family ownership above a certain level leads to a higher degree of long-term orientation. There is, however, no evidence that such a higher degree of long-term orientation leads to higher financial performance: As Chapter 5 has shown, it is family management and not family ownership that has a positive effect on financial performance.
7 Family firms and employment downsizing213 This chapter is connected in a direct way to the main research question of this thesis: whether family firms are more long-term-oriented than non-family firms are. Under particular circumstances, downsizing can be regarded as a short-term strategy. This is the case if, for the sake of shareholders’ (short-term) profits, the firm engages in deep job cuts and releases the skills and knowledge embodied in their workforce. Thus, if family firms are more long-term-oriented than non-family firms, they should also be less likely to downsize than non-family firms are. Besides comparing family and non-family firms, I also distinguish between family management and family ownership as two dimensions of family firms and analyze their respective influence on downsizing. In general, little is known about the relationship between family firms and their employees. This chapter aims to shed more light on this issue. Section 7.1 introduces and presents the research questions. Section 7.2 summarizes relevant prior empirical studies. In Section 7.3, theory is used to develop hypotheses about the impact of family management and family ownership regarding workforce reduction. Section 7.4 describes the construction of the dataset as well as the measures and methods used. Section 7.5 reports the results of univariate and multivariate analyses. Section 7.6 presents the implications for theory and practice. Section 7.7 summarizes and concludes the chapter.
7.1
Introduction When sales and profits fall, downsizing and cost-cutting are usually among the first manage-
ment reactions. As an example, consider Xerox Corp. and Merck & Co, which announced job cuts of 5% and 12% of their workforce, respectively, in the wake of the recent financial crisis (Chernikoff and Howell, 2008; Pollack, 2008; Uchitelle, 2008).214 Both firms refer to a slowdown in profits and sales as the main reason for the layoffs. Yet, research has shown that large job cuts do not necessarily lead to higher profits or stock prices (see Cascio, 2002; Gerpott, 2007). Downsizing often fails to achieve higher profits because it is used to achieve short-term goals in the face of
213
214
This chapter is based on Block (2008a). The paper was presented at the 2008 annual IFERA conference and the 2008 annual conference of the Verein für Socialpolitik. For an overview about layoffs since the beginning of the financial crisis in August 2008, see http://www.techcrunch.com/layoffs (accessed January 5th, 2008).
134
7 Family firms and employment downsizing
long-term problems such as overpriced products (Appelbaum et al., 1999). Downsizing may, in fact, introduce huge costs. The ‘surviving’ workforce becomes demoralized, and skilled and experienced workers may feel betrayed and thus be more likely to leave. Production schedules are disrupted. Unionization and strikes become more likely (for studies about the costs of downsizing, see e.g. Appelbaum et al., 1999; Cappelli, 2004; Cascio, 1993; Chadwick et al., 2004; O’Neill and Lenn, 1995). This chapter analyzes whether family firms treat their employees differently from nonfamily firms, in particular whether they are less likely to make deep job cuts. Previous research has not led to clear conclusions regarding the relationship between family firms and their employees. Two arguments suggest that family firms are less likely to engage in downsizing as compared to non-family firms. Some scholars argue that family firms pursue a more long-term-oriented strategy than non-family firms do (e.g., James, 1999; Miller and Le Breton-Miller, 2005), and they see this strong long-term orientation as one of the main competitive advantages of family firms (Anderson and Reeb, 2003; Bertrand and Schoar, 2006, Miller and Le Breton-Miller, 2005). Deep, mainly short-term-oriented job cuts do not fit with such a strategy. In a similar way, other authors argue that the large overlap between the family and the firm in terms of people and assets leads to strong feelings of social identity in relation to the firm among family managers (Dyer and Whetten, 2006; Guzzo and Abbott, 1990; Stavrou et al., 2007). Such strong feelings of social identity make deep job cuts less likely. Yet, there exist also counter-arguments. An alternative view describes family owners as being primarily interested in gaining private benefits as a result of control (Claessens et al., 2002; Morck and Yeung, 2003, 2004; Shleifer and Vishny, 1997)215 or as lacking self-control that leads to nepotism (Schulze et al., 2001, 2003)216. Both managerial entrenchment and nepotism are arguments against the idea of a socially responsible family firm. Consequently, according to this view, the likelihood of deep job cuts should not differ between family and non-family firms. These conflicting views lead to the following research questions: (1) Are family firms more or less likely than non-family firms to make deep job cuts? (2) What is the relationship between family management and family ownership, two important dimensions of family firms, and the likelihood of making deep job cuts?
215 216
See the paragraph on the conflicts between dominant and minority shareholders in Section 2.3.1. See the paragraph on the agency costs from altruism in Section 2.3.1.
7 Family firms and employment downsizing
7.2
135
Family firms and their employees: a literature review Little empirical work has been done on the relationship between family firms and their em-
ployees. This section summarizes what is known so far. In a qualitative study, Miller and Le Breton-Miller (2005) find that successful familycontrolled firms aim to minimize layoffs in hard times, and involve those employees that are affected in a direct way in the restructuring process.217 They also find that these firms are very choosy in the hiring process. The goal is to hire only those workers that fit with the (often strong) corporate culture. They want to hire for a career not a (short) job. Accordingly, these firms tend to invest deeply in training and development of their employees so that the employees can develop their full potential. Some authors have analyzed the corporate social responsibility policy (CSR) of family firms. Dyer and Whetten (2006), using a broad definition of family firms comparable to Anderson and Reeb (2003), find that family firms to a greater extent refrain from socially irresponsible actions, supporting the view of family owners as being more interested in a positive reputation. In the same line, Uhlaner et al. (2004) find that the inclusion of the family surname in the business name increases perceived social responsibility. Deniz and Suarez (2005) argue that owning families have a deep sense of personal responsibility towards their employees and the communities they live in. Using a sample of Spanish family firms, they find that family owners often sit on boards of local charitable organizations. However, they also find that in terms of their orientation towards stakeholders, family firms are not a homogenous group, supporting an approach that differentiates between several types of family firms and refers to several dimensions of family firms as variables of interest. In a paper close in nature to this chapter, Stavrou et al. (2007) explore the relationship between family firms and the likelihood of downsizing. They find that family-owned firms generally seem to be less likely to downsize relative to other firms. This chapter extends their study in several ways: I do not rely exclusively on family ownership as the defining characteristic of family firms. It is distinguished between family ownership and family management, understanding them as being two distinct dimensions of family firms. In addition, I use a large number of additional control variables such as ownership share of financial investors, the structure of executive pay and CEO characteristics. Finally, I use longitudinal data, which allow the inclusion of lagged values, therefore making the findings more robust regarding issues of causality. As can be seen in the results section, this chapter’s findings allow a more fine-grained assessment of downsizing in family firms. 217
Miller and Le Breton-Miller (2005) do not provide an explicit definition what they regard as a family firm. Most of their examples, however, are from firms that classify as a family firm using a narrow definition.
136
7 Family firms and employment downsizing
7.3
Theory and hypotheses Before developing the hypotheses, some general comments on downsizing are necessary. This
chapter explores whether family managers and family owners have a stronger commitment to their employees than other types of managers and owners have. More precisely, this chapter investigates the likelihood of downsizing for various types of family and non-family firms. Downsizing is defined as deep job cuts. To measure downsizing, the number of employees as reported by the company in a particular period is compared to the number of employees in the previous period. Downsizing differs from layoffs and from the concept of organizational decline. A layoff is defined as a termination of the employment contract with or without prior notice. Layoffs differ in two ways from downsizing. First, a firm may reduce its workforce (i.e., downsize) without layoffs by relying on incentives for early retirement, outplacement, or non-extension of a fixed-term contract. Layoffs thus represent only one of several operational mechanisms for workforce reduction (Greenhalgh et al., 1988). Second, downsizing is a concept that focuses on the organizational level, whereas investigations of layoffs mostly take place on the individual level (Freeman and Cameron, 1993). There is a wealth of literature on organizational decline (for a summary, see Cameron et al., 1988). In most definitions, organizational decline is described as something that ‘happens’ to an organization, not something intended by the organization or its managers (Freeman and Cameron, 1993). Furthermore, decline does not necessarily involve a reduction in personnel. For example, an organization may experience a decline in market share, but this does not necessarily lead to a reduction in workforce. In my view, a reduction in workforce can be seen as one of several potential responses to decline. To account for organizational decline in my regressions, I use decline in financial performance and decline in sales as control variables. 7.3.1 Family management and its influence on downsizing
This section develops a hypothesis about the impact of family management on the likelihood of deep job cuts. This is done using stewardship and agency theory. Stewardship theory has increasingly been employed in the context of family business research (e.g., Chrisman et al., 2007; Corbetta and Salvato, 2004; Eddleston and Kellermanns, 2007). Stewardship theory, rooted in psychology and sociology, describes situations in which executives, acting as stewards, are motivated to work for the best interests of their organizations (Davis et al., 1997; Donaldson and Davis, 1991).218
218
Stewardship theory is described in detail in Section 2.3.2.
7 Family firms and employment downsizing
137
Contrary to agency theory (Eisenhardt, 1989; Fama, 1980; Jensen and Meckling, 1976), stewardship theory proposes that pro-organizational, collectivist behavior produces a higher utility than individualistic, self-serving behavior does. Davis et al. (1997) compare the management philosophies of stewards and agents. The timeframe of stewards is suggested to be long-term, versus that of agents, which is short-term. Additionally, the main objective of stewards is suggested to be performance enhancement, while cost control is the main objective of agents. Both arguments suggest that managers who act as stewards are less likely to undertake short-term, cost-oriented job cuts. Why should family managers be assumed to act as stewards rather than as agents? Some arguments are listed below. Two of the four arguments relate directly to agency theory, showing how strongly agency and stewardship theory are interconnected. First, with their status and family membership, family managers often stay in the job for lengthy tenures (Le Breton-Miller and Miller, 2006), allowing them to benefit from decisions not to downsize (and thereby not to drain the resources that have been built up in the past). Second, for family managers, the firm is part of the family identity. Often, it is their main intention to pass the firm and its inherent resources (e.g., employees’ knowledge) to the next generation (Casson, 1999; James, 1999; Tagiuri and Davis, 1992). Large job cuts would contradict this goal. Third, due to their rather safe position, family managers do not need to engage in signaling in order to increase their reputation on the corporate executive market. They are therefore not obliged to show regular increases in operating efficiency or profitability. This argument can be extended even further. The fact that family managers are related to the founding (and business-owning) family by kinship ties might prevent them from moving to another (rival) company. Effectively, they do not take part in the market for executives, thus reducing the problems of managerial myopia (Campell and Marino, 1994; Holmstrom, 1982b). Finally, using social identity theory, one can argue that family managers are more likely to be emotionally attached to their firm and their employees than non-family managers are. Generally, social identity theory asserts that group membership can create in-group enhancement in ways that favor group members at the expense of non-group members (Ashforth and Mael, 1989; Tajfel and Turner, 1986). Due to the large overlap between the people and assets of the family and the firm and due to the co-evolution of the two systems (Kepner, 1983), family managers have a strong identification with their firm (Guzzo and Abbott, 1990). Strong feelings of social identity may then lead to stewardship-like behavior that favors the members of the firm over pressures from outside groups or institutions, such as financial analysts or the capital market, thus making deep job cuts less likely. Based on these four arguments, the following hypothesis is proposed:
138
7 Family firms and employment downsizing
Hypothesis 7-1:
7.3.2
There is a negative relationship between family management and the likelihood of deep job cuts.
Family ownership and its influence on downsizing
It is argued that family owners provide ‘patient capital’ (Habbershon and Williams, 1999; James, 1999; Miller and Le Breton-Miller, 2005; Porter, 1992).219 They do not require high quarterly returns, but instead aim to transfer a healthy business to the next family generation. As argued above, deep job cuts as a reaction to declining sales or profits are often shortsighted. Although they may create short-term cost benefits, deep job cuts are often associated with enormous costs over the long-term and may endanger the firm’s ability to produce high-quality products (Appelbaum et al., 1999; Cappelli, 2004; Cascio, 1993; Chadwick et al., 2004; O’Neill and Lenn, 1995). Because of their long-term perspective, family firms are said to be less reactive to economic cycles. If this is true, they should also be less likely to make deep job cuts. Agency theory can also be used (Eisenhardt, 1989; Fama, 1980; Jensen and Meckling, 1976) to form a hypothesis about the role of family ownership in workforce reduction.220 The first argument concerns the relationship between the management and owners of a family firm. It is argued that information asymmetry between owners and management should be less of a problem with family owners than with non-family owners. Unlike other owners, family owners often have known the business for a long time and have a good understanding of the underlying business model and the particulars involved. Consequently, there is less need for a manager (family or non-family) to use sharp short-term changes (e.g., those achieved through means of cost-cutting) as a signaling device (Thakor, 1990). The second argument uses agency theory on a different level by considering the relationship between family owners and society. Family owners now correspond to agents, being monitored and sanctioned by society (which is the principal). In line with Wiklund (2006), it is argued that family owners can be more easily monitored and sanctioned by society than other types of owners can. Unlike financial owners, family owners often have their wealth tied to a particular firm and are more easily identifiable, as they are often well-known and bear the same name as the firm does. Thus, compared to other types of owners, family owners should be more likely to care about their reputation for social responsibility. This greater concern for reputation makes them more fearful of the negative image associated with deep job cuts than non-family owners are. Consequently, the following hypothesis is proposed:
219 220
Refer also to the discussion on long-term orientation in family firms in Section 3.3. Agency theory is described in detail in Section 2.3.1.
7 Family firms and employment downsizing
Hypothesis 7-2:
7.4
139
There is a negative relationship between the extent of family ownership and the likelihood of making deep job cuts.
Data
7.4.1
Sample
To analyze downsizing in family firms, I use a subsample of the dataset described in Section 4.1. The reduced dataset contains only 2,638 observations from 414 firms (full dataset: 4,856 obs. from 499 firms). To construct the subsample, the following ‘data cleaning’ steps are undertaken. First, all observations with missing data on the number of employees are excluded (295 obs.). Next, I lag the independent variables by one year and exclude all observations with missing data (1,923 obs.). The dataset contains the following groups of family and non-family firm observations: nonfamily firm (1,565 obs.), family firm1 (1,073 obs.), family firm2 (1,015 obs.), family firm3 (411 obs.), family firm4 (302 obs.), family firm5a (177 obs.).221 7.4.2
Measures
Dependent variables
The variable workforce decreased is an indicator variable that takes the value of one if the firm’s workforce decreased compared to the previous year and is otherwise zero. Similarly defined are the variables workforce decreased>5%, workforce decreased>6%, workforce decreased>8%, and workforce decreased>10%. Finally, the variable percentage decrease in workforce indicates the percentual change by which the workforce is decreased as compared to the previous period.222 Independent variables
The main interest in this chapter lies in determining the impact of family management and family ownership on the likelihood of downsizing. The variable family management is constructed as an indicator variable that equals one if a member of the founding family is either CEO or chairman. The variable ownership share of family gives the percentage of stock owned by the founding family.
221
222
Note that the groups are not mutually exclusive. For example, observations falling into the group family firm3 also belong to the group family firm1. Note that the variable has a positive sign. For example, if the variable has a value of 4, then this refers to a decrease of 4%.
140
7 Family firms and employment downsizing
Besides these two variables, a large number of controls are included. The variable ownership share of financial investors measures the percentage of stock owned by financial institutions. The following firm-specific variables are used: firm age (number of years since the firm was founded), firm size (value of assets), leverage (value of debt divided by the value of assets), average sales growth in the last five years, and personnel-intensity (number of employees divided by the value of assets). Return on assets (ROA), change in sales (sales decrease or increase divided by sales in the previous period), and market-to-book value are included to account for firm performance, market valuation, and investment opportunities. To control for the impact of corporate restructuring activities such as divestitures or acquisitions, the variable change in property, plant and equipment (PPE) (PPE t-PPE
t-1)
is constructed.223 To measure the impact of the CEO’s strength and experience, the
variables CEO tenure (the number of years the individual is CEO) and CEO duality (an indicator variable that equals one if the CEO also serves as chairman of the board of directors) are included in the regressions. In order to measure the effect of incentive compensation, I calculate the variables share of option-based compensation and share of stock-based compensation. Both variables are measured in percent of total payment. Finally, two-digit SIC codes are used to construct indicator variables for the industries in the sample (53 categories), and time dummies for the years 1994-2003 are used to control for macro-economic effects. As the distributions of the variables firm size and firm age are highly skewed, logarithmic values are taken. Most of the independent variables are lagged by one year to avoid problems of endogeneity. For more details regarding variable construction, see Table A-2 in the Appendix.
7.5
Results
7.5.1 Univariate analysis
Downsizing firms are compared to firms with a stable workforce using descriptive statistics. The unit of analysis is observations. Downsizing observations are defined as those in which a firm’s workforce has declined compared to the previous period by more than 5%.224 The total sample encompasses 2,638 observations (from 414 firms). The downsizing category contains 530 observations (from 262 firms). Note also that there is a substantial variation in the size of the workforce change. The mean percentage change in workforce is about +8% (median is +3%); the standard deviation is 33% (Table 7-1).
223 224
See also Morris et al. (1999). The threshold of 5% has been used in a number of studies analyzing downsizing (e.g., Cascio et al., 1997, Morris et al., 1999). In the multivariate analyses, I also use 0.1%, 6%, 8%, and 10% as thresholds.
7 Family firms and employment downsizing
141
Table 7-2 presents the means and the medians of the independent variables grouped by downsizers and stable employers.225 Table 7-2 also reports the results of tests for equality of means or proportions and the results of Wilcoxon rank-sum tests. The proportion of family-managed firms is found to be higher in the stable than in the downsizing group (38% vs. 28%, p<0.01). In addition, the share of family ownership is higher in the group of stable employers (mean 5.7% vs. 4.0%, p<0.01). Univariate results thus seem to indicate the negative impact of both family ownership and family management on the likelihood of downsizing. Surprisingly, the results of the univariate analysis do not indicate a significant relationship between the degree of ownership by financial investors and the likelihood of belonging either to the downsizing or the stable group (mean 13.9% vs. 13.2%, p=0.17). This contradicts those who argue that the presence of institutional investors influence managers to act based only on the short-term (Johnson and Kaplan, 1987; Stein, 1988). Concerning the remaining firm characteristics, it is found that larger, older, and faster-growing firms, as well as firms with more debt, are more likely to belong to the downsizing group. Some interesting results emerge regarding individual CEO characteristics. The average tenure of a CEO is higher in the stable than in the downsizing group (mean 7.0 yrs vs. 5.7, p<0.01). The univariate results do not indicate a relationship between option-based payment, stock-based payment, or CEO duality and the likelihood of downsizing. Finally, both firm performance and investment opportunities are strong indicators of downsizing decisions. Market-to-book value, ROA, and change in sales are all significantly higher in the group of stable employers than in the group of downsizers.
225
I decided to make a comparison of downsizers versus stable employers and not a comparison of family versus non-family firms for three reasons: First, the comparison justifies the use of such an extensive list of control variables included in the multivariate analysis. Second, the hypotheses are about the effects of family management and family ownership and not family firms versus non-family firms. Finally, a pragmatic reason: there are already a lot of comparisons of family versus non-family firms in the text (see Sections 4.2, 5.2.1, and 6.4). As the sample in this chapter is comparable to the samples used in the other chapters, I don’t expect the results to differ in a substantial way.
0.8 6.7 0.07 0.43 2.2 5.3 6.8
CEO duality b CEO duration b Share of stock-based payment b Share of option-based payment b Market-to-book value b ROA b Change in sales b (in %)
11 12 13 14 15 16 17
-94.9 0 0 0 0.04 -11.9 3.6 0 -23.8 0
0 7.1 0 0.15 0 0.30 0 2.9 0.1 14.0 -458.3 23.9 -483.5
0.13 0.12 13.6 1.1 1.4 0.9 28.0 0.17
32.7
Min.
1 42 0.94 1 77.5 54.8 89.0
6,727 1 0.89 0.86 292.7 19.2 13.2 5.4 491.7 0.95
Max.
-0.03 0.08 -0.04 0.07 0.09 0.26 0.41
0.13 0.08 0.07 -0.03 0.19 -0.19 -0.24 0.36 -0.09
1
-0.20 0.32 -0.12 0.04 0.02 0.18 0.09
0.37 -0.07 0.13 0.01 -0.21 -0.39 0.24 -0.14
2
-0.10 0.08 -0.10 -0.07 0.03 0.11 0.06
-0.23 0.06 0.03 -0.11 -0.10 0.15 -0.10
3
0.03 -0.03 -0.01 0.09 -0.00 0.01 0.04
0.05 -0.03 -0.20 -0.12 0.02 0.02
4
-0.01 0.05 0.07 -0.02 0.08 0.04 0.00
-0.02 -0.20 -0.01 -0.03 -0.08
5
Correlations 7 8 9
-0.02 0.04 0.04 0.00 0.03 -0.01 0.25
0.08 -0.04 0.16 -0.05 -0.15 0.34 0.13
0.15 -0.06 0.16 -0.23 -0.00 -0.35 -0.16
-0.11 0.04 -0.06 0.15 -0.04 0.46 0.30
0.15 -0.01 0.40 0.09 -0.16 -0.45 0.08 0.34 0.27 -0.10
6
0.05 -0.09 0.14 -0.11 -0.09 -0.27 -0.03
10
0.23 0.04 -0.03 -0.05 -0.13 -0.06
11
-0.10 -0.08 0.06 0.03 0.07
12
-0.27 -0.01 -0.07 -0.04
13
15
16 1.82 1.54 1.28 1.33 1.22 2.56 2.29 1.83 1.97
VIF a
1.27 1.46 1.25 1.35 -0.07 1.72 0.18 0.21 1.21 0.01 0.22 0.19 1.38
14
Notes: a VIF=Variance inflation factor, b variables are lagged by one year, c Logarithmized, N=2,638. The variables are reported in the way they go into the multivariate analyses. The Pearson correlation coefficient is used for metric variables, the point-biserial correlation coefficient is used in case one variable is dichotomous, and Cramer’s V is used if both variables are dummy variables. Values greater than 0.04 refer to a p-value smaller than 5%.
8.4 0.33 0.05 0.13 6.3 0.2 8.6 4.0 16.1 0.24
Percentage change in workforce Family management b Ownership share of family b Ownership share of financial investors b Personnel-intensity b Change in PPE/ 1000 Firm size b, c Firm age c Sales growth in last 5 years (in %) Leverage b
Mean Std. dev.
1 2 3 4 5 6 7 8 9 10
Variables
Table 7-1: Summary statistics and correlations
142 7 Family firms and employment downsizing
7 Family firms and employment downsizing
143
Table 7-2: Univariate analysis Workforce downsized a
Variables
Mean Median
Family variables Family management t-1 (yes/no) Ownership share of family t-1 (in %) t-1
Stable workforce b
Workforce downsized vs. stable workforce
Mean Median
Test for Wilcoxon equality of rank-sum means/ test c proportions
0.28 4.0
0
0.35 5.7
0
p=0.002 p=0.006
p=0.140
13.9
12.2
13.2
11.5
p=0.170
p=0.176
6.3 -0.04 22.3 74.9 11.0 26.8
3.7 -0.03 6.9 70 5.3 27.3
6.2 0.31 15.9 70.7 17.3 23.5
3.8 0.06 4.7 76 10.6 22.6
p=0.905 p<0.001 p=0.004 p=0.063 p<0.001 p<0.001
p=0.343 p<0.001 p<0.001 p=0.042 p<0.001 p<0.001
CEO characteristics CEO duration t-1 (in yrs) CEO duality t-1 (yes/no) Share of option-based pay t-1 (in %) Share of stock-based pay t-1 (in %)
5.7 0.81 43.0 7.1
4
5 42.8 0
p<0.001 p=0.524 p=0.471 p=0.487
p<0.001
44.2 0
7.0 0.79 44.0 6.6
p=0.545 p=0.133
Firm performance Market-to-book value t-1 ROA t-1 (in %) Change in sales t-1 (in %)
1.5 0.8 -9.0
1.1 3.0 -2.4
2.3 6.4 10.8
1.5 5.9 9.2
p<0.001 p<0.001 p<0.001
p<0.001 p<0.001 p<0.001
Firm characteristics Ownership share of financial investors %) Personnel-intensity t-1 Change in PPE/ 1000 Firm size t-1 (in bn $) Firm age (in yrs) Sales growth in last 5 years (in %) Leverage t-1 (in %)
t-1
(in
N obs.
530 a
b
2,108
Notes: Workforce decreased by more than 5%. Workforce is increased or decreased by less than 5%. c The Wilcoxon rank-sum test analyzes whether the two samples are from different distributions (sample1: workforce downsized, sample 2: workforce is stable).
144
7 Family firms and employment downsizing
7.5.2 Multivariate analysis
In order to analyze whether family-owned or family-managed firms are more likely to downsize than other types of firms, several random effects logit models have been estimated. The respective dependent variables are workforce decreased>0.1%, workforce decreased>5%, workforce decreased>6%, etc. (model I-V). Table 7-3 below shows the results of these regressions. Before starting with the description of the regression results, I first comment on multicollinearity and unobserved heterogeneity. Table 7-1 above gives summary statistics and correlations for the dependent and independent variables in the regressions. Multicollinearity seems to be of minor concern, as is indicated by the low variance inflation factors (the maximum is 2.56)226. To account for unobserved heterogeneity227, the following steps are undertaken. The individual level error component G i controls for the potential influence of unobserved individual characteristics on the likelihood of large job cuts. I model random individual effects and assume that G i is normally distributed with zero mean and that G i is independent from all observable characteristics. Conveniently, this allows me to measure the extent to which the unobserved individual characteristics influence the dependent variable. The relative importance of G i is measured as U
V 2 i , which is V 2i V 2e
the proportion of the total unexplained variance (²i + ²e) that is contributed by individual-specific effects (²i) (²i is the variance of G i ). Or, in simple words: the panel-level variance component is unimportant. The result is clear. Except in model I (in which workforce decreased>0.1% is used as dependent variable), unobserved heterogeneity does not appear to be a problem (Wooldridge, 2002, pp. 477-478). This is indicated by , which is not significantly different from zero. Note also that a Hausman test (Hausman, 1978) of a random- versus fixed-effects model is not necessary in such a case.228 Now, I turn to the results. Hypothesis 7-1 is not supported. Controlled for firm age, firm size, the CEO’s tenure, and other firm and CEO characteristics, it is found that family-managed firms are not less likely to decrease their workforce than are non-family-managed firms (model III: ß=-0.15, p>0.1). This is true for all five models. Hypothesis 7-2, however, is supported. The variable ownership share of family exhibits a statistically significant negative influence in all models except model 226 227
228
A variance inflation factor of 2.56 corresponds to a tolerance level of 0.61. That is, some unobserved (time-invariant) variables influence both the dependent variables and the independent variables that are of interest. The test would anyway be problematic to implement since many variables of interest do not vary over time (e.g., industry dummies), which would lead them to drop out in a fixed-effects model.
7 Family firms and employment downsizing
145
I (e.g., model III: ß=-1.36, p<0.05). Family ownership thus seems to reduce the likelihood of large job cuts. For example, in firms in which the family owns more than 20% of equity, the odds ratio of A (job cut>10%) over B (no deep job cut) decreases by a factor of 0.59 (p=0.1). Note that the impact of family ownership only applies to deep job cuts; the variable ownership share of family is not found to have a statistically significant effect when it comes to job cuts less than 6% (see model I and II).229 The results concerning the control variables are as follows. Personnel-intensive and larger firms are more likely to engage in deep job cuts. In contrast, firms with a high ROA and a positive development in sales are less likely to downsize. Interestingly, there is a negative relationship between the length of a CEO’s tenure and the likelihood of job cuts greater than 10%. Year dummies are found to be jointly significant, while industry variables are not. Figure 7-1 shows a histogram that displays the distribution of the variable workforce decrease for family and non-family owned firms. This figure illustrates the main conclusion from the multivariate analysis. Family-owned firms shy away from deep job cuts. The distribution of job cuts is more right-skewed in the subgroup of family-owned firms than in the subgroup of non-familyowned firms (skewness 1.1 vs. 0.9; median 4.3 vs. 5).230
Figure 7-1: Histogram of workforce decrease (in %) by family ownership % of obs.
0
Ownership share of family > 10%
5 10 15 workforce decrease in %
N=82; mean=5.3; median=4.3; skewness=1.1
20
% of obs.
0
Ownership share of family < 10%
5 10 15 workforce decrease in %
20
N=792; mean: 6.2; median=5; skewness=0.9
Note: Observations with workforce decrease >20% are not displayed.
229
230
I have also tested for an interaction effect between the variables ownership by family and family management. The results are not very strong and inconsistent over the differently specified models. Therefore, they are not reported. A one-sided t-test is marginally significant (p<0.1), while a Wilcoxon rank-sum test is not (p>0.1).
2,638 (414) 971 1; 6.4; 10 1,309.98 p<0.001 0.041 p=0.035
-0.217 (0.154) 0.007 (0.533) -0.026 (0.509) 0.010 (0.004) ** -0.396 (0.084) *** 0.221 (0.065) *** -0.049 (0.097) -0.003 (0.004) 0.675 (0.443) -0.011 (0.009) 0.190 (0.145) -0.273 (0.199) -0.429 (0.373) -0.083 (0.051) * -0.033 (0.009) *** -0.060 (0.005) *** p=0.023 p<0.001
Model II
2,638 (414) 530 1; 6.4; 10 1,036.50 p<0.001 0.028 p=0.162
ß (SE) -0.168 (0.170) -1.020 (0.634) * 0.148 (0.563) 0.011 (0.004) *** -0.315 (0.086) *** 0.144 (0.071) ** -0.132 (0.105) 0.003 (0.004) 0.755 (0.488) -0.013 (0.010) -0.124 (0.160) -0.248 (0.221) -0.744 (0.435) * -0.077 (0.057) -0.040 (0.009) *** -0.046 (0.004) *** p=0.113 p<0.001
Workforce decreased>5%
Notes: a Variables are lagged by one year b Variable is logarithmized * p0.1 ** p0.05 *** p0.01; two-sided tests are used.
N obs. (groups) N downsizers Obs. per group: min./avg./max. Minus log likelihood Chi²-test Rho () LR-test of Rho=0
Family management a Ownership share of family a Ownership share of financial investors a Personnel-intensity a Change in PPE/ 1000 Firm size a, b Firm age b Sales growth in last 5 years Leverage a CEO duration a CEO duality a Share of option-based payment a Share of stock-based payment a Market-to-book value a ROA a Change in sales a Industry dummies (52 categories) Time dummies (9 categories)
Workforce Dependent variable decreased>0.1% Independent variables ß (SE)
Model I
2,638 (414) 457 1; 6.4; 10 947.96 p<0.001 0.018 p=0.276
ß (SE) -0.146 (0.175) -1.356 (0.673) ** 0.112 (0.586) 0.011 (0.004) *** -0.257 (0.083) *** 0.059 (0.073) -0.084 (0.107) 0.004 (0.004) 0.850 (0.509) * -0.015 (0.011) -0.005 (0.170) -0.315 (0.231) -1.000 (0.466) ** -0.101 (0.063) -0.041 (0.009) *** -0.048 (0.004) *** p=0.238 p<0.001
Workforce decreased>6%
Model III
2,638 (414) 357 1; 6.4; 10 813.20 p<0.001 0 p=0.496
ß (SE) -0.155 (0.185) -1.447 (0.724) ** -0.613 (0.645) 0.013 (0.004) *** -0.229 (0.083) *** 0.071 (0.077) -0.207 (0.113) * 0.001 (0.004) 1.303 (0.540) ** -0.010 (0.011) 0.082 (0.187) -0.271 (0.250) -0.534 (0.494) -0.053 (0.064) -0.043 (0.009) *** -0.046 (0.004) *** p=0.406 p=0.009
Workforce decreased>8%
Model IV
Table 7-3: Random-effects logit regressions of workforce decreased (>0.1%, >5%, >6%, >8%, and >10%) Model V
2,638 (414) 281 1; 6.4; 10 682.24 p<0.001 0 p=0.495
ß (SE) -0.039 (0.203) -2.083 (0.856) ** -0.496 (0.857) 0.011 (0.004) *** -0.240 (0.090) *** 0.015 (0.086) -0.170 (0.122) 0.000 (0.005) 0.557 (0.600) -0.029 (0.014) ** 0.216 (0.211) -0.014 (0.282) -0.076 (0.543) -0.095 (0.073) -0.036 (0.009) *** -0.046 (0.004) *** p=0.712 p=0.002
Workforce decreased>10%
146 7 Family firms and employment downsizing
7 Family firms and employment downsizing
147
Robustness checks
In this section, I aim to find out whether a linear model produces similar results as the logit models shown above. The problem is that the results of a linear model cannot be simply compared to the results of the logit regressions. A linear would model also include observations from firms that have increased their workforce and it would be impossible to distinguish between the effects of a particular variable on workforce increase or workforce decrease. A way out would be to include only observations from firms that have cut their workforce and estimate a linear model with these observations. However, this would introduce a great selection bias, which can lead to wrong conclusions. Heckman models are considered a way to deal with such problems related to sample selection (Heckman, 1979; Heckman et al., 1999).231 Heckman models usually consist of two steps. In the first step, the researcher estimates a model for the probability of belonging to the estimation sample (usually a logit or a probit model). Estimates of this model can then be used to predict the individual probability of each observation to be part of the estimation sample. In the second step, the researcher corrects for selection by including a transformation of these probabilities as an additional independent variable in the estimation equation. This way, the coefficients of the independent variables in the estimation equation are corrected for sample selection. In more detail, to test whether the coefficients in the linear models are influenced by selection, I estimated a two-step Heckman model (Heckman, 1979), with the percentage change in workforce as the dependent variable and a logistic regression model of workforce decrease>0.1% as the selection equation. Table 7-4 shows this Heckman model. The selection equation differs from the estimation equation by the variable personnel-intensity.232 The results of the Heckman model show that selection bias is an issue. Rho (), which measures the correlation between the error terms of the selection and the estimation equation, is significantly different from zero (=-0.59, p<0.001). The results regarding the main variables of interest are confirmed. Both family management and family ownership are found not to have an impact on workforce decrease>0.1% (selection equation: ß=0.14, p>0.1; ß=0.08, p>0.1); family ownership is found to decrease the level of job cuts given that the firm has decided to engage in downsizing (estimation equation: ß=-5.5, p<0.1). That is, an increase of family ownership by 10% leads to a decrease in the extent of downsizing by 0.6 percent-
231
232
For these selection models, James J. Heckman received the Nobel Prize in Economics in the year 2000 (together with Daniel McFadden). This makes sense because the variable is found to have a significant impact on the likelihood of downsizing (see Table 7-3) but is not found to have an impact on the degree of downsizing (ß=0.04 with p>0.1).
148
7 Family firms and employment downsizing
age points. The linear equation of the heckman model thus confirms the direction of the effects of the logit regressions shown above. Effect of being a family firm
As in the previous chapters, I also estimated regression models including the family firm definitions family firm1-5a and 5b as independent variables.233 However, in contrast to the results in Chapter 6, no significant difference exists between family and non-family firms. Note also that the use of a different family firm definition does not change these findings. Family firm age also seems not to have an impact (family firm5a refers to young family firms). To save space, the results are not reported in this thesis. Interested readers, however, may request the respective regression tables from the author.
233
For the family firm definitions, see Table 2-2.
7 Family firms and employment downsizing
149
Table 7-4: Heckman model Selection equation (Dep. variable: workforce decreased>0.1%)
Estimation equation (Dep. variable: percentage decrease in workforce)
Variables
E (SE)
Family management t-1
-0.141 (0.092)
1.324 (1.002)
Ownership share of family t-1
-0.080 (0.311)
-5.540 (3.074) *
Ownership share of financial institutions t-1 Change in PPE/ 1000 Firm size t-1
0.001 (0.289)
-0.355 (2.870)
-0.160 (0.067) **
0.752 (0.351) **
0.106 (0.043) **
-0.950 (0.304) ***
Firm age
-0.021 (0.059)
Sales growth in last 5 years t-1
-0.001 (0.003)
Leverage t-1
E (SE)
-0.199 (0.514) 0.059 (0.022) ***
0.329 (0.264)
-1.667 (2.474)
CEO duration t-1
-0.006 (0.005)
CEO duality t-1
0.119 (0.084)
-0.092 (0.923)
Share of option-based payment t-1
-0.107 (0.115)
-0.727 (1.281)
Share of stock-based payment t-1
-0.270 (0.234)
-0.447 (2.162)
Market-to-book value t-1
-0.045 (0.028)
ROA t-1
-0.018 (0.006) ***
-0.043 (0.011) ***
Change in sales t-1
-0.032 (0.004) ***
-0.089 (0.019) ***
Personnel-intensity t-1
-0.087 (0.053) *
0.130 (0.291)
0.006 (0.003) **
Industry dummies (52 categories)
p<0.001
p<0.001
Time dummies (9 categories)
p<0.001
p<0.001
N obs. Rho Wald test (Rho=0) Minus log pseudolikelihood LR test
2,638
971 -0.584 p<0.001 4,782.76 p<0.001
Notes: * p0.1 ** p0.05 *** p0.01; standard errors are robust and clustered; two-sided tests are used.
150
7.6 7.6.1
7 Family firms and employment downsizing
Discussion Implications for theory
Family firms and employment downsizing
One of this chapter’s central findings is that family management and family ownership have different impacts regarding the respective firms’ engagement in downsizing. Family management is found to have no effect with regard to downsizing, whereas family ownership decreases the likelihood of downsizing.234 This chapter contributes to the debate on whether family firms are more socially responsible towards their employees than non-family firms are (e.g., Dyer and Whetten, 2006; Stavrou et al., 2007; Wiklund, 2006). The main conclusion is that there is not a clear answer to this question. Any study that aims to analyze employment downsizing in family and non-family firms will run into difficulties of finding a ‘clear’ definition of a family firm. A polarized approach that only compares family and non-family firms runs into the danger of producing incomplete results. It seems more promising instead to have a closer look at the different dimensions that characterize family firms, such as family ownership and family management. The finding that family ownership makes deep job cuts less likely is in line with those authors who argue that family owners provide ‘patient capital’ (Habbershon and Williams, 1999; James, 1999; Miller and Le Breton-Miller, 2005). Deep job cuts as a reaction to declining sales or profits are often shortsighted and may lead to huge costs over the long term. In addition, they may endanger the firm’s ability to produce competitive products (Appelbaum et al., 1999; Cappelli, 2004; Cascio, 1993; Chadwick et al., 2004; O’Neill and Lenn, 1995). The finding is also in line with those who argue that family owners can be better monitored and sanctioned by society than other owners can be. Family owners usually have their wealth tied to a particular firm and are thus more easily identifiable than other types of owners. In many cases, they bear the same name as the firm. Compared to other types of owners, family owners should thus be more likely to care about their reputation for social responsibility, in particular with regard to the local community, in which the firm is located. This greater concern for reputation makes them more fearful than other owners of the negative image associated with deep job cuts. Previous literature has also argued that family owners use their firm as a vehicle to generate private benefits of control not shared by other shareholders, e.g., through the use of tunneling or
234
No difference is found between family and non-family firms (using the definitions family firm1-5).
7 Family firms and employment downsizing
151
self-dealing (Johnson et al., 2000; Morck and Yeung, 2003, 2004).235 In addition, it has been argued that family firms suffer from agency costs of altruism (Schulze et al., 2001, 2003). In this view, family CEOs have a great capacity to make altruistic transfers that benefit the members of the owning family. Other family members in turn have an incentive to overstate their needs in order to receive higher transfers, a practice which may put family CEOs in a difficult position. Both arguments suggest that family management, and partially also family ownership, increases the likelihood of short-term job cuts. Although these arguments make sense, the empirical results in this chapter do not support these views. Agency or stewardship theory?
This chapter’s findings are more supportive of an agency view than a stewardship view of the family firm. In the arguments concerning the relationship between family ownership and downsizing, agency theory is used. It is argued that, from the perspective of society, family owners are easier to monitor and sanction. The reason for this is that family owners are not faceless and are more visible to the public than other types of owners such as financial investors. I find some support for this argument in the fact that family ownership decreases the likelihood of deep job cuts above 6%. Furthermore, it is argued that there is a low information asymmetry between owners and management in family firms, making managerial myopia and short-term-oriented job cuts less likely. Stewardship theory is used to argue that family managers differ from non-family managers in their behavior. Due to the large overlap between family and place of work, family managers should be more likely to have strong feelings of social identity than do non-family managers (Guzzo and Abbott, 1990). Strong feelings of social identity should lead to stewardship-like behavior that favors the members of the family firm (i.e., its employees) over pressures from the (outside) capital market, which in turn should make downsizing less likely. The findings in this chapter, however, do not support such a view. Family managers are not found to differ in their behavior from non-family managers, at least when it comes to making deep job cuts. 7.6.2
Implications for practice
Downsizing and its consequences
The extant literature suggests that pure employment downsizing seems to have neither a positive impact on shareholder value (Cascio et al., 1997; Worrell et al., 1991; for a summary, see Ger-
235
See the paragraph on the agency costs due to a conflict of interest between dominant and minority shareholders (Section 2.3.1).
152
7 Family firms and employment downsizing
pott, 2007) nor a positive impact on profitability measures such as ROA (Cascio et al., 1997). Two interesting and related questions now arise: (1) Why do firms, often as a first reaction to decline in sales, engage in employment downsizing if, on average, it does not lead to significant improvements in terms of shareholder value or profitability? (2) Why, according to this chapter’s findings, are family-owned firms less likely to make deep job cuts? This chapter’s findings, as well as the arguments in the theory section, help to answer these questions. It is suggested that family owners provide ‘patient capital’. In contrast to many institutional investors, their main goal is to enhance firm value over the long term. Consequently, they do not force managers to act just for the sake of acting. They only consider deep job cuts as an option when the firm’s survival is at stake. From the perspective of corporate governance, the question is how to stop managers from engaging in value-destroying (or at least non-value-enhancing) programs of downsizing. I suggest that, in a situation of falling sales or profits, investors should not overstate short-term results. Otherwise, the management will focus on cost-cutting. In addition, one might argue that, even in times of crisis, it might not be a good idea to replace a manager who grew up in the organization with an outsider, as the latter is more likely to make deep job cuts and endanger the firm’s long-term competitive position. Family-owned firms as attractive employers
This chapter’s findings relate directly to the attractiveness of family firms as employers. Concerning job cuts, family-owned firms seem to be more stable employers than other types of firms. In particular, employees who undertake high relationship-specific investments (e.g., employees in an R&D or a specialized sales department) benefit from this relatively greater job stability. Familyowned firms could use this higher degree of job stability as an argument when recruiting (specialized) personnel.
7.7
Summary and conclusions This chapter analyzes downsizing in family firms. I define downsizing as deep job cuts (above
5%). To measure these deep job cuts, the number of employees as reported by the company in a particular period is compared to the number of employees in the previous period. Using agency and stewardship theory, it is argued that both family ownership and family management, two dimensions of family firms, reduce the likelihood of making deep job cuts. The chapter extends the literature on the relation between family firms and their employees. Contrary to most other studies, this chapter’s findings relate to different types and dimensions of family firms. The main finding is that the impact of family management and the extent of family
7 Family firms and employment downsizing
153
ownership have different effects. Family ownership is found to reduce the likelihood of downsizing, while family management does not. These results indicate strong differences regarding social responsibility within the population of family firms, differences that should be accounted for in further studies. This work could be extended in various ways. A promising avenue would be to look at the individual level of the employee. Using linked employer-employee datasets, already widely used in labor economics, could help to answer questions such as: (1) Are employees in family firms more satisfied with their job than employees in non-family firms are? (2) Do family firms invest more in their employees than non-family firms? (3) What happens when the skills of the workers and the skills that the firm requires no longer match? Do they invest in training of their workers or do they refer to other measures, such as outside recruiting or outsourcing? (4) What is the level and structure of salary in family and non-family firms? The findings of this chapter support the findings of the previous chapter regarding the level of R&D spending. Again, it is family ownership and not family management that has a positive effect on long-term orientation. Another result is in line with the findings of the previous chapter. Family firms in general are not found to be more long-term-oriented, as measured by the likelihood to make shortsighted job cuts.
8 Family firms and executive pay236 The chapter analyzes the structure of executive pay in family and non-family firms. It is connected to the overall research question of long-term orientation in family firms in that incentive pay is regarded as a source of myopic managerial behavior. If family firms indeed pursue more longterm goals, this should be reflected in the structure of executive pay. Accordingly, there should be a low share of short-term incentive pay and a high proportion of base salary in total pay. Section 8.1 introduces. Section 8.2 summarizes related literature. The data are described in Section 8.3, and the results of the empirical analyses are reported in Section 8.4. Section 8.5 discusses implications of the findings from a research and a management perspective. Finally, Section 8.6 gives a brief conclusion.
8.1
Introduction The relationship between CEO pay and performance has been a central and recurrent issue in
the management and economics literatures alike (e.g., Frey and Osterloh, 2005; Hall and Liebman, 1998; Jensen and Murphy, 1990a; Tosi et al., 2000). In 1990, Jensen and Murphy argued that executive pay is virtually independent of performance and that, on average, most CEOs are paid like bureaucrats (Jensen and Murphy, 1990b). With the strong increase of stock options in CEO pay over the last decades, however, the situation has changed: The link between performance and executive pay has become much stronger (Hall and Liebmann, 1998; Hall, 2003; Hall and Murphy, 2003). The discussion in the media and the academic literature has also turned and now concerns as well the adverse effects of an overly strong link between executive pay and firm performance (e.g., Frey and Osterloh, 2005; Useem, 2003; The Economist, 2006b). It is stated that an overly strong link between executive pay and stock performance leads managers to focus too strongly on shortterm profits at the expense of long-term opportunities (Fuller and Jensen, 2002), and that they neglect to pay dividends to the firm’s shareholders (Lambert et al., 1989). So far, however, the discussion has largely ignored the fact that a sizeable number of publicly listed firms in the US and other
236
This chapter is based on Block (2008b). The paper was presented at the TIME Kolloquium, which is jointly organized by Technische Universität München and Ludwig-Maximilians-Universität München.
8 Family firms and executive pay
155
industrialized countries are family firms.237 This chapter aims to close this gap and analyzes executive pay in family firms, in particular its respective shares of base salary, annual bonus, and stock options. The main focus of this chapter is on the role of incentive pay. From an agency theory perspective, it is unclear whether family firms have a higher or a lower share of incentive pay as compared to non-family firms. On the one hand, standard principal-agent theory predicts a low share of incentive pay. Family firms often resemble owner-managed firms, in which an agency conflict between owners and management is unlikely to occur. If a manager of a firm owns a large number of the firm’s shares, then (risk aversion and self control problems aside) the manager’s decision would be presumed to maximize long-run shareholder value. Moral hazard should not be a problem and bonding mechanisms are not necessary (Jensen and Meckling, 1976).238 The second argument concerns monitoring. Since family owners typically own large blocks of shares, they have a strong incentive to engage in monitoring. Economic benefits from monitoring are high; a free-rider problem associated with firms with only dispersed shareholders does not exist (Fama, 1980; Maug, 1998).239 Since monitoring and bonding are substitutes to some degree, incentive pay should be lower in family firms as compared to non-family firms. On the other hand, when executive pay in general and incentive pay in particular is regarded as a potential source rather than a solution to an existing agency problem (Bebchuk and Fried, 2003), an alternative view is possible. Clever executives have the power to manipulate the remuneration process to benefit themselves at the expense of the company (Bebchuk et al., 2002). In particular, stock- and stock option-based pay is being criticized as being vulnerable to manipulation (Dechow et al., 1996). It has been argued that managerial rent extraction explains both rising trends in pay levels and several (inefficient) features in option plans benefiting mainly the CEO (such as that exercise prices of options are usually not indexed to general market movements). This managerial rent extraction hypothesis is argued to apply in particular to incumbent executives with long tenures with a strong position in the firm, and not to executives hired from outside (Hall and Mur-
237
238
239
In the US, depending on the exact definition used, the share of family firms among the Fortune 500 firms ranges from 7 to 37% (Villalonga and Amit, 2006). For an international comparison on the importance of family firms, see La Porta et al. (1999). Note that Schulze et al. (2001) take a different view and argue that private ownership and owner-management do not eliminate the agency conflict but just create another type of agency conflict. In their view, family firms are exposed to a self-control problem caused by the altruism of family managers towards members of their own family. In such a situation, bonding mechanisms would then help to better align the interests of a family manager with the interests of the firm and its non-family shareholders. Shivdasani (1993), Bertrand and Mullainathan (2001) and Dennis and Serrano (1996) provide empirical evidence on the efficacy of monitoring by large shareholders.
156
8 Family firms and executive pay
phy, 2003). Family executives resemble very much these incumbent executives. Almost by definition, family executives are not hired from outside. They are in a strong position to manipulate the remuneration process. Since they have known the business for a long time (they have often grown up with the business), they have a strong information advantage over non-family board directors. Furthermore, due to their relatively strong position in the firm (they are often important shareholders themselves), they have a strong influence on the composition of the remuneration board. Thus, a family CEO might use her firm as a vehicle to generate private benefits of control, which are to the detriment of the company and its less powerful shareholders (Claessens et al., 2000, 2002; GomezMejia et al., 2001; McConnell and Servaes, 1990; Morck and Yeung, 2003; Stulz, 1988). The chapter’s findings support the argument that agency conflicts between owners and management are lower in family firms relative to non-family firms and do not support the hypothesis of managerial rent extraction. Using several definitions of family firms, it is found that CEOs in family firms have a higher share of base salary and annual bonus and a lower share of stock options in their compensation package as compared to CEOs in non-family firms. Note that stock options in particular are said to be vulnerable to managerial rent extraction. Thus, there seems to be a stronger alignment of interests between the CEO and the firm in family firms than this is the case in nonfamily firms. The main contribution of this chapter is to show that family firm characteristics actually make a difference in regards to the structure of executive pay. Thus far, the discussion about the use and effectiveness of particular components of executive pay such as stock options has largely ignored family business variables (Dittmann and Maug, 2007; Hall and Murphy, 2003).
8.2
Related literature I summarize empirical literature about the relation between executive pay and corporate gov-
ernance, with a focus on studies about family firms. For more comprehensive reviews about executive pay in general, I refer to Murphy (1999). Bertrand and Mullainathan (2001) find that executives in better-governed firms are less likely to be rewarded for luck.240 In particular, they show that adding a large shareholder to the board decreases the sensitivity of pay to luck by 23 percent. Moreover, in another publication, they can show that well-governed firms are more likely to structure their option payment in an effective way (Bertrand and Mullainathan, 2000). Since rewarding luck is inefficient from a principal’s point of view,
240
As an example consider CEOs in petroleum firms whose pay increases due to windfall profits resulting from an increase in world oil prices.
8 Family firms and executive pay
157
they conclude that CEO pay is more effective when there are in fact individuals around to act as principals, which is often the case in family-owned firms. Empirical research on CEO pay in family firms so far has mainly compared family-member CEOs with non-family CEOs working in family firms. There is evidence that family-member CEOs generally receive lower levels of compensation than non-family CEOs who work in family-owned firms (Gomez-Mejia et al. 2003; McConaughy, 2000). This effect becomes stronger with an increase in family ownership (McConaughy, 2000). Regarding incentive pay, Chrisman et al. (2007) and Schulze et al. (2003) find that family executives receive incentive compensation and that it has a positive effect on performance, which suggests that there exists also agency behavior among family executives.241 However, McConaughy (2000) shows that family CEOs receive less incentivebased pay than non-family CEOs who are employed by family firms (McConaughy, 2000). So far, little research exists which compares executive pay of family and non-family firms. This chapter aims to close this gap.
8.3 8.3.1
Data Sample
To analyze executive pay in family and non-family firms, I use a subsample of the dataset described in Section 4.1. The reduction is due to excluding observations from the year 2003242 and due to missing values in the compensation variables and some covariates. The reduced dataset contains 2,578 observations from 393 firms (full dataset: 4,856 obs. from 499 firms) and consists of the following groups of family and non-family firm observations: non-family firm (1,488 obs.), family firm1 (1,090 obs.), family firm2 (1,031 obs.), family firm3 (430 obs.), family firm4 (316 obs.), family firm5a (190 obs.).243
241
242
243
Schulze et al. argue that family CEOs suffer from a self-control problem caused by the altruism towards members of their own family. Incentive pay then helps to align the interests of the family manager with the interests of the firm (see the paragraph on the agency costs of altruism in Section 2.3.1). Observations from the year 2003 are excluded, since I can only make a statement about the CEO’s status as a member of the owning family from January to July 2003 (see also the construction of the dataset in Section 4.1). The compensation data relate to the the entire year. Note also that unlike the regressions in the previous sections, this section does not use lagged values for most of its independent variables. Note that the different groups are not mutually exclusive. For example, observations falling into the group family firm3 also belong to the group family firm1. See also Table 2-2 above.
158
8.3.2
8 Family firms and executive pay
Measures
The following measures are used as dependent variables: total pay, share of base salary, share of annual bonus, and share of stock option payment244 (the latter three measures are all in percent of total pay). Table A-2 in the Appendix gives more details about the exact ExecuComp data items that were used. The independent variables can be categorized into variables relating to family firms and control variables. The variable family CEO is an indicator variable that equals one if a member of the founding family is CEO and is zero otherwise. The variable ownership share of family gives the percentage of stock owned by the founding family. The variables family firm1-5 indicate family firms as defined by the definitions in Table 2-2 above. To distinguish between family ownership and ownership by financial investors, the variable ownership share of financial investors was constructed. The variable measures the percentage of stock owned by financial investors such as large banks, insurance companies, or mutual funds. To control for other firm-, industry-, and timespecific influences, many more variables are included in the regression models. Table A-2 in the Appendix describes their construction. 8.3.3
Regression models
Quantile regressions are used in addition to GLS random-effects regressions. These regressions estimate conditional quantile functions, that is models in which quantiles245 of the conditional distribution of the dependent variable are expressed as functions of several independent variables (Koenker and Bassett, 1978).246 Quantile regressions are more robust than OLS regressions to problems with outliers in the dependent variable (Koenker and Hallock, 2000). Quantiles such as the median are robust to outliers. This is not true for the mean, which is used in OLS regressions. Outliers are often a great problem in executive compensation data. This is also the case in my dataset, which is clearly illustrated in the histograms shown in the next section (e.g., Figure 8-1). Specifically, I estimate quantile regressions on the median of the dependent variable, to which I refer to hereafter as median regressions.
244
245
246
The value of options is determined by ExecuComp’s Black-Scholes calculations. The value is calculated at the date the option is granted. Quantiles are points taken at regular intervals from the cumulative distribution of a random variable. The most widely used quantile is the median (50%-quantile). For an introduction to quantile regressions, see Koenker and Hallock (2000, 2001).
8 Family firms and executive pay
8.4 8.4.1
159
Results Descriptive statistics and univariate analysis
Table 8-1 gives descriptive statistics about the level and structure of executive pay in the sample. A median CEO earns $4.4 mn (mean: $8.2 mn). Yet, with the lowest CEO pay at $0.001 mn and the highest CEO pay at $293 mn, the range of executive pay is large (the standard deviation is $15.52 mn). Figure 8-1 below shows a box plot of the variable total pay, which depicts the large number of outliers that exist. Figure 8-2 displays the development over time. Executive pay rose steadily from $2.2 mn in 1994 to $6.8 mn in 2001 and dropped thereafter slightly to $6.2 mn in 2002. A large portion of this increase is explained by the increased use of stock options, which usually comes as additional pay (i.e., if it is granted, it does not lead to a reduction in pay in the other components).247 Regarding the components of executive pay, the results are as follows: the median CEO base salary is $0.78 mn (about 18% of total pay), the median annual bonus is $0.66 mn (about 17% of total pay), and median stock option pay is $1.78 mn (about 44% of total pay). Compared to the other components of executive pay, stock pay and pay due to long-term incentive plans are of lesser importance. Most of the variance in executive pay comes from stock options (the coefficient of variation is 2.74). The histograms in figure 8-3 illustrate the large variance that exists with regard to the structure of executive pay. With annual bonus, stock pay and stock option pay, there is a peak at 0%. Obviously, not all firms in my sample rely on all components of executive pay. For example, about 18% of the compensation packages do not include stock options at all. Table 8-2 uses t-tests to compare the structure of executive pay in family versus non-family firms. In regard to family management, it is found that, on average, family CEOs have a higher share of base salary (28% vs. 22%, p<0.01), a lower share of stock pay (4% vs. 7%, p<0.01), and a lower share of pay due to long-term incentive plans (2% vs. 5%, p<0.01). Regarding family ownership, it is found that, in firms where a family shareholder is present who has 5% or more of shares, the CEOs have a higher share of base salary (30% vs. 22%, p<0.01) and annual bonus (21% vs. 19%, p<0.01). Furthermore, it is found that CEOs in firms with a family shareholder have a lower share of pay due to long-term incentive plans (2% vs. 5%, p<0.01), a lower share of stock pay (3% vs. 7%, p<0.01) and a lower share of stock option pay (41% vs. 44%, p<0.05).
247
The upward trend is comparable to the numbers in Hall and Murphy (2003). However, since I use median CEO pay and not mean CEO pay, the trend is slightly less dramatic.
160
8 Family firms and executive pay
Figure 8-1: Box plot of total pay
0
50
100
150 200 Total pay (in mn $)
250
300
Notes: The box plot provides a summary of the variable total pay. The left and right limits of the box represent the first and third quartiles of total pay (Q0.25 = $2.2 mn; Q0.75 = $8.6 mn). The median is shown by the vertical line in the middle of the box (Q50 = $4.4 mn). The extreme left line gives the minimum; the extreme right line is calculated as 1.5 times the interquartile distance plus the third quartile. Outliers are represented by diamonds.
0.79 0.99 0.33 0.69 5.22 0.21 8.22
Base salary Annual bonus Long-term incentive plan Stock pay Stock option pay Miscellaneous other pay a Total pay
in mn $
0.78 0.66 0 0 1.78 0 4.41
Median 0 0 0 0 0 0 <0.01
Min. 3.65 16.50 29.25 66.99 290.59 96.33 293.10
Max. 0.37 1.35 1.42 2.88 14.32 2.10 15.52
Std. dev. 24% 19% 4% 6% 44% 3%
Mean 18% 17% 0% 0% 44% 0%
Median 0% 0% 0% 0% 0% 0%
Min. 100% 100% 82% 94% 99% 100%
Max.
Structure of executive pay (share of total pay in %)
0
1
2
3
4
5
6
7
1994
1995
1996
Total pay
1997
1999 Stock options
1998
2000
2001
Figure 8-2: Median CEO pay over time
2002
21% 17% 11% 15% 31% 9%
Std. dev.
1994: 21% 1995: 25% 1996: 30% 1997: 33% 1998: 40% 1999: 48% 2000: 47% 2001: 59% 2002: 47%
Share of stock options:
Notes: N=2,578 obs. from 393 firms a Miscellaneous other pay includes the ExecuComp data items ALLOTHPD (all other paid) and OTHANN (other annual).
Mean
Components of executive pay
Level of executive pay (in million $)
Table 8-1: Level and structure of executive pay
8 Family firms and executive pay 161
obs. in %
20
15
10
Share of base salary (in%)
obs. in %
0% Share of stock pay Notes: N=2,578; stock pay is 0% in 1,957 cases. mean=6%, median=0%
Share of stock pay (in%)
0% Share of stock option pay 100% Notes: N=2,578; stock option pay is 0% in 457 cases. mean=44%, median=44%
Share of stock option pay (in%)
5
0
10
15
20
obs. in %
0% Share of base salary Notes: N=2,578; base salary is 0% in 21 cases. mean=24%, median=18%
100%
obs. in %
0% Share of annual bonus Notes: N=2,578; annual bonus is 0% in 376 cases. mean=19%, median=17%
Share of annual bonus (in %)
Figure 8-3: Histograms regarding the structure of executive pay
0
5
10 8 6 4 2 0 80 60 40 20 0
100%
100%
162 8 Family firms and executive pay
3.68 (12.29) 44.04 (29.04)
Stock pay (in %)
Stock option pay (in %)
3.27 (11.07) 41.08 (33.53)
p<0.001 b p=0.694 c
7.32 (15.11) 43.51 (34.82) 2.63 (8.32)
2.64 (10.98)
2.20 (7.66)
p<0.001 b
5.00 (11.25)
p=0.875 b
21.14 (18.41)
p=0.886 b
19.39 (15.50)
22.14 (17.73)
29.66 (25.04)
Mean (Std. dev.)
p<0.001 b
p-value of t-test
2.60 (8.68)
44.47 (29.74)
7.29 (15.28)
4.86 (11.35)
18.87 (15.94)
21.89 (18.57))
Mean (Std. dev.)
No (N=1,960)
p=0.914 b
p=0.017 b
p<0.001 b
p<0.001 b
p=0.003 b
p<0.001 b
p-value of t-test
Family owns more than 5%? Yes (N=618)
Notes: N=2,578; p-values refer to a two-sided t-test. a Miscellaneous other compensation includes the ExecuComp data items ALLOTHPD (all other paid) and OTHANN (other annual). b A Wilcoxon rank-sum test is significant (p0.05). c A Wilcoxon rank-sum test is not significant (p>0.05).
2.56 (11.45)
2.10 (8.50)
Long-term incentive plan (in %)
a
19.49 (19.21)
Annual bonus (in %)
Miscellaneous other compensation (in %)
28.05 (26.24)
Mean (Std. dev.)
Mean (Std. dev.)
Base salary (in %)
Components of executive pay
Non-family CEO (N=1,874)
Family CEO (N=704)
Family versus non-family CEOs
Table 8-2: Structure of executive pay – family versus non-family firms
8 Family firms and executive pay 163
164
8 Family firms and executive pay
8.4.2 Multivariate analysis – full sample
Table 8-3 shows a GLS random-effects model and a median regression on total pay. As can be expected from the large number of outliers shown in the box plot diagram above (figure 8-1), the two models produce different results. Outliers strongly influence the results of the random-effects model. The median regression, however, is more robust to outliers. Thus, in what follows, I focus only on the results of the median regression. The coefficients of most covariates are as expected, e.g., firm size and CEO duality lead to an increase in CEO pay. The variables family CEO (ß=544.25, p<0.01) and ownership share of family (ß=-19.12, p<0.01) have a negative impact on the level of CEO pay. The variable ownership share of financial investors does not have a significant impact (=5.28, p>0.1). A high level of debt decreases executive pay (ß=-21.43, p<0.01), whereas a high market-to-book value leads to an increase in executive pay (ß=757.98, p<0.01). Regarding the structure of executive pay, Tables 8-4 to 8-7 display median regressions on the share of base salary, annual bonus and stock options in total pay.248 Pseudo-R² values of the models range from 0.13 to 0.19.249 Base salary
Table 8-4 shows a median regression to estimate the effect of family CEO and ownership share of family on the percentage share of base salary.250 The model shows that being a family CEO does not have a significant impact on the share of base salary (ß=1.28, p>0.1). Contrary, the variable ownership share of family has a positive effect (ß=0.20, p<0.01). An increase of family ownership by 10% thus leads to an increase of the share of base salary of about 2 percentage points. No effect is found with regard to ownership by financial investors (ß=-0.01, p>0.1). Table 8-5 displays several regression models analyzing the effect of being a family firm on the share of base salary. The result is clear. The narrower a family firm is defined, the higher is the share of base salary in the CEO’s compensation package (ß family firm1 = 1.47 vs. ß family firm3 = 7.95, both p<0.05). Some control variables have a strong effect. CEOs of large firms are found to have a lower share of base salary in their compensation as compared to CEOs of small firms (ß=-2.76, p<0.01),
248
249
250
The results of the GLS random effects regressions regarding the main variables of interest are reported in the notes of the tables. STATA calculates the pseudo-R² value in median regressions as one minus the ratio of the sum of the weighted distances from the estimated median to the sum of the weighted distances from the original median. The calculation is similar to the calculation of the R² value in OLS regression except for that the median is used instead of the mean. If not stated otherwise, the reported coefficients in this section refer to Table 8-4.
8 Family firms and executive pay
165
whereas CEO tenure has a positive effect (ß=0.07, p<0.01). A higher market-to-book value decreases the share of base salary (ß=-1.15, p<0.01). This result makes sense. A stock price increase has a positive effect on option-based or stock-based pay, which then reduces the share of base salary. Industry dummies as well as time dummies are found to be jointly significant. Annual bonus
The share of annual bonus in the compensation package does not differ between family and non-family CEOs (ß=-0.53, p>0.1). The variable ownership share of family, however, has a positive effect (ß=0.11, p<0.01). Table 8-6 gives the results of a median regression on the share of annual bonus in family and non-family firms. CEOs in family firms (defined as family firm3) have a larger share of annual bonus in total pay (ß=4.58, p<0.01).251 Regarding the control variables, it is found that both firm age (ß=2.21, p<0.01) and ROA (ß=0.16, p<0.05) have a positive effect. The latter result makes sense, because the annual bonus is often tied to profitability measures such as ROA. Finally, as with the share of base salary, a high market-to-book value decreases the share of annual bonus (ß=-0.50, p<0.01). Stock option pay
To some degree, the results of the median regressions on the share of stock options mirror the results of the models that seek to explain the share of base salary and annual bonus. If a particular variable has a positive effect in the base salary regression, most likely it has a negative effect negative in the stock options regression and vice versa. Family ownership has a negative effect on the share of stock options (ß=-0.39, p<0.01), whereas being a family CEO has no significant effect (ß=0.20, p>0.1). An increase of family ownership by ten percent thus leads to a four percentage points lower share of stock options in total pay. The results of the regressions comparing family and nonfamily firms go in the same direction (Table 8-7). The narrower a family firm is defined, the lower is the share of stock option pay (ß family firm3 = -8.82, p<0.01). The control variables have the following effects: The variables firm size (ß=1.78, p<0.05) and market-to-book value (ß=1.97, p<0.01) have a positive effect. The variable firm age has a negative effect (ß=-8.35, p<0.01).
251
For the definition, see Table 2-2 above.
166
8 Family firms and executive pay
Table 8-3: Regressions of total pay (in $000s) Random-effects GLS (dep. variable: total pay) Variables Family CEO Ownership share of family Ownership share of financial investors Firm size Firm age Leverage Market-to-book value ROA CEO duration CEO duality Industry dummies (52 categories) Time dummies (9 categories) N obs. (firms) Obs. per group: min./avg./max. p-value Chi²-test Rho (fraction of variance due to ui) p-value Breusch-Pagan test R² within, R² between, R² overall Pseudo-R²
ß (SE) 304.27 (1,000.98) -52.83 (26.10) ** -38.65 (18.91) ** 3,819.17 (585.06) *** -2,061.83 (811.09) ** -100.12 (32.88) *** 822.48 (552.83) 14.80 (22.27) -65.23 (19.23) *** 1,480.71 (696.57) ** p<0.001 p<0.001 2,578 (393) 1; 6.6; 10 p<0.001 0.23 p<0.001 0.08; 0.39; 0.20
Random-effects GLS (dep. variable: log of total pay) ß (SE)
Median regression (dep. variable: total pay) ß (SE)
-0.332 (0.167) ** -0.016 (0.006) *** 0.002 (0.002) 0.335 (0.044) *** -0.075 (0.061) -0.005 (0.002) ** 0.071 (0.024) *** -0.001 (0.001) -0.002 (0.003) 0.069 (0.078) p<0.001 p<0.001
-544.25 (204.72) *** -19.12 (7.07) *** 5.28 (7.85) 1,646 (113.40) *** -117.08 (146.91) -21.43 (7.12) *** 757.98 (131.66) *** -26.57 (13.77) * -9.47 (6.34) 831.00 (212.08) *** p<0.001 p<0.001
2,578 (393) 1; 6.6; 10 p<0.001 0.20 p<0.001 0.13; 0.42; 0.25
2,578
0.16
Notes: * p0.1 ** p0.05 *** p0.01 Random-effects GLS regression: Standard errors (SE) are robust and clustered; two-sided tests are used. Median regression: number of bootstraps is 500.
ß (SE)
2,578 0.17
1.28 (0.89) 0.20 (0.04) *** -0.01 (0.03) -2.76 (0.34) *** 2.18 (0.49) *** 0.05 (0.02) ** -1.15 (0.19) *** 0.01 (0.02) 0.07 (0.03) *** -1.20 (0.80) p<0.001 p<0.001 2,578 0.13
-0.53 (1.05) 0.11 (0.05) ** 0.01 (0.03) -0.68 (0.43) 2.21 (0.62) *** 0.04 (0.03) -0.50 (0.16) *** 0.16 (0.04) ** 0.05 (0.04) -1.11 (0.94) p<0.001 p<0.001
ß (SE)
Annual bonus (share of total pay in %)
Base salary ß Family CEO = 4.07 (1.95) ** ß Ownership share of family = 0.25 (0.08) ***
Annual bonus ß Family CEO = 0.23 (1.36) ß Ownership share of family = 0.06 (0.05)
Random-effects GLS regressions yield the following results:
2,578 0.17
-0.20 (2.14) -0.39 (0.10) *** -0.01 (0.09) 1.78 (0.89) ** -8.35 (1.45) *** -0.03 (0.08) 1.97 (0.52) *** -0.08 (0.09) -0.10 (0.07) 1.07 (1.84) p<0.001 p<0.001
ß (SE)
Stock option pay (share of total pay in %)
Stock option pay ß Family CEO = -4.66 (2.40) * ß Ownership share of family = -0.23 (0.09) ***
Notes: * p0.1 ** p0.05 *** p0.01; number of bootstraps is 500; two-sided tests are used.
Obs. Pseudo-R²
Family CEO Ownership share of family Ownership share of financial investors Firm size Firm age Leverage Market-to-book value ROA CEO duration CEO duality Industry dummies (52 categories) Time dummies (9 categories)
Variables
Base salary (share of total pay in %)
Table 8-4: Median regressions of the structure of executive pay – full sample
8 Family firms and executive pay 167
2,578 0.16
-0.04 (0.03) -2.90 (0.33) *** 1.89 (0.49) *** 0.02 (0.02) -1.26 (0.20) *** 0.01 (0.03) 0.08 (0.03) *** -1.13 (0.78) p<0.001 p<0.001
1.47 (0.71) **
Model I ß (SE)
2,578 0.17
-0.04 (0.03) -2.94 (0.34) *** 2.02 (0.48) *** 0.03 (0.02) -1.28 (0.21) *** 0.01 (0.02) 0.08 (0.03) *** -0.96 (0.76) p<0.001 p<0.001
2.10 (0.73) ***
Model II ß (SE)
2,578 0.17
-0.03 (0.03) -2.75 (0.35) *** 1.97 (0.49) *** 0.05 (0.02) ** -1.21 (0.19) *** 0.01 (0.02) 0.07 (0.03) *** -0.83 (0.77) p<0.001 p<0.001
7.95 (1.57) ***
Model III ß (SE)
2,578 0.17
-0.02 (0.03) -2.91 (0.35) *** 1.83 (0.49) *** 0.05 (0.02) ** -1.20 (0.21) *** 0.01 (0.02) 0.08 (0.03) *** -1.17 (0.77) p<0.001 p<0.001
8.57 (1.89) ***
Model IV ß (SE)
ß Family firm1 = 3.51 (1.89) * ß Family firm2 =4.94 (1.79) *** ß Family firm3 = 8.95 (2.87) *** ß Family firm4 = 10.51 (3.12) *** ß Family firm5 = 7.89 (3.65) **
Random-effects GLS regressions yield the following results:
Notes: * p0.1 ** p0.05 *** p0.01; number of bootstraps is 500; two-sided tests are used.
Obs. Pseudo-R²
Family firm1 Family firm2 Family firm3 Family firm4 Family firm5a Ownership share of financial investors Firm size Firm age Leverage Market-to-book value ROA CEO duration CEO duality Industry dummies (52 categories) Time dummies (9 categories)
Variables
Table 8-5: Median regression of the share of base salary (in %) – family versus non-family firms
2,578 0.16
4.17 (2.03) ** -0.04 (0.03) -3.02 (0.35) *** 2.07 (0.49) *** 0.04 (0.02) * -1.15 (0.19) *** 0.01 (0.03) 0.09 (0.03) *** -1.39 (0.77) * p<0.001 p<0.001
Model V ß (SE)
168 8 Family firms and executive pay
2,578 0.13
-0.01 (0.03) -0.73 (0.42) * 2.19 (0.61) *** 0.03 (0.04) -0.35 (0.15) ** 0.16 (0.07) ** 0.08 (0.04) * -1.71 (0.95) * p<0.001 p<0.001
-1.42 (1.03)
Model I ß (SE)
2,578 0.13
-0.01 (0.03) -0.69 (0.42) * 2.32 (0.60) *** 0.03 (0.03) -0.34 (0.15) ** 0.16 (0.07) ** 0.07 (0.04) ** -1.48 (0.91) * p<0.001 p<0.001
-0.98 (1.03)
Model II ß (SE)
2,578 0.13
0.001 (0.035) -0.49 (0.43) 2.58 (0.61) *** 0.04 (0.03) -0.47 (0.16) *** 0.17 (0.07) ** 0.05 (0.03) -1.00 (1.00) p<0.001 p<0.001
4.58 (1.58) ***
Model III ß (SE)
2,578 0.13
0.02 (0.04) -0.63 (0.44) 2.48 (0.61) *** 0.04 (0.03) -0.49 (0.17) *** 0.17 (0.07) ** 0.05 (0.04) -1.24 (1.00) p<0.001 p<0.001
4.71 (2.14) **
Model IV ß (SE)
ß Family firm1 = -0.47 (1.04) ß Family firm2 = -0.06 (1.02) ß Family firm3 = 1.44 (1.52) ß Family firm4 = 0.81 (1.68) ß Family firm5 = 2.46 (2.24)
Random-effects GLS regressions yield the following results:
Notes: * p0.1 ** p0.05 *** p0.01; number of bootstraps is 500; two-sided tests are used.
Obs. Pseudo-R²
Family firm1 Family firm2 Family firm3 Family firm4 Family firm5a Ownership share of financial investors Firm size Firm age Leverage Market-to-book value ROA CEO duration CEO duality Industry dummies (52 categories) Time dummies (9 categories)
Variables
Table 8-6: Median regressions of the share of annual bonus (in %) – family versus non-family firms
2,578 0.13
3.60 (3.20) 0.01 (0.04) -0.78 (0.44) * 2.53 (0.59) *** 0.04 (0.03) -0.43 (0.18) ** 0.14 (0.07) ** 0.06 (0.04) * -1.22 (0.94) p<0.001 p<0.001
Model V ß (SE)
8 Family firms and executive pay 169
2,578 0.16
0.06 (0.08) 1.61 (0.89) * -8.48 (1.41) *** -0.01 (0.07) 1.70 (0.50) *** -0.08 (0.08) -0.10 (0.07) 2.26 (2.03) p<0.001 p<0.001
0.45 (2.00)
Model I ß (SE)
2,578 0.16
0.06 (0.08) 1.57 (0.87) * -8.52 (1.37) *** -0.02 (0.08) 1.72 (0.48) *** -0.09 (0.08) -0.08 (0.07) -2.10 (2.02) p<0.001 p<0.001
-0.23 (1.88)
Model II ß (SE)
2,578 0.16
0.01 (0.10) 1.52 (0.89) * -8.54 (1.29) *** 0.03 (0.07) 1.83 (0.51) *** -0.08 (0.08) 0.09 (0.07) 0.68 (1.94) p<0.001 p<0.001 2,578 0.17
-0.04 (0.09) 1.21 (0.86) -8.45 (1.30) *** -0.02 (0.08) 1.77 (0.52) *** -0.08 (0.08) -0.07 (0.07) 0.16 (1.93) p<0.001 p<0.001
-15.33 (3.57) ***
Model IV ß (SE)
ß Family firm1 = -1.87 (2.13) ß Family firm2 = -3.02 (2.00) ß Family firm3 = -7.62 (3.06) ** ß Family firm4 = -9.16 (3.47) *** ß Family firm5 = -9.66 (4.27) **
Random-effects GLS regressions yield the following results:
Model III ß (SE)
-8.82 (3.18) ***
Notes: * p0.1 ** p0.05 *** p0.01; number of bootstraps is 500; two-sided tests are used.
Obs. Pseudo-R²
Family firm1 Family firm2 Family firm3 Family firm4 Family firm5a Ownership share of financial investors Firm size Firm age Leverage Market-to-book value ROA CEO duration CEO duality Industry dummies (52 categories) Time dummies (9 categories)
Variables
Table 8-7: Median regressions of the share of stock option pay (in %) – family versus non-family firms
2,578 0.16
-13.29 (6.16) *** 0.03 (0.08) 1.43 (0.89) -8.83 (1.38) *** -0.03 (0.08) 1.81 (0.54) *** -0.08 (0.08) -0.07 (0.08) 1.15 (1.96) p<0.001 p<0.001
Model V ß (SE)
170 8 Family firms and executive pay
8 Family firms and executive pay
8.4.3
171
Multivariate analysis – only non-family CEOs
In this paragraph, I regard only firms that are managed by non-family CEOs. I am interested in the effect of family ownership when a family firm is governed by a non-family CEO.252 Table 88 displays the results of median regressions on the share of base salary, annual bonus, and stock options in total pay. The variable ownership share of family has a positive effect as regards the share of base salary (ß=0.15, p<0.01), but has no significant effect regarding the share of annual bonus or stock options in total pay (ß=0.01, p>0.1; ß=0.17, p>0.1). Table 8-8: Median regressions of the structure of executive pay – only non-family CEOs Base salary (share in %) Variables Ownership share of family Ownership share of financial investors Firm size Firm age Leverage Market-to-book value ROA CEO duration CEO duality Industry dummies (52 categories) Time dummies (9 categories) Obs. Pseudo-R²
Annual bonus (share in %)
Stock option pay (share in %)
ß (SE)
ß (SE)
ß (SE)
0.15 (0.05) *** 0.03 (0.03) -2.44 (0.38) *** 0.86 (0.60) 0.08 (0.04) ** -1.06 (0.32) *** 0.01 (0.06) 0.10 (0.03) *** 0.18 (0.83) p<0.001 p<0.001
0.01 (0.05) 0.02 (0.04) -1.18 (0.46) ** 2.28 (0.60) *** 0.07 (0.03) ** 0.36 (0.15) ** 0.18 (0.06) *** 0.10 (0.03) *** 0.40 (1.02) p<0.001 p<0.001
-0.17 (0.17) -0.13 (0.09) 2.20 (1.02) ** -6.90 (1.57) *** -0.20 (0.09) *** 1.72 (0.65) *** -0.29 (0.11) *** 0.18 (0.08) ** -0.86 (2.35) p<0.001 p<0.001
1,874 0.19
1,874 0.14
1,874 0.17
Notes: * p0.1 ** p0.05 *** p0.01; number of bootstraps is 500; two-sided tests are used. Random-effects GLS regressions yield the following results: Base salary ß Ownership share of family = 0.08 (0.05)
252
Annual bonus ß Ownership share of family = 0.10 (0.08)
The number of non-family CEO observations in family firms is 297.
Stock option pay ß Ownership share of family = -0.10 (0.11)
172
8.5 8.5.1
8 Family firms and executive pay
Discussion Implications for theory
This chapter contributes to the discussion of whether members of the founding family use their firm to extract private benefits of control (Claessens et al., 2002; Morck and Yeung, 2003). The results of this chapter do not support this view. In line with prior studies (Gomez-Mejia et al., 2003; McConaughy, 2000), family CEOs are found to have a lower salary than non-family CEOs. Moreover, after controlling for firm and industry characteristics, the share of stock option pay is lower in family firms relative to non-family firms. This is an interesting result because executive pay scandals and their coverage in the media often have addressed the exorbitant value of stock option pay (e.g., The Economist, 2006b; Useem, 2003).253 Options are seen as a form of hidden compensation that is not perceived by the market (Dechow et al., 1996), a view that then can lead to a principal-agent problem concerning the remuneration process (Bebchuk and Fried, 2003). According to the results in this chapter, managerial rent extraction through stock options is less likely to happen in family firms as compared to non-family firms. Still, the share of incentive-based pay is sizeable. Even with cautious estimates (i.e., not considering annual bonus), the average share of a CEO’s incentive pay is about 50%, which supports the idea that a family CEO’s altruism towards the members of her own family creates agency costs (Schulze et al., 2001, 2003), which can be reduced by bonding mechanisms. An alternative interpretation is that non-family shareholders in family firms such as mutual funds do not believe that family CEOs behave in the firm’s best interest and therefore prefer to tie executive pay to performance. The main contribution of this chapter to the executive compensation literature is to show that the structure of executive pay is different in family firms versus non-family firms, in particular the share of stock options and base salary. This result is important for the literature that analyzes the use and effectiveness of incentive pay (Jensen and Murphy, 1990a; 1990b; for a summary, see Tosi et al., 2000) or executive options (e.g., Dittmann and Maug, 2007; Feltham and Wu, 2001; for a summary, see Arnold and Gillenkirch, 2007). Dittmann and Maug (2007) calibrate a standard principal-
253
It should be noted, however, that it cannot be ruled out that members of the founding family seek private benefits of control. They might simply use means other than manipulating the remuneration process to extract personal rents. For a discussion of the various types of private benefits of control, see Dyck and Zingales (2004) or La Porta et al. (2000). For an example of an unusual and creative way to extract rents from shareholders, see Liu and Yermack (2007).
8 Family firms and executive pay
173
agent model and find that most CEOs should not receive any stock options at all.254 Their second finding is that through better designed contracts, compensation costs could be reduced by an average of 20%. They also make the argument that these potential savings are higher in firms with weak corporate governance. This chapter’s results support this view. It is found that CEOs in family firms receive less option-based pay than CEOs in non-family firms do.255 Family firms usually have a strong corporate governance, since, in most cases, family owners are blockholders and have a good knowledge about the firm and the underlying business model, which is why monitoring is more effective (Bertrand and Mullainathan, 2000; Holderness, 2003). Another argument is that family owners substitute incentive pay for monitoring (Mehran, 1995). Finally, it might also be the case that the interests of a CEO in a family firm are in itself more strongly aligned with the firm’s longterm interests than this is the case in non-family firms. CEOs in family firms could be intrinsically motivated to act in the firm’s best interest. Giving high-powered incentives in such a situation is not a good idea. Individuals shift their focus from the activity itself to the reward or sanction; intrinsic motivation is crowded out (Deci and Ryan, 2000; Frey and Osterloh, 2005). 8.5.2
Implications for practice
From a practical perspective, this chapter’s main contribution is to show that there exist great differences between family and non-family firms regarding the structure of executive pay. The share of option-based pay in family firms is found to be lower, and the share of base salary is found to be higher than in comparable non-family firms. CEOs in family firms thus receive a lower share of incentive pay in their compensation package and are paid more like bureaucrats. This is an important result for members of remuneration boards of non-family firms and executive pay consultants to consider, because family firms are not found to have a lower corporate performance relative to non-family firms (Anderson and Reeb, 2003; Villalonga and Amit, 2006).256 Giving too highpowered incentives might indeed be inefficient from a shareholder’s perspective; it only leads to an increase in the executive’s pay but does not have an impact on corporate performance.
254
255 256
That is, they use executive compensation data to estimate the intensity of the incentives in the principal-agent contract and compare this contract to the optimal contract. This leads to a savings variable which is calculated as the difference between in compensation costs between observed and optimal contracts. The share of options decreases further if a family firm is defined in a more narrow way. Some of the studies event point to a superior performance of family firms as compared to non-family firms. See also Chapter 5 for an analysis of the performance of family and non-family firms using the same dataset as used in this chapter. In the chapter, there is also a literature review about the performance of family firms.
174
8.6
8 Family firms and executive pay
Summary and conclusions Turning back to the question of whether CEOs in family firms are paid like bureaucrats, the
evidence presented in this chapter provides a mixed picture. On the one hand, it is found that family ownership increase the share of base salary in total pay. Furthermore, it is found that family ownership decreases the share of stock option pay. Thus, the idea that CEOs in family-owned firms are paid more like bureaucrats is supported. On the other hand, the share of incentive pay in family firms is still very high. For example, the mean share of stock-option pay of a CEO that belongs to the business-owning family is about 44% (Table 8-2). The mean share of annual bonus of a family CEO is about 19% (Table 8-2). In conclusion, the link between pay and performance seems to be less strong in family firms than in non-family firms. Nevertheless, a large amount of pay is still dependent on performance. The discussion of the structure of executive pay and the exorbitant value of stock option pay in the media and the academic literature should take into account differences between family and nonfamily firms.257 The chapter is connected to chapter 6 and chapter 7 in that it gives an explanation why family ownership is found to increase the level of R&D spending and decrease the likelihood of employment downsizing – used as indicators of long-term orientation. This chapter’s finding that family ownership decreases the share of stock options in total pay supports the argument that defectively designed incentive pay may lead managers to behave myopic (Frey and Osterloh, 2005; Reichelstein, 2000; Schotter and Weigelt, 1992).258 There are two main reasons as to why stock options can induce short-term behavior: First, vesting periods (that is the period of time the executive must wait until she can exercise her stock options) are often only one or two years and are often renegotiated. Second, stock options are often granted to the CEO based on strong short-term performance. There are several promising avenues for further research: It would be interesting to go deeper into the issue of stock option pay and to have a closer look at the design of the stock options. Are option exercise prices more likely to be indexed to the market when there is a family owner sitting in the remuneration board? With a family owner being present, does option payment merely come as additional pay or does its introduction lead to a reduction in the other components of executive pay (as it should be in efficient contracting)? Another avenue would be to follow the approach of Dittmann and Maug (2006) and calibrate a principal-agent model for family and non-family firms.
257 258
See Section 8.1 for appropriate references. See also the discussion about executive pay as a source of managerial myopia in Section 3.3.3.
8 Family firms and executive pay
175
This approach would allow a direct comparison of the compensation costs of both types of firms and lead to a statement whether family firms use more effective and/or efficient contracts than nonfamily firms do. A third avenue would be to regard the pay-for-performance relationship. If CEOs from the owning family are indeed more intrinsically motivated, then incentive pay should have a lower impact on corporate performance.
9 Executive pay in family firms: a principal-agent model259 A large number of family firms employ non-family executives. This chapter analyzes the optimal compensation contract for non-family executives using principal-agent analysis. It is shown that if the business-owning family is interested in long-term performance but the non-family executive is more interested in short-term performance, the compensation contract should be lowpowered with regard to short-term performance. The level of short-term incentives depends on the measurement error of effort for short-term performance, the executive’s level of risk aversion, and the executive’s responsiveness to short-term incentives. This chapter is connected to Chapter 6 and Chapter 7 by explaining how the compensation contract of a non-family executive should be designed if the business-owning family wants to pursue a long-term strategy. Furthermore, this chapter explains some of the empirical findings about the differences in executive pay between family and non-family firms (Chapter 8). The remainder of the chapter proceeds as follows. Section 9.1 introduces and motivates the topic. Section 9.2 reviews the related literature. Section 9.3 introduces the model. Section 9.4 proceeds with the analysis. Section 9.5 discusses the implications for theory and practice. Section 9.6 summarizes and concludes the chapter.
9.1
Introduction In his classic article “On the folly of rewarding A, while hoping for B,” Steven Kerr writes
that “we hope for long-term growth […], but we often reward quarterly earnings” (Kerr, 1995, p. 14) and cites this as a common management reward folly.260 Given the current financial crisis, stock options and annual bonus payments have been particularly criticized for setting shortsighted incentives (Fockenbrock and Terpitz, 2008; Thielemann, 2008; von Preen, 2008). This problem is of particular concern for family business owners, who regard their firms as long-term investments and who need to pay non-family executives who do not necessarily share this view.
259
260
This chapter is based on Block and Henkel (2008). The paper was presented at European Business School (European Family Business Center and Department of Law, Governance, and Economics), the 2008 annual conference of the International Family Enterprise Research Academy (IFERA), the 2009 WHU Campus for Finance Research Conference, and the 2008 Academy of Management Meetings. See also Kerr (1975).
9 Executive pay in family firms: a principal-agent model
177
Most family firms employ non-family executives. Non-family executives occupy 80% of all board seats in large US family firms; non-family CEOs manage about 55% of these firms (Anderson and Reeb, 2003, 2004). Klein (2000) reports that 56% of all German family firms have at least one non-family member on their top management teams. Klein (2000) also finds that the importance of non-family executives increases with the size of the firm. When family firms expand, the family often has difficulty providing enough management talent. Furthermore, there are a number of other reasons to employ non-family executives, such as added knowledge, lack of a family successor, or family conflict mediation (Dyer, 1989; Schein, 1968). Most of these reasons are linked to the limited human resource pool of the owning family (e.g., Bennedsen et al., 2007; Burkart et al., 2003; Perez-Gonzales, 2006). Non-family executives have a difficult job. The family often has a lot of power and can easily replace the executive.261 Furthermore, their performances in the job are often evaluated by less senior executives who are members of the business-owning family. Problems with non-family executives arise when their preferences and time horizons differ strongly from those of the owning family, which is likely to be the case (Chua et al., 2003; Dyer, 1989, Schein, 1968, 1983). In a flexible labor market262, the wage of an executive depends strongly on her short-term performance, which is determined by short-term measures such as the increase in operating efficiency, sales, or shareholder return (Campbell and Marino, 1994; Narayanan 1985). A non-family executive who wants to increase her wage or move to a new employer needs to score high on these measures. The personal goals of the non-family executive might differ strongly from the goals of the family. As discussed in the previous chapters, the goals of the business-owning family are often long-term (Bertrand and Schoar, 2006; James, 1999; Le Breton-Miller and Miller, 2006). A conflict between the family and the non-family manager is thus likely to occur.263, 264 This chapter deals with this situation and determines the optimal compensation contract for a non-family executive when the interests of the two parties diverge. To analyze this situation, I extend the well-known principal-agent multi-task model of Holmstrom and Milgrom (1991). It is assumed that the business-owning family has a long-term horizon and that its main interest is to trans261
262
263
264
The Financial Times Deutschland lists a number of examples in which the business-owning family and the nonfamily managers were in a conflict. See http://www.ftd.de/koepfe/319268.html (accessed October 21st, 2008). Flexible labor markets are characterized by low costs of hiring and firing, flexibility in terms of location and pay arrangements and short-term employment contracts (e.g., Brodsky, 1994). In a flexible labor market, individuals change their jobs more frequently than in other labor markets. For research on the goals of family firms, see Casson (1999), Guzzo and Abbott (1990), or Tagiuri and Davis (1992). As an example, consider an investment in a promising technology project. The family sees the investment as a chance to gain a lasting competitive advantage. The non-family manager might worry about the R&D expenditures incurred in the short term.
178
9 Executive pay in family firms: a principal-agent model
fer the firm to the next generation. The manager, however, wants to demonstrate her management skills to the outside market for executives and therefore wants to produce strong short-term results. I determine the optimal compensation contract and calculate the optimal level of short-term incentives. It is found that (1) the optimal sensitivity of pay to short-term performance decreases with the executive’s willingness to send a signal of strong short-term performance to the market for executives; (2) the level of this signaling discount depends on the error in measuring short-term performance, the executive’s risk aversion, and the executive’s responsiveness to incentives. These results suggest that it can be advantageous to offer a fixed-wage contract to a non-family executive (Frey and Osterloh, 2005) or to rely on subjective performance evaluations or voluntary bonus payments265 (Baker et al., 1994; MacLeod and Malcomson, 1989).
9.2
Literature review Four strands of literature were identified that are particularly important in the context of this
chapter: (1) differences between family and non-family executives, (2) empirical research on CEO pay in family firms, (3) principal-agent models on executive compensation, and (4) theoretical models analyzing the problem of managerial myopia. 9.2.1
Non-family executives in family firms
Given the importance of non-family executives for the growth of family firms, it is surprising that there is relatively little research on the role of non-family executives in family firms (Chua et al., 2003). I summarize what is known about the differences between family and non-family executives.266 Founders or family executives often have a particular vision of their product. Their decisions are often made intuitively, their power is based on ownership, and they try to motivate their employees by charismatic behavior (Dyer, 1989; Schein, 1983). Non-family executives, by contrast, often make decisions based on logic and rational analysis rather than intuition. Their management style is often impersonal (Dyer, 1989; Schein, 1983). The differences between the two types of managers are explained by their organizational and occupational experiences. Family executives have grown up in the organization and have learned skills and practices idiosyncratic to their organization. They have little experience outside their own firm. Their view of the firm is strongly influenced by what their family thinks is good for the firm. Their peer group consists of family 265
266
That is, the remuneration board grants a bonus based on its subjective assessment of the executive’s performance, in particular the more subjective aspects of performance (e.g., willingness to cooperate with the board, level of customer satisfaction). For a more complete overview, I refer to Klein and Bell (2007) or Blumentritt et al. (2007).
9 Executive pay in family firms: a principal-agent model
179
members, employees of the firm, and managers of other family firms. Non-family executives are socialized differently: after graduation, they often go to work in large firms, change jobs frequently, and gain a broad range of organizational experiences.267 Most of them have a management education in which training is formal and generic. Their peer group consists of their fellow students working in other (often large) organizations or in consultancy firms (Dyer, 1989; Schein, 1983). They are connected to their peers through professional or university alumni associations.268 Although this view does not apply to every family firm and may change as more family executives are educated at business schools, the main point remains valid: family executives and nonfamily executives are socialized in a different way and are likely to have different goals and views of the firm. 9.2.2
Executive pay in family firms
There is a wealth of empirical research on CEO pay.269 I summarize only studies that involve family firms. This research is classified into (1) studies about the level of CEO pay and (2) studies about the structure of CEO pay. Gomez-Mejia et al. (2003) and McConaughy (2000) find that family CEOs receive lower compensation than non-family CEOs working in family firms. This effect becomes stronger with an increase in family ownership concentration (Gomez-Mejia et al., 2003). Similarly, studying the relation between insider ownership and CEO pay, Core et al. (2003), Allen (1981), and Lambert et al. (1993) find that CEO compensation decreases with the CEO’s ownership stake.270 Finally, Deckop (1988) shows that CEOs recruited from outside the firm earn significantly more than founder CEOs. There is also some counterevidence: Holderness and Sheehan (1988) demonstrate that managers in publicly held corporations who are also majority shareholders (defined as shareholders with more than 50% of common stock) have higher salaries than managers in more diffusely held firms. Regarding the structure of CEO pay, Chrisman et al. (2007) find that family executives (1) receive incentive compensation and (2) that it has a positive effect on performance, suggesting that family executives also exhibit agency behavior. However, McConaughy (2000) discovers that fam267
268
269 270
This is true for non-family executives who are hired from the outside labor market. Note that there exist also a large number of non-family executives that do not change their employers and are with one firm over their entire working life. The contrasts between family and non-family executives are stronger when hired non-family executives are considered. Non-family executives that spend their entire working life in one firm can be considered as falling in between the two categories family executive and hired non-family executive. Note that this argument refers to the situation in the US. In Germany, for example, most students of state universities do not enter a university alumni association. For a summary of this literature, I refer to Murphy (1999). Note that there is a strong overlap between insider ownership and family ownership.
180
9 Executive pay in family firms: a principal-agent model
ily CEOs receive less incentive-based pay than non-family CEOs working in family firms, a finding which he termed the family incentive alignment hypothesis. The results of Gomez-Mejia et al. (2003) are similar. The greater the percentage of stock owned by institutional investors, who are assumed to vigilantly protect their interests, the lower is the share of long-term income in the total compensation of a family CEO. 271, 272 Gomez-Mejia et al. (2003) also find that the positive effect of R&D intensity on the share of long-term income in CEO pay is stronger for non-family CEOs than for family CEOs. They argue that a non-family CEO wants to be compensated for the increased ambiguity and uncertainty associated with a higher level of R&D intensity. 9.2.3
Principal-agent models on executive pay
Theoretical work on executive compensation draws mainly on principal-agent models and contract theory (e.g., Ross, 1973). Hart and Holmstrom (1987) and Prendergast (1999) provide surveys of this literature. Since I use a multi-task principal-agent model, my literature review is limited to studies using this type of model. Holmstrom and Milgrom (1991) were the first to analyze the optimal provision of incentives in a multi-task environment. They show the existence of important interaction effects between the incentives given for one task and the incentives given for another task. With multiple tasks, incentive pay not only allocates risk and incentivizes hard work but also distributes the agent’s attention between her various tasks. If one task can be observed and the other cannot, then the agent will focus her efforts on the task that can be observed, which may not be in the principal’s interest. This central result has been taken to explain why incentive contracts are observed less frequently than one-dimensional principal-agent models predict. The costs of this reallocation of efforts are higher than the benefits obtained from the incentives. In line with this argument, MacLeod and Malcomson (1989) and Baker et al. (1994) suggest that if some aspects of the agent’s performance cannot be contracted upon, it is better to rely on subjective performance evaluations or voluntary bonus payments. Empirically, Brown (1990) has shown that incentive pay is less likely in a job with a variety of different duties than in a job with a narrower set of duties. In a similar vein, MacLeod and Parent (1999) find that jobs with high-powered incentives involve greater worker autonomy and fewer tasks than hourly paid or salary jobs.
271
272
Interestingly, long-term income is calculated as the number of stock options granted multiplied by 25 percent of the exercise price of these stock options. Stock options are thus considered as being long-term income, which is in not in line with those who argue about the adverse effects of stock options (e.g., Fuller and Jensen, 2002; Frey and Osterloh, 2005). It is argued that financial investors know about the family incentive alignment hypothesis and thus push for a lower share of incentive pay in the CEO’s compensation package.
9 Executive pay in family firms: a principal-agent model
9.2.4
181
Principal-agent models on managerial myopia
There is a considerable literature on managerial myopia and its explanations.273 This review is limited to explanations of managerial myopia that involve managerial opportunism. Narayanan (1985) finds that if an executive has private information regarding her decisions, she may be biased towards shortsighted decisions that are not in shareholders’ long-term interests. The reason is that such a manager wants to enhance her reputation in the labor market (which primarily values the manager’s short-term performance) and improve her wage. Narayanan (1985) also demonstrates that this short-term behavior is negatively related to the manager’s experience, the duration of her contract, and firm risk. Campbell and Marino (1994) extend this idea and show that the problem of managerial opportunism becomes more severe with mobile managers and flexible labor markets. They argue that in countries such as Japan, where inter-firm mobility is almost non-existent, the interests of management and shareholders are more strongly aligned, spurring investment in longterm projects. By contrast, in countries or industries with more flexible labor markets, there is a strong incentive to behave myopically, because a manager’s compensation increases with her perceived ability. An optimal contract should therefore induce an executive to stay with the firm using deferred contingent compensation. The problem is that if the terms of employment are determined on the labor market, there exist minimum levels of total compensation and the share of incentive pay. In this case, the problem of myopic behavior cannot be solved through compensation contracts. Campbell and Marino’s (1994) research is in line with that of Holmstrom (1982b), who was the first to introduce the idea that flexible labor markets may actually foster a managerial incentive problem. Going in the same direction, Hirshleifer and Thakor (1992) argue that concerns about reputation distort managers’ investment decisions in favor of relatively safe projects that lead to short-term payoffs but have low expected returns for the firm. This puts a manager’s interests closer to the interests of bondholders, who favor short-term, low-risk projects over long-term, high-risk projects (Bester and Hellwig, 1987; Biais and Casamatta, 1999; Jensen and Meckling, 1976; Stiglitz and Weiss, 1981).
9.3
The model Before I describe the model, two important points deserve mention. First, my model falls into
the category of principal-agent models (Ross, 1973). Principal-agent models have been criticized by management scholars (e.g., Lubatkin et al., 2007a, b; Lubatkin, 2007; Zahra, 2007), because they
273
For a review of this literature, see Laverty (1996) and Section 3.3.2.
182
9 Executive pay in family firms: a principal-agent model
use a reductionist approach and because they assume individuals are rationally behaving agents that only seek to maximize their respective individual utility (Jensen and Meckling, 1976; Davis et al, 1997). 274 Agency problems and agency costs occur when the utility functions of agents and principals are not aligned. Monitoring and provision of incentives can solve these agency problems. Critics argue that the economic model of man does not explain the complexity of human behavior because intrinsic motivation, fairness, and pro-social behavior are not considered. Although I partially agree with this criticism, I deem these concerns to be less relevant for the particular principal-agent relationship investigated in this chapter. As the literature review in the previous section shows, the goals of non-family executives and the business-owning family are likely to diverge (Section 9.2.1). Thus, in line with other scholars (Blumentritt et al., 2007; Chrisman et al., 2004; Chua et al., 2003; Gomez-Mejia et al., 2001), I assume an agency relationship between the two parties.275 The second comment refers to the selection process itself. My model aims to determine the optimal compensation contract after the non-family executive has been selected. No statement is made about the contract that should be offered before this selection has taken place. The firm can use screening and offer different types of contracts to overcome the problem of information asymmetry, thereby avoiding a lemons market (Akerlof, 1970; Rothschild and Stiglitz, 1976).276 The model is as follows. I consider a principal-agent environment in which the businessowning family (hereafter family) constitutes the principal and the non-family executive (hereafter executive), the agent. The executive is hired from the outside labor market and has not been with the firm before. The value of output for the family is B(e), where e is the executive’s efforts associated with private costs C(e) for the executive. B(e) is increasing in e (in a linear way). C(e) is increasing and convex.277 The executive has two tasks: achieving short-term performance, e1 (e.g., increases in dividends, increases in sales, etc.), and achieving long-term performance, e2 (e.g., technological edge, good relations with stakeholders, survival of the firm, etc.). The value the family
274 275 276
277
See also Section 2.3.1. Note that the results of the model do not apply to situations in which this assumption does not hold. Off course, in management practice, the contract used to attract a new manager is linked to the contract after the hiring has taken place. Yet, in order to make a statement about the incentive effects (and not the selection issue), I have to abstract from the selection phase and assume the type of non-family manager as given (that is I treat the type of non-family manager as exogenous variable). That is, the executive’s marginal costs of effort increase with additional effort.
9 Executive pay in family firms: a principal-agent model
183
attributes to each task is denoted by p1 (p2).278 The family is assumed to be risk-neutral. The family’s gross benefit is given by B(e1,e2) = p1e1 + p2e2.
(9-1)
The family separately observes the results of each task: x1 = e1 + 1 and
(9-2)
x2 = e2 + 2,
(9-3)
where the pair of observational noises (1, 2) follows a centered normal distribution with variance
¦
§ V 2V · ¨ 1 12 ¸ , ¨V V 2 ¸ © 21 2 ¹
(9-4)
where 12 and 21 are assumed to be 0. The manager is assumed to be risk-averse. Her utility function is given by u(z) = -e-rz,
(9-5)
where r is a positive constant that indicates the manager’s risk aversion and z denotes the executive’s utility if she were risk-neutral. z is given by z = (w+Jx1) - C(e),
(9-6)
where w denotes the executive’s compensation and J indicates the value the executive attributes to sending a signal of short-term performance (x1) to the market for executives. The inclusion of J constitutes the main difference between my model and the model of Holmstrom and Milgrom (1991). Equation 9-4 assumes a CARA utility function279 for the executive, which allows me to use the results of Holmstrom and Milgrom (1987) and to limit the attention to linear contracts.280 The executive’s compensation is given by
278
Note that the values attributed to each task might differ across the population of family firms. A business-owning family that wants to sell the firm is interested in short-term performance, whereas a family that wants to transfer the business to the next generation cares more about long-term performance.
184
9 Executive pay in family firms: a principal-agent model
w = 1x1 + 2x2 + ß,
(9-7)
where 1 (2) denotes the intensity of the incentives for short-term performance (x1) or long-term performance (x2). ß is a fixed component greater than zero. Making use of the exponential form of the utility function, the executive’s certainty equivalent281 is CE=1e1 + 2e2 + ß + Je1 - C(e) - 0.5r1212 - 0.5r2222 - 0.5rJ212.
(9-8)
That is, the executive’s certainty equivalent consists of the expected wage plus the value of signaling to the market for executives minus her costs of effort and minus a risk premium. The net benefit of the family is given as B(e) = p1e1 + p2e2 - w = p1e1 + p2e2 - (1x1 + 2x2 + ß).
(9-9)
That is, the net benefit of the family equals its gross benefit minus the costs of the compensation contract. Applying the value maximization principle now allows calculating the optimal contract for the family by maximizing the total certainty equivalent of the contract, that is the certainty equivalent of the family plus the certainty equivalent of the executive (Milgrom and Roberts, 1992, pp. 35-37 and p. 218). The value maximization principle states that any contract is efficient only if it maximizes the total sum of the certainty equivalents of the two parties. The argument is that if a contract exists that leads to a higher total certainty equivalent, it is always possible to distribute it in a way that makes both parties better off. The value maximization principle thus separates the issue of the distribution of value from the issue of the creation of value. The value maximization principle can only be assumed if the no wealth condition applies, which states that the family’s decisions are not affected from increased wealth due to cash transfers from the contract (Milgrom and Roberts, 1992, pp. 3536). This assumption makes sense if the cash transfers between the family and the executive are small relative to the family’s overall financial resources.
279
280
281
CARA stands for constant absolute risk aversion. That is, risk aversion is independent of the agent’s level of consumption (level of income). An alternative would be to assume relative risk aversion, which states that risk aversion depends on the agent’s level of consumption (level of income). Holmstrom and Milgrom (1987) show that, in many cases, a linear contract is the unique optimal compensation scheme. That is, a certain payment which gives the same level of utility as the (uncertain) incentive contract.
9 Executive pay in family firms: a principal-agent model
185
Using the value maximization logic described above, the optimal contract for the family can be obtained by maximizing the total certainty equivalent of the contract, which is TCE=p1e1 + p2e2 + Je1 - C(e) - 0.5r1212 - 0.5r2222 - 0.5rJ212,
(9-10)
with regard to e1 and e2 under the participation constraint that the certainty equivalent is greater than zero. Furthermore, the contract should be compatible with the two incentive constraints C1 d D 1 J and
(9-11)
C2 d D 2 .
(9-12)
In the optimum, (9-11) and (9-12) are binding, i.e., C1 = 1 + J or C2 = 2. The participation constraint is fulfilled if ß, the fixed component of the compensation contract, is assumed to be greater than zero. Figure 1 below describes the sequence of events; Table 9-1 summarizes the notation used in the model. Figure 9-1: Sequence of events
Nature selects J
Family picks a
Family makes a
Manager ac-
Manager under-
Contingent
randomly.
manager, then
contract offer.
cepts the con-
takes effort.
payments
learns about her J.
tract or not.
are made.
186
9 Executive pay in family firms: a principal-agent model
Table 9-1: Notation used in the model
9.4
Variable
Meaning
B(e)
Benefit of the family
e1, e2
The executive’s efforts for short-term performance (long-term performance)
p1, p2
The value the family attributes to short-term performance (long-term performance)
1, 2
Measurement errors
12, 22
Variances of measurement errors
x1, x2
Observed short-term performance (long-term performance)
1, 2
Sensitivity of pay to performance regarding short-term goals (long-term goals)
ß
Fixed component in the compensation contract
r
Level of manager’s risk aversion
z
Net benefit of the executive if she were to be risk-neutral
u(z)
The executive’s utility function
J
Value the executive attributes to signaling of short-term performance
C(e)
Costs of efforts
C1, C2
First derivatives of cost function; marginal cost of efforts
C11, C22
Second derivatives of cost functions
C12, C21
Cross derivative of cost function; determines whether tasks are cost substitutes or cost complements.
Analysis The goal of this chapter is to determine the optimal sensitivity of pay to short-term perform-
ance (1). I aim not to make a statement about the optimal sensitivity of pay to long-term performance (2), which is why it is treated as an exogenously given variable. In the case of non-family executives, long-term incentive pay is usually low. Two examples illustrate this. The most common form of long-term incentive pay consists of restricted stock that the executive is not allowed to sell for a certain period of time (Murphy, 1999).282 However, most family firms are reluctant to transfer ownership rights to family outsiders. Second, non-family executives are often hired only for a short period until a junior family member is supposed to take over the business (Miller et al., 2003), which makes it difficult to assess their impact on long-term performance. The model is now analyzed using the following steps. First, some general compensation rules are established. Next, based on the model’s results, the compensation of a non-family executive is
282
Alternatives are phantom stock (a bonus based on the value of a stated number of shares that is paid out at the end of a specified period of time) or other forms of long-term bonuses. Chapter 8 shows, however, that these forms of compensation are not widely used (ExecuComp item LTIP).
9 Executive pay in family firms: a principal-agent model
187
discussed. Finally, the optimal compensation contract of a non-family executive is compared to the optimal compensation contract of a family executive. 9.4.1
General compensation rules
The goal in this chapter is to determine the optimal sensitivity of pay to short-term performance (1). In the optimal contract, the executive’s marginal benefit of a unit of effort equals her marginal costs of this unit of effort, as described by (9-11) and (9-12). Hence, it is possible to derive the optimal contract with respect to sensitivity of pay to short-term performance (1) by calculating the partial derivative TCE/e1: wTCE we1
p1 J C1 rV 1 (C1 J )C11 rV 2 C 2 C 21 , 2
2
(9-13)
Substituting the incentive constraints (9-11) and (9-12) into (9-13) then yields
D 1 J
p1 J D 1 J rV 1 D 1C11 rV 2 D 2 C 21 2
2
(9-14)
Solving (9-14) for 1 gives p1 J rV 2 D 2 C 21 2
D1
2 rV 1 C11 2
(9-15)
Following (9-15), the intensity of incentives for short-term performance (1) should be high if
p1 (the value the family attributes to short-term performance) is high.
r is low: The less risk-averse the executive, the more effective it is for the family to provide the executive with incentives.
J is low: The executive has less interest in signaling a strong short-term performance to the market for executives.
V 1 is low: e1 is observable to a certain degree.
2 is low: a decrease in the compensation for long-term performance shifts the executive’s
2
attention to achieving short-term performance.
C11 is low: a slight change in 1 results in a big change in e1. The executive is responsive
to incentives.
188
9 Executive pay in family firms: a principal-agent model
C 21 is low: the efforts toward achieving short-term goals and the efforts toward achieving
long-term goals are cost substitutes to a low degree (that is, increasing the marginal costs of effort for short-term performance leads only to a minor increase of the marginal costs of effort for long-term performance). This substitution effect increases with high V 2 (meas2
urement error in efforts for long-term performance) and high 2 (incentives for long-term performance). To explore some of the model’s properties, Figure 9-2 provides a sensitivity analysis. Each graph displays the intensity of incentives for short-term performance as a function of the executive’s risk aversion, the measurement error regarding efforts for short-term performance, etc. It is assumed that the two tasks are substitutes in the cost function. That is, high incentives for one task cause the manager to divert effort from the other task.
9 Executive pay in family firms: a principal-agent model
189
Figure 9-2: Sensitivity analysis Risk attitude
Signaling 0.50
0.50
0.40
0.40
0.30
0.30
0.20
0.20
0.10
0.10
0.00
0.00 0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
Value of signaling for the executive ()
Executive’s risk aversion (r)
Notes: p1=1, r=0.5, C11=0.5, 21= 22=0.5, C21=0.5, 2=0.5
Notes: p1=1, =0.5, C11=0.5, 21= 22=0.5, C21=0.5, 2=0.5
1
Preference for short-term performance
Measurement error 0.50
1.00
0.40
0.80
0.30
0.60
0.20
0.40
0.10
0.20
0.00 0
0.2
0.4
0.6
0.8
1
0.00 0
0.2
0.4
0.6
0.8
Measurement error for short-term performance (21)
Family’s preference for short-term performance (p1)
Notes: p1=1, =0.5, r=0.5, C11=0.5, 22=0.5, C21=0.5, 2=0.5
Notes: =0.5, r=0.5, C11=0.5, 21=22=0.5, C21=0.5, 2=0.5
1
190
9.4.2
9 Executive pay in family firms: a principal-agent model
Compensation of a non-family executive
Equation 9-15 is used to explain the optimal contract of a non-family executive. Beyond the general rules derived in the previous sections, the model gives some further insights regarding the compensation of a non-family executive. Generally, the model predicts that a signaling-oriented executive should be offered a contract with lower incentives for short-term performance than other executives are offered. I call this difference the signaling discount. I argue that the signaling discount should be greater for non-family executives than for family executives. The argument is simple. Non-family executives will not necessarily remain in the firm in the long run, and they will want to leave with a successful track record. Unlike family executives, non-family executives do not have the patience to wait for longterm strategies to ultimately pay off. In some situations, this tendency is reinforced, e.g., if the non-family manager is only employed on an interim basis, if the labor market for executives is flexible, or if the manager is young and ambitious. Each of these situations is discussed in more detail below. Employment on an interim basis
A family firm may hire a non-family executive on an interim basis (Miller et al., 2003) because of the sudden death of a family manager or to bridge a gap between two family generations. Almost by definition, such an executive has a strong incentive to demonstrate her skills with strong short-term results. After all, she needs to look for a new job after her contract expires, so signaling has a strong value for her. Ceteris paribus, her contract should be lower-powered with regard to short-term performance than the average contract. Market for executives
Labor markets differ. Some labor markets are flexible, whereas others are not (e.g., Brodsky, 1994). The nature of the labor market for executives plays an important role in the design of the compensation contract. On the one hand, it restricts the options available. Given that the family firm competes for the best executives, the family members may not have the power to design the compensation contract and its incentive structure as they wish to. On the other hand, the nature of the labor market determines the value of signaling. In a labor market with low inter-firm mobility (e.g., in a country like Japan or in a highly specialized industry), the value of signaling and consequently the risk of myopic behavior are low. In such a situation, the contract may include short-term incentives.
9 Executive pay in family firms: a principal-agent model
191
Age of the manager
If the family hires a young executive, it is fair to assume that this executive sees her job as a stepping stone to another position in the firm or outside the firm. She wants to signal through early efforts that she intends to make a difference. These efforts may or may not be in the family’s best interest. In any case, the executive wants her efforts to be visible to the labor market. She is likely to engage in activities that have a strong effect on short-term performance. For example, she might push sales with shortsighted price promotions or advertising. Such an executive has a strong inherent incentive to achieve short-term performance. This tendency to act in the short term should not be further intensified by high-powered incentive contracts. Hence, ceteris paribus, the contract of a young executive should include lower-powered incentives than the average contract has. The contrary should be the case with an executive close to pension age, who has a low incentive to demonstrate her management skills by means of strong short-term results, as she is unlikely to think about changing jobs. 9.4.3
Family versus non-family executives
Although the model is designed to analyze the structure of compensation for non-family executives, some predictions regarding the compensation of family executives can also be derived. The question is very relevant for family firms, as family and non-family managers often work in similar positions and are members of the same board of directors. Should family and non-family managers be offered similar contracts? To answer this question, the following issues need to be considered. Family executives283 differ in two important aspects from non-family executives. First, they find it more difficult to leave the firm, reducing their willingness to send a signal of strong shortterm performance to the market for executives (low J). Second, family managers are likely to inherit the firm in the future. Following a long-term strategy that leads to better long-term performance increases their future wealth; wealth of family managers and long-term performance of the firm are thus strongly linked (which leads to a high 2). The family executive is strongly incentivized to achieve long-term performance. How do these two aspects influence the intensity of incentives for short-term performance?
The first aspect, which I have referred to as the signaling discount, implies a higher level of short-term incentives for family executives than for non-family executives.
283
I define family managers as managers who belong to the business-owning family and work in the firm owned by their family.
192
9 Executive pay in family firms: a principal-agent model
The second aspect, which I refer to as the ownership discount, decreases the level of shortterm incentive pay. Equation 9-15 shows that the optimal sensitivity of pay to short-term performance is reduced by the term r2 V 2 C 21 . The level of this ownership discount is 2
moderated by the executive’s level of risk aversion (r), the observability of efforts for long-term performance ( V 2 ), and the degree to which the two tasks are cost substitutes 2
( C 21 ). Should short-term incentives be higher or lower for family managers than for non-family managers? The answer is not clear, since the two effects counteract each other. The signaling discount implies higher levels of short-term incentives for family executives than for non-family executives, whereas the ownership discount induces lower levels. The answer becomes clearer in cases where one of the two effects does not apply. Consider, for example, a family manager working in a multi-generation family firm with several (hundreds of) owners (e.g., Haniel Group284 or Freudenberg Group285). In this case, the ownership discount is low, but the signaling discount is present. Ceteris paribus, such a family executive should be offered a higher-powered contract with regard to short-term incentives as compared to a contract offered to a non-family executive in a similar position. The following numerical example illustrates this situation.
284
See http://www.haniel.com (accessed November 15th, 2008).
285
See http://www.freudenberg.com (accessed November 15th, 2008).
9 Executive pay in family firms: a principal-agent model
193
Table 9-2: Numerical example comparing a family and a non-family executive Family executive
Non-family executive
Value attributed to signaling () Value the family attributes to short-term performance Incentives for long-term performance (2) Risk aversion (r) Responsiveness for short-term incentives (C11) Degree to which the tasks are substitutes (C12) Measurement error in efforts for short-term performance (21) Measurement error in efforts for long-term performance (22)
0.2 1 0.5 0.5 0.5 0.5 0.5 0.5
0.8 1 0.5 0.5 0.5 0.5 0.5 0.5
Incentives for short-term performance (1)
0.35
0.06
The signaling discount is greater for the non-family executive than for family executive, which leads to a higher-powered contract for the family manager (1=0.35 vs. 1=0.06).
9.5 9.5.1
Discussion Implications for theory: avenues for further empirical research
The model’s results have implications for the succession debate, research about the role of non-family executives, and research about executive pay in family firms. Some of the model’s normative predictions call for empirical research. Succession
The model’s results contribute to the literature on succession issues.286 Succession is the most important concern in family firms (Chua et al., 2003). The model shows that designing a compensation contract for a non-family executive is difficult. The family needs to know about the executive’s value attributed to sending a signal of strong short-term performance to the labor market for executives, her degree of risk aversion, and her responsiveness for short-term incentives. The implementation of the contract can be even more difficult; this is why it can be advantageous to choose a family successor even if the family candidate is less qualified than a non-family executive is. The model also shows that these problems become more severe when the executive’s efforts for shortterm performance are more difficult to observe and when the executive is risk-averse and highly responsive to short-term incentives. Following this logic, there should be more within-family successions in industries in which the executive’s efforts are difficult to observe (e.g., cyclical or fast-
286
For a summary of this literature, see Handler (1994) or Chua et al. (2003).
194
9 Executive pay in family firms: a principal-agent model
moving industries). Empirical research on succession in family firms should thus consider the industry context. Another interesting conclusion concerns the nature of the labor market for executives. From the model, it also follows that a flexible labor market increases the value of signaling for the executive and thus reduces the optimal intensity of short-term incentives in the contract. If, due to restrictions from the market for executives, this contract cannot be implemented, a family candidate might be a better choice for successor than a non-family executive, even if this candidate is less qualified. This leads to an interesting phenomenon: on the one hand, a flexible labor market for executives increases the chances of finding a suitable non-family executive. On the other hand, it is just this flexibility in the labor market that leads to problems in the design of the optimal contract. Non-family executives
The model’s findings also contribute to the issue of non-family executives, which is the second most important concern for many family firms (Chua et al., 2003). The business-owning family should be aware that the executive may trade off between her efforts for long- and short-term performance and should have a close look at the candidates and their motivations. So far, little is known about the motivations of non-family executives to join family firms. Empirical research in this area would be promising. Interesting questions are to what extent non-family executives are interested in sending a signal of strong short-term performance to the market for executives; how this interest in signaling depends on age, industry experience, and the industry context; whether this interest in signaling is stronger for interim managers than for other non-family executives, and whether a non-family executive regards her job in the family firm as an interesting career path rather than as merely a stepping stone to a higher-paid job outside the family firm. Another issue of great relevance concerns the selection process of finding a non-family executive. For family business owners who want to grow the business, it is of great importance to find non-family executives who share their long-term view of the firm. Empirical research might help family business owners in this selection process. What are the criteria they should use in the selection? For example, is it better to have a track record with many firms? Is it better to have an MBA from a renowned business school? How important is experience as a consultant? Is there a trade-off between the qualifications of the candidate and her desire to make a strong short-term contribution? Executive pay
Empirical research on executive pay in family firms (e.g., Chrisman et al., 2007; Deckop, 1988; Gomez-Mejia et al., 2003; McConaughy, 2000) can benefit from the results of this model.
9 Executive pay in family firms: a principal-agent model
195
McConaughy (2000) and Gomez-Meija et al. (2003) find that family CEOs receive less incentivebased pay than non-family CEOs working in family firms. McConaughy (2000) termed this observation the family incentive alignment hypothesis. It is argued that a family CEO’s interests are closer to the goals of the firm, thus reducing the necessity for incentives. The results of this model show that there is also an effect going in the opposite direction. Relative to non-family executives, family executives are less interested in signaling strong short-term performance, making their optimal contracts higher-powered with regard to short-term incentives. Empirical research could analyze how the two effects interact with each other and in which context one or the other is more important. An interesting case would be to look at family and non-family executives who are active on the same board and have similar responsibilities. As mentioned above, subjective performance measures and voluntary bonus payments have been suggested as an alternative to objective performance measures (Baker et al., 1994; MacLeod and Malcomson, 1989). Given that many family firms are often not listed on the stock market and that long-term goals are particularly difficult to measure, it would be interesting to see whether family firms rely more on subjective performance measures and voluntary bonus payments. 9.5.2
Implications for practice: some guidelines for family business owners
The results of the model have a number of implications for family business practices in the category of compensation guidelines. Some guidelines are specific to family firms; others are more general. It is assumed that the family is interested in long-term performance and that high incentives for short-term performance cause the manager to divert effort from achieving long-term performance. 1.
If non-family executives want to signal strong short-term performance, then they should be offered low incentives for short-term performance.
2.
If the labor market for executives is flexible, then the value of signaling increases and the executive should receive low incentives with regard to short-term performance.
3.
If the non-family executive is employed on an interim basis, then signaling is of great importance, and the executive should receive low incentives with regard to short-term performance.
4.
If the non-family executive is responsive to short-term incentives, then the contract should be low-powered with regard to short-term performance.
196
5.
9 Executive pay in family firms: a principal-agent model
If short-term performance is difficult to observe, then the contract should be low-powered with regard to short-term performance. If short-term performance cannot be observed at all, then the family should offer a flat wage or rely on subjective performance evaluations and voluntary bonus payments.
6.
Family and non-family executives with similar tasks and similar responsibilities should not necessarily be offered the same contract. The structure of pay may differ across board members. If non-family executives want to signal strong short-term performance, then they should be offered a lower-powered contract with regard to short-term performance as compared to a contract offered to a non-family executive in a similar position.
7.
If, due to restrictions from the market for executives, the optimal contract cannot be implemented, then it might be optimal to choose a family executive, even if this family executive is less qualified.
9.6
Summary and conclusions In this chapter, I have analyzed the compensation contract of a non-family executive working
in a family firm. Due to the increasing importance of non-family executives (not only as CEOs, but also in other positions such as CFO or COO), this topic is of great importance for family firms that deserves more research. To the best of my knowledge, this is the first work that treats this problem in a theoretical, normative way. The good thing about this approach is that concrete advice can be given, as summarized in the guidelines above. The caveat is that the results often depend on the specific assumptions made (as is always the case with economic models). My model assumes that the executive is risk-averse, maximizes her utility, and that efforts for long-term and short-term performance are substitutes in the executive’s cost function. Principal-agent analysis and contract theory constitute well-developed tools in standard microeconomics. I believe that there is a huge gap in applying such models to problems of family business research. To illustrate, this chapter’s analysis could be extended in a way that it includes varying degrees of bargaining power and monitoring intensity. Also, it might be promising to have a closer look at the stage before the non-family executive is selected and a contract is designed. The chapter is connected to previous chapters (and the overall research question in this thesis) in that it provides (normative) recommendations about the optimal structure of executive pay in family firms that want to pursue a long-term-oriented strategy. In addition to that, the chapter (and the model) offers an explanation as to why CEOs in family firms are found to have a lower share of stock options in their pay as compared to CEOs in non-family firms (Chapter 8).
10 Summary, implications, and outlook This chapter summarizes the dissertation’s main findings, discusses its limitations, outlines its implications for practice and theory, and provides directions for further research.
10.1
Summary In recent years, firms and their executives have been criticized for focusing too strongly on the
short term and for neglecting investments in assets and capabilities required for long-term success, such as investments in R&D or in employee training. There are many explanations for this shortterm managerial behavior, and most refer to flawed management practices, defective incentive compensation, managerial opportunism, and stock market myopia. At the same time, there has been an evolving discussion on the competitive advantage of family firms relative to non-family firms. As a group, family firms have traditionally been dismissed as old-fashioned and rife with conflict. Family owners were said to be interested in gaining private control, resulting in overconsumption and underinvestment. On the other hand, in recent years, it has been argued that family firms may be more oriented towards the long-term than non-family firms. This is attributed to a stronger alignment of interests between management and owners, as well as to protection from stock market pressures. Moreover, it has been argued that the desire to pass the firm over to the next family generation encourages family firms to focus on the long-term. A large number of firms worldwide are family firms. This is true for small and large firms, as well as for those listed and not listed on stock exchanges. Many of the family firms that exist today are predicated on the success of previous generations, and family firms are often market leaders in their field. So far, however, there is little systematic empirical evidence on the long-term orientation of family firms. This dissertation aims to close this gap using large-scale quantitative data to analyze whether and under which circumstances family firms follow a more long-term-oriented corporate strategy than non-family firms. To this end, the dissertation proceeded as follows. I surveyed the literature on the definition of family firms. There is no clear-cut definition of what constitutes a family firm. The two main dimensions used to define family firms are family management and family ownership. Scholars have proposed various definitions of family firms, depending on the country’s system of corporate governance, the availability of data, and the field to which the study aims to contribute. After a review of existing definitions, I focused on six. The
198
10 Summary, implications, and outlook
definitions covered the entire spectrum of family firms and allowed for differentiation between young and old family firms, as well as between family-managed and non-family-managed family firms. I summarized agency and stewardship theory, the two main theoretical frameworks used in the field of family business research. After establishing the theoretical foundations of family firms, I reviewed the extant literature on the issue of long-term orientation. The concept of long-term orientation is strongly linked to the concept of intertemporal choice. Problems of intertemporal choice occur when the costs and benefits of a particular decision fall into different time periods. Benefits that occur in later periods should be discounted more heavily than benefits occurring in earlier periods. Long-term orientation refers to the discount factor. Firms with a high degree of long-term orientation use low discount rates. This intertemporal trade-off exists with many top-level management decisions. The concept of long-term orientation has many dimensions. For example, it is linked to the level and nature of R&D spending, investments in employee training, brand building, as well as building long-term firm-supplier relationships. In a next step, I discussed whether family firms are more likely to follow strategies oriented towards the long-term. To this end, I relied on the literature on family firms and the literature on the causes of managerial myopia. The results of the discussion show that longterm orientation is an empirical question. A number of arguments exist for both a strong and a weak degree of long-term orientation of family firms. I then formulated research questions and used large-scale, quantitative data to analyze whether and under what conditions family firms have a strong degree of long-term orientation with regard to (1) R&D spending, (2) employment downsizing, and (3) the structure of executive pay. Furthermore, I compared the financial performance of family and non-family firms. The empirical analysis was based on US data. The sample was constructed from the set of Standard & Poor’s 500 firms (as of July 31, 2003). I collected detailed data on the firms’ ownership structures and management compositions from corporate proxy statements submitted to the US Securities and Exchange Commission for 1992-2003. I then expanded the dataset using information from other sources. Financial databases were used to get additional firm, manager, and market data. The final sample is an unbalanced panel dataset with 4,856 observations from 499 firms. I used the six family firm definitions derived in the theory section to divide the sample into several groups of family and non-family firm observations. Depending on the definition used, the share of family firm observations varied between 9% and 36%. This analysis relies on uni- and multivariate methods. Multivariate statistical analysis was performed with both Bayesian and classical regression techniques. Bayesian methods do not rely on null hypothesis testing and assumptions about significance levels. Contrary to classical methods, they are able to determine whether no relationship between
10 Summary, implications, and outlook
199
two variables exists. This is useful if theory is unclear and leads to competing hypotheses. Bayesian methods allow the researcher to state which hypothesis is more likely, instead of rejecting one (or both) of the two hypotheses as irrelevant. In addition, Bayesian methods have strong small sample properties and are robust to multicollinearity problems. The results of the (Bayesian and classical) regression analyses were as follows. Using ROA and market-to-book value as indicators of financial performance, a narrow definition of family firms was found to imply better financial performance. However, when using a broad family firm definition, family and non-family firms showed no difference in financial performance. A similar result was obtained using R&D spending as an indicator of long-term orientation. No evidence was found that family firms in general invest more resources in R&D than non-family firms. Rather, only family firms in which the family owned more than 30% of stock were found to invest more. Furthermore, the results showed that family management is negatively associated with the level of R&D spending, whereas the degree of family ownership was found to have a positive effect. Also, family firm age was found to have a negative impact on the level of R&D spending. The central finding with regard to employment downsizing as an indicator of corporate myopia was that family management and family ownership have different impacts regarding the firms’ engagement in downsizing. Similar to the findings with regard to R&D spending, family management was found to have no effect, whereas family ownership decreased the likelihood of large job cuts. Finally, an analysis of the structure of executive pay showed that family ownership increases the share of base salary in total pay, whereas it decreases the share of stock option pay. Stock option pay has been criticized for inducing short-term managerial thinking. Vesting periods (that is the period of time the executive must wait until she can exercise her stock options) are often only one or two years and are often renegotiated; the executives may be tempted to fool the market by temporarily prop up the stock prices of their firms and then cash out their stock options. In conclusion, the empirical analyses showed that the definition of family firms matters and impacts the results. In particular family firms in which the family still owns a large amount of stock were found to be more oriented towards the long-term than other types of family firms. Furthermore, irrespective of the dimension of long-term orientation analyzed, I found that family ownership has a positive impact on long-term orientation, whereas the impact of family management seems to be neutral. The last chapter is normative and determines the optimal compensation contract for nonfamily executives using principal-agent analysis. This chapter is connected to the empirical chapters through an explanation of how the compensation contract of a non-family executive should be designed if the business-owning family wants to pursue a long-term strategy. I showed that, in a situa-
200
10 Summary, implications, and outlook
tion where the business-owning family is interested in long-term performance but the non-family executive is more interested in short-term performance, the compensation contract should be lowpowered with regard to short-term performance. The intensity of incentives for short-term performance depends on the value of performance signaling to the outside market for corporate executives, the measurement error of the executive’s efforts for short-term performance, the executive’s level of risk aversion, and the executive’s responsiveness to short-term incentives. I formulated guidelines with concrete advice for family business owners.
10.2
Limitations The interpretation of the results comes with some limitations. The data suffer from limitations
that preclude generalizations to the universe of all family firms. First, I cover only large, publicly listed US firms in the years 1992-2003. These findings do not necessarily apply to small firms or to firms that are not publicly listed. Second, most of the family firms in this sample are rather young, and the founder is often the CEO. The mean age of the family firm (using the broad definition) is only 36 years (Table 4-3). The mechanisms described in this dissertation might be different in old, multi-generation family firms. Another important point concerns the national context. Some of this study’s results might be specific to the US where the relationship between firms and the capital market is presumed to be more short-term than in other countries (Porter, 1992). Finally, this dissertation focuses on particular dimensions of long-term orientation, namely R&D spending and human resources. The results may change when other dimensions of long-term orientation are analyzed (e.g., brand building).
10.3
Implications
10.3.1 Theoretical implications
This dissertation contributes to the general management/strategy literature on corporate strategy, to the family business literature, and to the finance literature. Furthermore, the use of Bayesian methods adds to the knowledge of research methods in the field of management research. The main contributions are summarized as follows. General management/strategy literature on corporate myopia. The findings of this disser-
tation contribute to the discussion of the causes of corporate myopia (e.g., Bushee, 1998; Campbell and Marino, 1994; David et al., 2001; Hansen and Hill, 1991; Hirshleifer and Thakor, 1992;
10 Summary, implications, and outlook
201
Laverty, 1996; Narayanan, 1985; Porter, 1990; Thakor, 1990, von Thadden, 1995).287 Corporate governance and ownership structures play an important role in the temporal orientation of corporate strategy. Family ownership has a positive impact on the level of R&D spending (an indicator of long-term orientation), decreases the likelihood of large job cuts (an indicator of myopia), and decreases the share of stock options in the executive pay package. The latter has been criticized for its adverse effects on managerial myopia. The positive effect of family ownership on long-term orientation is in line with explanations of myopic corporate behavior that invoke information asymmetry (Thakor, 1990; von Thadden, 1995) and fluid and impatient capital (Porter, 1990). The first explanation postulates that managers know more about the firm than investors and are thereby forced to use strong short-term results to signal the quality of their management to the owners. According to the latter explanation, corporate myopia may be a result of the short-term relationship between firms and capital providers. Both explanations suggest a positive relationship between the degree of family ownership and long-term orientation, which is supported by the results of this dissertation. Contrary to the effect of family ownership, the effect of family management is neutral. Thus, the results in this dissertation do not support the managerial opportunism explanation of corporate myopia (Hirshleifer and Thakor, 1992; Narayanan, 1985). This explanation postulates that, in order to enhance his or her personal reputation on the external job market or to secure a position, it is optimal for the CEO to adopt a short-term perspective. Given that it is hard for an owner to lay off a family member, family managers should be in safe positions, where job security does not depend on strong short-term performance. In addition, family managers need not worry about their reputation on the external job market. If the managerial opportunism explanation were correct, then long-term orientation would be higher in firms managed by a family member than in firms managed by a nonfamily manager; this is not the case. Family business literature. This dissertation contributes to the evolving literature on family
businesses in several ways. In the family business literature, it is often assumed that family firms are more long-termoriented (Anderson and Reeb, 2003; Berghoff, 2006; Bertrand and Schoar, 2006; Casson, 1999; Guzzo and Abbot, 1990; James, 1999; Le Breton-Miller and Miller, 2006; Miller and Le BretonMiller, 2005; Tagiuri and Davis, 1992; Zellweger, 2007). Apart from some qualitative studies, no study has rigorously analyzed this critical assumption with large-scale, quantitative data. To the best of my knowledge, this dissertation is the first attempt to scrutinize this assumption. The results of
287
See Section 3.3 for a summary of this literature.
202
10 Summary, implications, and outlook
this dissertation show that family firms are not per se more long-term-oriented relative to nonfamily firms. The family firm definition plays an important role. The use of several family firm definitions allows for a comparison of different types of family firms. The findings point to a large heterogeneity among family firms. If family firms are narrowly defined in the way that the family needs to own at least 30% of stock, there seems to be a positive effect of being a family firm on long-term orientation. However, there is no difference between family and non-family firms given a broad family firm definition. This justifies the use of several definitions of family firms (Miller et al., 2008) or the use of continuous measures of family influence, such as the F-PEC (Astrachan et al., 2002; Klein et al., 2005) or the familiness construct (Habbershon and Williams, 1999).288 A dichotomous world consisting of only family and nonfamily firms does not exist. This is an important point to consider for future family business research, regardless of whether the research interest lies in the area of financial performance or corporate strategy. Furthermore, I distinguish between the effects of family ownership and family management, the two main characteristics of family firms. These results have some implications for the discussion on incentive and entrenchment effects of family ownership and management (Claessens et al., 2002; McConnell and Servaes, 1990; Shleifer and Vishny, 1997; Stulz, 1988). In a nutshell, family ownership and family management are found to differ with regard to long-term orientation. I find that the effect of family ownership (above a certain level) is positive, whereas the effect of family management seems to be neutral. This finding does not support the entrenchment hypothesis (Claessens et al., 2002). Family owners do not appear to use their firms as vehicles to gain private control or to exploit minority shareholders. Relative to non-family firms, family-owned firms seem to invest more in long-term capabilities. I also differentiate between young and old family firms. I find that young family firms tend to spend more on R&D than old family firms. This finding is in line with scholars who argue that multi-generation family firms are characterized by sibling rivalries or conflicts between two family generations (Eddleston and Kellermanns, 2007; Harvey and Evans, 1994), leading to short-term corporate behavior. This finding also agrees with authors who argue that over time, family firms become conservative and less entrepreneurial (Miller et al., 2003; Vago, 2004; Ward, 1987). There could be many reasons for such conservatism, including the founder’s reluctance to hand over the business at the right point in time, inadequate attention given to grooming future leaders, or difficulties integrating competent non-family employees into the firm. 288
See Section 2.1 and Table 2-2 for the large number of family firm definitions that have been used; see Section 2.2 for measures of family influence.
10 Summary, implications, and outlook
203
Finally, the dissertation includes a theoretical model analyzing the compensation contract of a non-family executive working in a family firm. The model is based on principal-agent theory. Principal-agent analysis and contract theory constitute a well-developed tool in standard microeconomics. So far, however, few principal-agent models deal with problems specific to family firms. There is a huge gap in applying principal-agent analysis to family business research. One benefit of this normative approach is that under some strict model assumptions, it allows for the derivation of testable hypotheses. Furthermore, concrete advice for family business owners and/or managers becomes possible. The contributions of this particular model to family business theory and practice are summarized in Chapter 9. Finance literature. Using the broad definition of a family firm applied in some finance pa-
pers (e.g., Anderson and Reeb, 2003), I find that family firms do not differ from non-family firms regarding long-term orientation. Hence, the superior performance of family firms cannot be explained by a higher degree of long-term orientation, a widely used argument in the finance and economics literature (e.g., Anderson and Reeb, 2003; Bertrand and Schoar, 2006). This argument requires a more narrow definition of a family firm. The empirical results in this chapter provide some suggestions (e.g., a high level of family ownership). Bayesian methods in management research. Most researchers in economics and manage-
ment rely on null hypothesis significance testing, which has been criticized for various reasons (e.g., Cohen, 1994; Schmidt, 1996; Starbuck, 2006).289 One criticism is that scientific journals almost never publish negative (not statistically significant) results and thus contribute to a biased picture of reality. Furthermore, the statistical significance required for publication (in most cases 5%) is arbitrary. It has no mathematical basis, but is rather based on tradition. Bayesian methods can be considered as an alternative method of hypothesis testing.290 This dissertation uses Bayesian methods in addition to classical statistical methods. Bayesian methods are widely used in the biological and medical sciences as well as in some areas of economics and marketing science. However, these methods have only rarely been used in the management and strategy literature. This dissertation shows that Bayesian methods have some unique advantages that can help to increase the understanding of management problems. My experiences with the use of Bayesian methods are as follows. (1) In the face of competing hypotheses291, Bayesian methods are particularly useful.
289
290 291
At the 2008 Academy of Management Meeting in Anaheim, Bill Starbuck, a former president of the Academy, organized a professional development workshop making the case against null hypothesis testing. Bayesian analysis is described in detail in Section 4.3. E.g., theory A predicts a positive influence of a particular variable, whereas theory B predicts a negative influence of this variable.
204
10 Summary, implications, and outlook
Contrary to classical methods, the Bayesian approach does not simply reject one (or both) of the two hypotheses as irrelevant, but allows the researcher to state which hypothesis is more likely. This enables a fine-grained approach to hypothesis testing. (2) Similarly, Bayesian methods can be used for sensitivity analysis. The result of Bayesian analysis – the posterior distribution of the effect of a particular variable – gives a very precise picture of whether the way to define an important construct is defined (in this dissertation: the family firm definition) has an impact on the results. (3) Since Bayesian analysis is not based on null hypothesis testing and statistical significance, multicollinearity and sample size are less of a problem. Bayesian methods are useful if the variables of interest are strongly correlated with control variables and if the size of the sample is small.292 (4) Finally, when classical methods lead to a highly significant result, Bayesian methods tend to agree. In this way, Bayesian methods can serve as a useful robustness check. 10.3.2 Practical implications
This dissertation offers practical implications for family and non-family firms, the owners of family firms, non-family investors, and stakeholders of family firms. Family firms. Firms with a large family shareholder can use the results of this dissertation in
their communication with the local community or the federal state. In general, the state will favor a firm oriented towards the long-term to one oriented towards the short-term, and might be willing to make concessions to family firms and their owners when they learn of their strong long-term commitment (e.g., a reduction in inheritance tax or a privileged treatment in public procurement). Moreover, those firms can use the results to market themselves as attractive employers. In particular, employees who undertake high relationship-specific investments (e.g., employees in an R&D or a specialized sales department) benefit from job stability and a long-term-oriented corporate strategy. Firms with a large family shareholder can highlight their relatively stronger level of job stability and long-term orientation when recruiting specialized personnel. Another implication concerns the result that older family firms tend to spend less on R&D than young family firms. This can be interpreted as multi-generation family firms suffering from conservatism and resistance to strategic change. Multi-generation family firms should be aware of this problem and take measures against it. They may wish to follow the example of the German multi-generation family firm Freudenberg Group, described in the introductory chapter. The Freudenberg Group has implemented a plan to limit dividends to family owners, thereby enabling the management to consistently invest in projects that are important to the future of the firm. 292
Note that in this thesis, family firm characteristics are strongly correlated with variables such as firm size and firm age.
10 Summary, implications, and outlook
205
Non-family firms. Generally, family firms do not seem to perform worse than non-family
firms. In contrast, the results here and from other studies indicate that family firms, narrowly defined, have superior performance. At the same time, research has shown that short-sighted strategies such as deep job cuts in times of financial distress can have a negative impact on shareholder value and/or profitability. Is the superior performance of some family firms linked to their tendency to avoid short-sighted actions? The results in this chapter provide some indications that this may be the case. Hence, if non-family firms as well as their owners and managers want to learn about the success of some extremely successful family firms, they should take a closer look at the overall strategy of the firm and particularly at the firm’s long-term orientation. Family business owners. The theoretical model on the compensation of non-family managers
in Chapter 9 offers some insights for family business owners who want to (or need to) employ a non-family manager. The model shows the difficulty of designing an effective compensation contract. The difficulty is to find a contract that incentivizes the non-family manager to behave in the firm’s long-term interest. This is difficult, because the manager may want to send a signal of strong short-term performance to the outside market for corporate executives and because the effort associated with long-term performance is usually more difficult to observe than the effort associated with short-term performance. The model in Chapter 9 derives the optimal contract. The conclusions for family business owners are summarized as guidelines. Non-family investors investing in firms with large family shareholders. These results have
implications for non-family shareholders investing in family-owned firms, such as pension funds, banks, insurance companies, or other financial investors. Managerial actions in firms with a large family shareholder seem to be oriented towards long-term goals. This may be good news for (nonfamily) investors who prefer this type of firm strategy. On the other hand, this may be bad news for investors, such as mutual funds, who prefer a stable stream of dividends. The latter type of investor will find it hard to convince the management of a firm with a large family shareholder to pay out dividends if this harms the firm’s long-term strategy (e.g., a deep cut in the firm’s R&D budget). Stakeholders dealing with firms who have large family shareholders. Stakeholders that
deal with firms include employees, important suppliers, large customers, or the local community in which the firm is located. These stakeholders are affected by the firm’s strategic choices, but are not owners and/or managers of the firm. These results indicate that firms with large family shareholders are more long-term oriented with regards to their R&D and human resources strategies than other firms. Stakeholders are thus less likely to suffer from their relationship-specific investments. For example, in times of financial distress, employees in R&D departments should be less afraid of losing their jobs. Similarly, suppliers who engage in collaborative R&D with the firm should be less
206
10 Summary, implications, and outlook
concerned about the future of their joint ventures. This is also true for the local community. For example, subsidies for infrastructure investments or basic research are less likely to be lost if the firm shows a long-term commitment to the community in which its R&D or production site is located. In conclusion, from the perspective of stakeholders, investing in a relationship with a firm that has a large family shareholder seems to be less risky than investing in a relationship with a firm that does not have a large family shareholder.
10.4
Directions for further research There are several promising avenues for further empirical research. One avenue would be to
look more closely at different types of family firms. This dataset is confined to large, publicly listed US firms, which is not representative. One may ask whether there is a difference in the long-term orientation of private and publicly listed family firms. In addition, what happens to long-term orientation when there is a conflict within the owning family? Another direction for future research is to focus on the aspects of long-term orientation that are not covered in this dissertation, such as brand building or relationships with suppliers or customers. A third possibility would be to address the issue of long-term orientation more holistically, focusing on the individual and organizational factors that determine long-term orientation (e.g., employees’ inclination to invest in firm-specific knowledge). Another related avenue would be to look at the firms’ actions towards their workers. The use of linked employer-employee datasets, which are already widely used in labor economics, could help to answer questions such as: do family firms invest more in the skills of their workers than non-family firms? What happens when the skills of the workers and the skills that the firm requires no longer match? Do family firms invest in the training of their workers or do they refer to other (short-term) measures, such as outside recruiting or outsourcing? What is the salary structure of workers in family and non-family firms? Does it include performance pay based on indicators of long-term success? Finally, another possibility would be to relate differences in long-term orientation to differences in financial performance, which would be of interest to finance and management scholars alike. There are also possibilities for further research connected to the principal-agent model presented in Chapter 9. Principal-agent analysis is a well-developed tool in standard microeconomics. So far, there is a gap in applying principal-agent analyses to family business topics. For example, such models could shed further light on the issue of succession, the role of non-family managers working in family firms, and the relationships within the owning family. The results of such principal-agent models relating to family firms are likely to differ from the results of standard principalagent models. This is due to the fact that family firms have unique characteristics, such as altruism
10 Summary, implications, and outlook
207
towards members of their own family or the pursuit of family rather than business goals. Moreover, family firms are characterized by a blend of two systems with differing goals, processes, and values: a family system and a business system.293 Building models that capture the uniqueness of family firms is both a challenge and an opportunity. I hope that this dissertation inspires future research in this area.
293
See Pieper and Klein (2007) for the systems approach to characterizing family firms. Pieper and Klein (2007) refer to four subsystems: a family system, an ownership system, a management system, and a business system.
Bibliography
A
Abrahamson, E. 1991. Managerial fads and fashions: the diffusion and rejection of innovations. Academy of Management Review, 16(3): 586-612. Abrahamson, E. 1996. Management fashion. Academy of Management Review, 21(1): 254-285. Achleitner, A.-K., Ampenberger, M. 2006, German und Swiss Entrepreneurial Index – StyleIndices zur Wahrnehmung eigentümerdominierter Unternehmen an öffentlichen Kapitalmärkten. Der Schweizer Treuhänder, 9: 621-626. Achleitner, A.-K., Kaserer, C., Moldenhauer, B. 2005a, German Entrepreneurial Index (GEX®) – ein Style-Index zur Performance eigentümergeführter Unternehmen. FinanzBetrieb, 2(1): 118126. Achleitner, A.-K., Kaserer, C., Moldenhauer, B. 2005b, Entrepreneurial Firms – Einfluss der Eigentümerstruktur auf Unternehmensperformance und Kapitalmarktfähigkeit, Betriebsberater, 60(16): 2-8. Achleitner, A.-K., Geidner, A., Klöckner, O. 2006. Der Beitrag von Private Equity und Venture Capital zur Beschäftigung in Europa. FinanzBetrieb, 8(3):140-147. Achleitner, A.-K., Schraml, S. C., Klöckner, O. 2007. Finanzierung von Familienunternehmen – Wie professionell ist die Unternehmensfinanzierung tatsächlich? SSRN working paper. http://ssrn.com/abstract=1117971. Achleitner, A.-K., Betzer, A., Gider, J. 2008a. Investment rationales of hedge funds and private equity funds in the German stock market. SSRN working paper. http://ssrn.com/abstract=1292896. Achleitner, A.-K., Andres, C., Weir, C. 2008b. Economic consequences of private equity investments on the German stock market. SSRN working paper. http://ssrn.com/abstract=1086598. Achleitner, A.-K., Nathusius, E., Herman, K., Lerner, J. 2008c. Impact of private equity on corporate governance, employment and empowerment of a family in a divisional buyout: the case of Messer Griesheim. In World Economic Forum (ed.). The Global Economic Impact of Private Equity Report: 91-102. Akerlof, G. 1970. The market for lemons: qualitative uncertainty and the market mechanism. Quarterly Journal of Economics, 84(3): 488-500. Alchian, A. A., Demsetz, H. 1972. Production, information costs and economic organization. American Economic Review, 62(5): 777-795.
210
Bibliography
Alderfer, C. P. 1972. Existence, relatedness, and growth: human needs in organizational settings. New York: Free Press. Aldrich, H. E., Cliff, J. E. 2003. The pervasive effects of family on entrepreneurship: toward a family embeddedness perspective. Journal of Business Venturing, 18(5): 573-596. Alford, W. E. B. 1994. Chandlerism, the new orthodoxy of U.S. and European corporate development. Journal of European Economic History, 23: 631-643. Allen, M. P. 1981. Power and privilege in the large corporation: corporate control and managerial compensation. American Journal of Sociology, 86(5): 1112-1123. Allen, M. P., Panian, S. K. 1982. Power, performance, and succession in the large corporation. Administrative Science Quarterly, 27(4): 538-547. Anderson, R. C., Lee, D. S. 1997. Ownership studies: the data source does matter. Journal of Financial and Quantitative Analysis, 32(3): 311-329. Anderson, R. C., Mansi, S. A., Reeb, D. M. 2003. Founding family ownership and the agency costs of debt. Journal of Financial Economics, 68(2): 263-285. Anderson, R. C., Reeb, D. R. 2003. Founding family ownership and firm performance: evidence from the S&P 500. Journal of Finance, 58(3): 1301-1328. Anderson, R. C., Reeb, D. M. 2004. Board composition: balancing family influence in S&P 500 firms. Administrative Science Quarterly, 49(2): 209-237. Andres, C. 2008. Large shareholders and firm performance – an empirical examination of foundingfamily ownership. Journal of Corporate Finance, 14(4): 431-445. Ang, J. S., Cole, J. S. Lin, J. W. 2000. Agency costs and ownership structure. Journal of Finance, 55(1): 81-106. Appelbaum, S. H., Close, T. G., Klasa, S. 1999. Downsizing: an examination of some successes and more failures. Management Decision, 37(5): 424-436. Arnold, M. C., Gillenkirch, R. M. 2007. Leistungsanreize durch Aktien oder Optionen? Eine Diskussion des State of the Art. Zeitschrift für Betriebswirtschaft, 77(1): 75-99. Ashforth, B. E., Mael, F. 1989. Social identity theory and the organization. Academy of Management Review, 14(1): 20-39. Astrachan, J. H., Klein, S. B. Smyrnios, K.X. 2002. The F-PEC scale of family influence. A proposal for solving the family business definition problem. Family Business Review, 15(1): 4558. Astrachan, J. H., Zellweger, T. 2008. Die Performance von Familienunternehmen: Literaturübersicht und Orientierungshilfe für künftige Forschungsarbeiten. Zeitschrift für KMU und Entrepreneurship, 56(1/2): 83-108.
Bibliography
211
B
Baber, W., Fairfield, P., Haggard, J. 1991. The effect of concern about reported income on discretionary spending decisions: The case of research and development. The Accounting Review, 66(4): 818-829. Baird, I. S., Thomas, H. 1985. Toward a contingency model of strategic risk taking. Academy of Management Review, 10(2): 230-243. Baker, G. P. 1992. Incentive contracts and performance measurement. Journal of Political Economy, 100(3): 598-614. Baker, G. P., Gibbons, R., Murphy, K. J. 1994. Subjective performance measures in optimal incentive contracts. Quarterly Journal of Economics, 109(4): 1125-1156. Baker, G. P., Gibbons, R., Murphy, K. J. 2002. Relational contracts and the theory of the firm. Quarterly Journal of Economics, 117(1): 39-84. Barney, J. B. 1991. Firm resources and sustained competitive advantage. Journal of Management, 17(1): 99-120. Barker, V. L., Mueller, G. C. 2002. CEO characteristics and firm R&D spending. Management Science, 48(6): 782-801. Barontini, R., Caprio, L. 2006. The effect of family control on firm value and performance: evidence from continental Europe. European Financial Management, 12(5): 689-723. Bayes, T. 1763. An essay towards solving a problem in the doctrine of chances. Philosophical Transaction of the Royal Society of London, 53: 370-418. Baysinger, B. D., Kosnik, R. D., Turk, T. A. 1991. Effects of board and ownership structure on corporate R&D strategy. Academy of Management Journal, 34(1): 205-214. Bebchuk, L. A., Fried, J. M. 2003. Executive compensation as an agency problem. Journal of Economic Perspectives, 17(3): 71-92. Bebchuk, L. A., Fried, J. M., Walker, D. I. 2002. Managerial power and rent extraction in the design of executive compensation. University of Chicago Law Review, 69(3): 751-846. Becker, G. 1974. A theory of social interactions. Journal of Political Economy, 82(6): 1063-1093. Becker, G. 1976. Altruism, egoism, and genetic fitness: economics and sociobiology. Journal of Economic Literature, 14(3): 817-826. Becker, G. 1981. A treatise of the family. Cambridge: Harvard University Press. Beckhard, R, Dyer, W. G. 1983. SMR forum: Managing change in the family firm – issues and strategies. Sloan Management Review, 24(3): 59-65. Becht, M., Roell, A. 1999. Blockholdings in Europe: an international comparison. European Economic Review, 43(6): 1049-1056.
212
Bibliography
Beldona, S., Inkpen, A. C., Phatak, A. 1998. Are Japanese managers more long-term oriented than United States managers? Management International Review, 38(3): 239-256. Bennedsen, M. K, Nielsen, M. K., Pérez-González, F., Wolfenzon, D. 2007. Inside the family firm: the role of families in succession decisions and performance. Quarterly Journal of Economics, 122(2): 647-691. Berghoff, H. 2006. The end of family business? The Mittelstand and German capitalism in transition, 1994-2000. Business History Review, 80(2): 263-295. Berle, A., Means, G. 1932. The modern corporation and private property. New York: Macmillan. Bernstein, J. I., Nadiri, M. I. 1998. Financing and investment in plant and equipment and research and development. In Preston, M. H., Quandt, R. E. (eds.), Prices, competition, and equilibrium. New York: Phillip Allan: 233-248. Bertrand, M., Mullainathan, S. 2000. Do CEOs set their own pay? The ones without principals do. NBER working paper no. 7604. Bertrand, M., Mullainathan, S. 2001. Are CEOs rewarded for luck? The ones without principals are. Quarterly Journal of Economics, 116(3): 801-932. Bertrand, M., Schoar, A. 2006. The role of family in family firms. Journal of Economic Perspectives, 20(2): 73-96. Bester, M., Hellwig, M. 1987. Moral hazard and equilibrium credit rationing: an overview of the issues. In G. Bambers, G., Spreeman, K. (eds.), Agency theory, information and incentives. Heidelberg: Springer. Biais, B., Casamatta, C. 1999. Optimal leverage and aggregate investment. Journal of Finance, 54(4): 1291-1323. Block, J. H. 2008. Family management, family ownership, and downsizing: evidence from S&P 500 firms. Working paper. http://ssrn.com/abstract=1099917. Block, J. H. 2008. Are CEOs in family firms paid like bureaucrats? Evidence from Bayesian and frequentist analyses. Working paper. http://ssrn.com/ab-stract=1123965. Block, J. H., Henkel, J. 2008. How to pay a non-family executive in a family firm – a multi-task principal-agent analysis. Working paper. Technische Universität München. Block, J. H., Jaskiewicz, P. 2007. Non-family executives in family businesses: monitoring, private benefits and firm value. Working paper. http://ssrn.com/abstract-=947047. Block, J. H., Köllinger, P. 2008. I can’t get no satisfaction: necessity entrepreneurship and procedural utility. Kyklos, Forthcoming. Block, J. H., Sandner, P. 2008. Necessity and opportunity entrepreneurs and their duration in selfemployment: evidence from German micro data. Journal of Industry, Competition, and Trade, Forthcoming. Block, J. H., Thams, A. 2007a. Bayesianische Ökonometrie. WiSt – Wirtschaftswissen-schaftliches Studium, 36(4): 205-208.
Bibliography
213
Block, J. H., Thams, A. 2007b. Long-term orientation in family firms: a Bayesian analysis of R&D spending. Working paper. http://ssrn.com/abstract=1019208. Block, J. H., Thams, A. 2008. Long-term orientation in family firms: a Bayesian analysis of R&D spending. Academy of Management Proceedings. Block, J. H., Wagner, M. 2007. Opportunity recognition and exploitation by necessity and opportunity entrepreneurs: empirical evidence from earnings equations, Working paper. http://ssrn.com/abstract=899968. Blondel, C. Rowell, N., Van der Heyden, L. 2002. Prevalence of patrimonial firms on Paris stock exchange: analysis of the top 250 companies in 1993 and 1998. Working paper. Insead. Bloom, N., van Reenen, J. 2007. Measuring and explaining management practices across firms and nations. Quarterly Journal of Economics, 122(4): 1351-1408. Blumentritt, T. P., Keyt, A. D., Astrachan, J. H. 2007. Creating an environment for successful nonfamily CEOs: an exploratory study of good principals. Family Business Review, 20(4): 321335. Bolton, P., Scheinkman, J., Xiong, W. 2005. Pay for short-term performance: executive compensation in speculative markets. Journal of Corporation Law, 30: 721-747. Boot, A. W. 2000. Relationship banking: what do we know? Journal of Financial Intermediation, 9(1): 7-25. Boot, A. W., Thakor, A. V. 2000. Can relationship banking survive competition? Journal of Finance, 55(2): 679-713. Böttcher, B., Linnemann, C. 2008. Erfolgsmodell Mittelstand – Fakten und Irrtümer. Finanz Betrieb, Heft 2: 163-167. Breusch, T. S., Pagan, A. R. 1980. The Lagrange multiplier test and its application to model specification in econometrics. Review of Economic Studies, 47(1): 239-253. Brodsky, M. A. 1994. Labor market flexibility: a changing international perspective. Monthly Labor Review, 117(11): 53-60. Brown, B., Hanson, M. E., Liverman, D. M., Merideth, R. W. 1987. Global sustainability: toward definition. Environmental Management, 11(6): 713-719. Brown, C. 1990. Firm’s choice of method of pay. Industrial and Labor Relations Review, 43(3): 165S-182S. Bruce, N., Waldman, M. 1990. The rotten kid theorem meets the Samaritans dilemma. Quarterly Journal of Economics, 105(1): 155-165. Buchanan, J. M. 1975. The samaritan’s dilemma. In Phelps, E. S. (ed.), Altruism, morality and economic theory. New York: Russell Sage Foundation. Burkart, M., Panuzi, F., Shleifer, A. 2003. Family firms. Journal of Finance, 58(5): 2167-2201.
214
Bibliography
Burrough, B., Helyar, J. 2003. Barbarians at the gate: the fall of RJR Nabisco. Ne York: HarperCollins. Bushee, B. J. 1998. The influence of institutional investors on myopic R&D investment behaviour. The Accounting Review, 73(3): 305-333.
BusinessWeek. 2003. Family Inc. November 10: 111-114.
C
Cambra-Fierro, J. J., Polo-Redondo, Y. 2006. The long-term orientation of firm-suppliers relationships analysis of moderating effect of belonging to an agro-food protection system. Journal of Food Products Marketing, 12(4): 79-108. Cambra-Fierro, J. J., Polo-Redondo, Y. 2008. Long-term orientation of the supply function in the SME context – reasons, determining factors and implications. International Small Business Journal, 26(5): 619-646. Cameron, K. S., Sutton, R. I., Whetten, D. A. 1988. Readings in organizational decline: frameworks, research, and prescriptions. Cambridge, MA: Ballinger. Campbell, T. S., Marino, A. M. 1994. Myopic investment decisions and competitive labor markets. International Economic Review, 35(4): 855-875. Cappelli, P. 2004. Why do employers retrain at-risk workers? The role of social capital. Industrial Relations, 43(2): 421-447. Carland, J. W., Hoy, F., Boyton, W., Carland, J. C. 1984. Differentiating entrepreneurs from small business owners: a conceptualisation. Academy of Management Review, 9(1): 354-359. Carlock, R. S., Ward, J. L. 2001. Strategic planning for the family business: parallel planning to unify the family and business. Houndsmill: Palgrave. Cascio, W. F. 1993. Downsizing: what do we know? What have we learned? Academy of Management Executive, 7(1): 95-104. Cascio, W. F. 2002. Strategies for responsible restructuring. Academy of Management Executive, 16(3): 80-91. Cascio, W. F., Young, C. E., Morris, J. R. 1997. Financial consequences of employment change decisions in major U.S. corporations. Academy of Management Journal, 40(5): 1175-1189. Casson, M. 1999. The economics of the family firm. Scandinavian Economic History Review, 47(1): 10-23. Chadwick, C., Hunter, L. W., Walston, S. L. 2004. Effects of downsizing practices on the performance of hospitals. Strategic Management Journal, 25(5): 405-427. Chandler, A. D. 1977. The visible hand: the managerial revolution in American Business. Cambridge: Harvard University Press.
Bibliography
215
Chandler, A. D. 1990. Scale and scope: the dynamics of industrial capitalism. Cambridge: Harvard University Press. Chernikoff, H., Howell, M. 2008. Wave of job cuts sweeps across corporate America. Reuters online, October 23rd, 2008. Chrisman, J. J., Chua, J. H., Litz, R. A. 2004. Comparing the agency costs of family and non-family firms: conceptual issues and exploratory evidence. Entrepreneurship Theory and Practice, 28(4): 335-354. Chrisman, J. J., Chua, J. H., Kellermanns, F. W., Chang, E. P. C. 2007. Are family managers agents or stewards? An exploratory study in privately held family firms. Journal of Business Research, 60(10): 1030-1038. Chua, J. H., Chrisman, J. J., Sharma, P. 1999. Defining the family business by behavior. Entrepreneurship Theory and Practice, 23(4): 19-39. Chua, J. H., Chrisman, J. J., Sharma, P. 2003. Succession and nonsuccession concerns of family firms and agency relationships with nonfamily managers. Family Business Review, 16(2): 89107. Church, R., Fishlow, A., Fligstein, N., Hughes, T., Kocka, J. Morikawa, H., Scherer, F. M. 1990. Review “Scale and Scope”: A Review Colloquium. Business History Review, 64(4): 690-735 Church, R. 1993. The family firm in industrial capitalism: international perspectives on hypotheses and history. Business History, 35(4): 17-43. Claessens, S., Djankov, S., Lang, L. H. P. 2000. The separation of ownership and control in East Asian corporations. Journal of Financial Economics, 58(1): 81-112. Claessens, S., Djankov, S., Fan, J. P. H., Lang, L. H. P. 2002. Disentangling the incentive and entrenchment effects of large shareholdings. Journal of Finance, 57(6): 2741-2771. Claessens, S., Tzioumis, K. 2006. Ownership and financing structures of listed and large non-listed corporations. Corporate Governance: An International Review, 14(4): 266-276. Coase, R. H. 1976. Adam Smith’s view of man. Journal of Law and Economics, 19(3): 529-546. Cohen, J. 1994. The earth is round (p<.05). American Psychologist, 49(12): 997-1003. Cohen, W. M., Levinthal, D. A. 1990. Absorptive capacity: a new perspective on learning and innovation. Administrative Science Quarterly, 35(1): 128-152. Corbetta, G., Salvato, C. 2004. Self-serving or self-actualizing? Models of man and agency costs in different types of family firms: a commentary on “comparing the agency costs of family and non-family firms: conceptual issues and exploratory evidence. Entrepreneurship Theory and Practice, 28(4): 335-354. Core, J. E., Holthausen, R. W., Larcker, D. F. 1999. Corporate governance, chief executive officer compensation, and firm performance. Journal of Financial Economics, 51(3): 371-406. Covin, J., Slevin, D. 1989. Strategic management of small firms in hostile and benign environments. Strategic Management Journal, 10(1): 75-87.
216
Bibliography
Cronqvist, H., Nilsson, M. 2003. Agency costs of controlling minority shareholders. Journal of Financial and Quantitative Analysis, 38(4): 695-719. Cumming, G., Finch, S. 2001. A primer on the understanding, use and calculation of confidence intervals that are based on central and non-central distributions. Educational and Psychological Measurement, 61(4): 532-575.
D
Daily, C. M., Dollinger, M. J. 1992. An empirical examination of ownership structure in family and professionally-managed firms. Family Business Review, 5(2): 117-136. Daily, C. M., Thompson, S. S. 1994. Ownership structure, strategic posture, and firm growth: an empirical examination. Family Business Review, 7(3): 237-250. David, P., Hitt, M., Gimeno, J. 2001. The influence of activism by institutional investors on R&D. Academy of Management Journal, 44(1): 144-157. Davis, J. H., Schoorman, F. D., Donaldson, L. 1997. Toward a stewardship theory of management. Academy of Management Review, 22(1): 20-47. DeAngelo, H., DeAngelo, L. 1985. Managerial ownership of voting rights: a study of public corporations with dual classes of common stock. Journal of Financial Economics, 14(1): 33-69. Deci, E. L., Ryan, R. M. 2000. The “what” and “why” of goal pursuits: human needs and the selfdetermination of behavior. Psychological Inquiry, 11(4): 227-268. Dechow, P. M., Sloan, R. G. 1991. Executive incentives and the horizon problem: an empirical investigation. Journal of Accounting and Economics, 14(1): 51-89. Dechow, P. M., Hutton, A. P., Sloan, R. G. 1996. Economic consequences of accounting for stockbased compensation. Journal of Accounting Research, 34(Supplement): 1-20. Deckop, J. 1988. Determinants of chief executive officer compensation. Industrial and Labor Relations Review, 41(2): 215-226. Demsetz, H. 1983. The structure of ownership and the theory of the firm. Journal of Law and Economics, 26(2): 375-390. Demsetz, H., Lehn, K. 1985. The structure of corporate ownership: causes and consequences. Journal of Political Economy, 93(6): 1155-1177. Denis, D. J., Serrano, J. 1996. Active investors and management turnover following unsuccessful control contests. Journal of Financial Economics, 40(2): 239-266. Denis, D. J., Sarin, A. 1999. Ownership and board structures in publicly traded corporations. Journal of Financial Economics, 52(2): 187-223. Denison, D. R., Mishra, A. K. 1995. Toward a theory of organizational culture and effectiveness. Organization Science, 6(2): 204-223.
Bibliography
217
Denison, D. R., Lief, C., Ward, J. L. 2004. Culture in family-owned enterprises: recognizing and leveraging unique strengths. Family Business Review, 17(1): 61-70. Deniz, D., Suarez, M. K. C. 2005. Corporate social responsibility and family business in Spain. Journal of Business Ethics, 56(1): 27-36. Dharwadkar, R., George, G., Brandes, P. 2000. Privatization in emerging economies: an agency theory perspective. Academy of Management Review, 25(3): 650-669. Dittmann, I., Maug, E. 2007. Lower salaries and no options? On the optimal structure of executive pay. Journal of Finance, 62(41): 303-343. Dittmann, I., Maug, E. G. and Schneider, C. 2009. Bankers on the boards of German firms: what they do, what they are worth, and why they are (still) there. Review of Finance, Forthcoming. Dlugosz, J., Fahlenbrach, R., Gompers, P., Metrick, A. 2006. Large blocks of stock: prevalence, size, and measurement. Journal of Corporate Finance, 12(3): 594-618. Donaldson, L., Davis, J. H. 1991. Stewardship theory or agency theory: CEO governance and shareholder returns. Australian Journal of Management, 16(1): 49-64. Donckels, R., Frohlich, E. 1991. Are family businesses really different? European experiences from STRATOS. Family Business Review, 4(2): 149-160. Doucouliagos, C. 1994. A note on the evolution of homo economicus. Journal of Economics Issues, 3: 877-883. Dutta, S., Reichelstein, S. 2003. Leading indicator variables, performance measurement, and longterm versus short-term contracts. Journal of Accounting Research, 41(5): 837-866. Dyck, A., Zingales, L. 2004. Private benefits of control: an international comparison. Journal of Finance, 59(2): 537-600. Dyer, W. G. 1989. Integrating professional management into a family owned business. Family Business Review, 2(3): 221-235. Dyer, W. G. 1994. Potential contributions of organizational behavior to the study of family-owned businesses. Family Business Review, 7(2): 109-131. Dyer, W. G., Whetten, D. A. 2006. Family firms and social responsibility: preliminary evidence from the S&P 500. Entrepreneurship Theory and Practice, 30(6): 785-802.
E
Eddleston, K. A., Kellermanns, F. W. 2007. Destructive and productive family relationships: a stewardship theory perspective. Journal of Business Venturing, 22(4): 545-565. Ehrhardt, O., Nowak, E. 2003a. The effect of IPO’s on German family-owned firms: governance changes, ownership structure, and performance. Journal of Small Business Management, 41(2): 222-232.
218
Bibliography
Ehrhardt, O., Nowak, E, 2003b. Private benefits and minority shareholder expropriation (or what exactly are private benefits of control?). Working paper. http://ssrn.com/abstract=423506. Einstein, A. 1916. Zur Quantentheorie der Strahlung. Physikalische Gesellschaft Zürich. Mitteilungen, 18: 47-62. Eisenhardt, K. 1989. Agency theory: an assessment and review. Academy of Management Review, 14(1): 57-74. Erickson, G., Jacobson, R. 1992. Gaining comparative advantage through the discretionary expenditures: the return to R&D and advertising. Management Science, 38(9): 1264-1279.
F
Faccio, M., Lang, L. H. P. 2002. The ultimate ownership of Western European corporations. Journal of Financial Economics, 65(3): 365-395. Fama, E. F. 1980. Agency problems and the theory of the firm. Journal of Political Economy, 88(2): 288-307. Fama, E., Jensen, M. C. 1983. Separation of ownership and control. Journal of Law and Economics, 26(2): 301-325. Feltham, G. A., Wu, M. G. H. 2001. Incentive efficiency of stock versus options. Review of Accounting Studies, 6(1): 7-28. Fisher, I. 1930. The theory of interest. New York: Macmillan. Flören, R. H. 2002. Crown princes in the clay: an empirical study on the tackling of succession challenges in Dutch family farms. Dissertation: Universiteit Nyenrode. Fockenbrock, D., Terpitz, K. 2008. Der Weg zum Bonus wird für Chefs länger. www.handelsblatt.com, October 31st. Franks, J., Mayer, C. 1994. Corporate control: a comparison of insider and outsider systems. Working paper. London Business School. Franks, J., Mayer, C. 2001. Ownership and control of German corporations. Review of Financial Studies, 14(4): 943-977. Freeman, S. J., Cameron, K. S. 1993. Organizational downsizing: a convergence and reorientation framework. Organization Science, 4(1): 10-29. Frey, B. S., Osterloh, M. 2005. Yes, managers should be paid like bureaucrats. Journal of Management Inquiry, 14(1): 96-111. Fryar, E. O., Arnold, F. T., Dunn, J. E. 1988. Bayesian evaluation of a specific hypothesis. American Journal of Agricultural Economics, 70(3): 685-692. Fuller, J., Jensen, M. 2002. Just say no to Wall Street: putting a stop to the earnings game. Journal of Applied Corporate Finance, 14(4): 41-46.
Bibliography
219
G
Gabler Wirtschaftslexikon 13th edition (1992). Wiesbaden: Gabler. Gale, B. 1972. Market share and rate of return. The Review of Economics and Statistics, 54(4): 412423. Gallo, M. A., Sveen, J. 1991. Internationalizing the family business: facilitating and restraining factors. Family Business Review, 4(2): 181-190. Gallo, M. A., Tapies, J., Cappuyns, K. 2004. Comparison of family and nonfamily businesses: financial logic and personal preferences. Family Business Review, 17(4): 303-318. Gerpott, T. J. 2007. Bewertung von Personalabbauprogrammen aus Aktionärssicht: eine Bestandsaufnahme der empirischen Ereignisstudienforschung. Journal für Betriebswirtschaft, 57(1): 3-35. Gibbons, R. 1998. Incentives in organizations. Journal of Economic Perspectives, 12(4): 115-132. Gibson, J. L., Ivancevich, J. M., Donneley, J. H. 1991. Organizations. Homewood: Irwin. Gildersleeve, R. 1999. Winning business: how to use financial analysis and benchmarks to outscore your competition. Amsterdam: Elsevier. Gomez-Mejia, L. R., Nunez-Nickel, M., Gutierrez, I. 2001. The role of family ties in agency contracts. Academy of Management Journal, 44(1): 81-95. Gomez-Mejia, L. R., Larraza-Kintana, M., Makri, M. 2003. The determinants of executive compensation in family-controlled public corporations. Academy of Management Journal, 46(2): 226237. Gomez-Mejia, L. R., Wiseman, R. M. 2007. Does agency theory have universal relevance? A reply to Lubatkin, Lane, Collin, and Very. Journal of Organizational Behavior, 28(1): 81-88. Goodman, S. N. 2005. Introduction to Bayesian methods I: measuring the strength of evidence. Clinical Trials, 2(4): 282-290. Gorton, G., Schmid, F. A. 2004. Capital, labor, and the firm: a study of German codetermination. Journal of the European Economic Association, 2(5): 863-905. Graves, S. B. 1988. Institutional ownership and corporate R&D in the computer industry. Academy of Management Journal, 31(2): 417-428. Greenhalgh, L., Lawrence, A. T., Sutton, R. I. 1988. Determinants of work force reduction strategies in declining organizations. Academy of Management Review, 13(2): 241-254. Griliches, Z., Mairesse, J. 1984. Productivity and R&D at the firm level, Griliches, Z. (ed.), R&D, patents and productivity. Chicago: University of Chicago Press. Guellec, D., van Pottelsberghe de la Potterie, B. 2007. The economics of the European patent system: IP policy for innovation and competition. Oxford: Oxford University Press.
220
Bibliography
Gugler, K. 2003. Corporate governance, dividend payout policy, and the interrelation between dividends, R&D, and capital investment. Journal of Banking and Finance, 27(7): 1297-1321. Guzzo, R., Abbott, S. 1990. Family firms as utopian organizations. Family Business Review, 3(1): 23-33.
H
Habbershon, T. G., Williams, M. L. 1999. A resource-based framework for assessing the strategic advantages of family firms. Family Business Review, 7(1): 1-26. Habbershon, T. G., Pistrui, J. 2002. Enterprising families domain: family influenced ownership groups in pursuit of transgenerational wealth. Family Business Review, 15(3): 223-237. Habbershon, T. G., Williams, M. L., MacMillan, I. C. 2003. A unified systems perspective of family firm performance. Journal of Business Venturing, 18(4): 451-465. Hackethal, A., Schmidt, R. H., Tyrell, M. 2005. Banks and German corporate governance: on the way to a capital market-based system. Corporate Governance: An International Review, 13(3): 397-407. Hackman, J. R., Oldham, G. R. 1976. Motivation through the design of work: test of a theory. Organizational Behavior and Human Performance, 16(2): 250-279. Hahn, J., Doh, J. P. 2006. Using Bayesian methods in strategy research: an extension of Hansen et al.. Strategic Management Journal, 27(8): 783-798. Hall, B. H. 1992. Investment and research and development at the firm level: does the source of financing matter. NBER working paper no. 4096. Hall, B. H. 1993. The stock market’s valuation of R&D investment during the 1980’s. American Economic Review, 83(2): 259-264. Hall, B. H., Mairesse, J., Branstetter, L., Crepon, B. 1999. Does cash flow cause R&D: an exploration using panel data for French, Japanese, United States scientific firms. In Audretsch, D., Thurik, R. (eds.), Innovation, industry evolution, and employment. Cambridge: Cambridge University Press. Hall, B. H., Ziedonis, R. H. 2001. The patent paradox revisited: an empirical study of patenting in the U.S. semiconductor industry, 1979-1995. Rand Journal of Economics, 32(1): 101-128. Hall, B. J. 2003. Six challenges in designing equity-based pay. Journal of Applied Corporate Finance, 15(3): 21-33. Hall, B. J., Liebman, J. B. 1998. Are CEOs really paid like bureaucrats? Quarterly Journal of Economics, 113(3): 653-691. Hall, B. J., Murphy, K. J. 2003. The trouble with stock options. Journal of Economic Perspectives, 17(3): 49-70.
Bibliography
221
Handelsblatt. 2007a. Anleger brechen mit Quartalsdenken: Finanzleute fordern mehr strategische Führung – Familienunternehmen schneiden an der Börse besser ab. February 19. Handelsblatt. 2007b. Familienunternehmen – der entzauberte Mythos. July 23. Handelsblatt. 2008. Exklusiv-Ranking: Deutschlands wachstumsstärkste Kleinunternehmen. October 24. Handler, W. C. 1989. Methodological issues and considerations in studying family businesses. Family Business Review, 2(3): 257-276. Handler, W. C. 1994. Succession in family business: a review of the research. Family Business Review, 7(2): 133-157. Hansen, G. S., Hill, C. W. L. 1991. Are institutional investors myopic? A time-series study of four technology-driven industries. Strategic Management Journal, 12(1): 1-16. Hansen, M., Perry, L. T., Reese, C. S. 2004. A Bayesian operationalization of the resource-based view. Strategic Management Journal, 25(13): 1279-1295. Harhoff, D., 1998. Are there financing constraints for R&D and investment in German manufacturing firms? Annales d’ Economie et de Statistique, 49/50: 421-456. Harris, D., Martinez, J. 1994. Is strategy different for family-owned business? Family Business Review, 7(2): 159-174. Hart, O., Holmstrom, B. 1987. The theory of contracts. In Bewley, T. (ed.), Advances in economic theory. Cambridge: University Press. Harvey, M., Evans, R. E., 1994. Family business and multiple levels of conflict. Family Business Review, 7(4): 331-348. Hausman, J. A. 1978. Specification tests in econometrics. Econometrica, 46(6): 1251-1271. Hayes, R. H., Abernathy, W. J. 1980. Managing our way to economic decline. Harvard Business Review, 58(4): 67-77. Hayes, R. H., Garvin, D. A. 1982. Managing as if tomorrow mattered. Harvard Business Review, 60(3): 71-79. Heckman, J. J. 1979. Sample selection bias as a specification error. Econometrica, 47(1): 153-161. Heckman, J. J., LaLonde R. J., Smith, J. A. 1999. The economics and econometrics of active labor market programs. In Ashenfelder, O., Cards, D (eds.), Handbook of labor economics, 3(2). Amsterdam: Elsevier. Heide, J. B., John, G.. 1990. Alliances in industrial purchasing: the determinants of joint action in buyer-supplier relationships. Journal of Marketing Research, 27(1): 24-36. Hennerkes, B. H. 2004. Die Familie und ihr Unternehmen – Strategie, Liquidität, Kontrolle. Frankfurt: Campus Verlag.
222
Bibliography
Henseler, N. 2006. Beiräte in Familienunternehmen: eine kritische Betrachtung der Ausgestaltung der Beiratsarbeit. Dissertation: Universität Bielefeld. Himmelberg, C. P., Peterson, B. C. 1994. R&D and internal finance: a panel study of small firms in high-tech industries. Review of Economics and Statistics, 76(1): 38-51. Hirshleifer, J. (1958) On the theory of the optimal investment decision. Journal of Political Economy, 66(4): 329-352. Hirshleifer, D., Thakor, A. V. 1992. Managerial reputation, project choice and debt. Review of Financial Studies, 5(3): 437-470. Hitt, M. A., Hoskission, R. E., Kim, H. 1997. International diversification: effects on innovation and firm performance in product-diversified firms. Academy of Management Journal, 40(4): 767-798. Hofstede, G. 1991. Cultures and organizations: software of the mind. London: McGraw-Hill. Holderness, C. G., Sheehan, D. P. 1988. The role of majority shareholders in publicly held corporations: an exploratory analysis. Journal of Financial Economics, 20(1-2): 317-346. Holderness, C. G. 2003. A survey of blockholders and corporate control. Economic Policy Review, 9(1): 51-64. Holmstrom, B. 1979. Moral hazard and observability. The Bell Journal of Economics, 10(1): 74-91. Holmstrom, B. 1982a. Moral hazard in teams. The Bell Journal of Economics, 13(2): 324-340. Holmstrom, B. 1982b. Managerial incentive problems – a dynamic perspective. In Essays in economics and management in honor of Lars Wahlbeck. Helsinki: Swedish School of Economics. Holmstrom, B., Kaplan, S. 2001. Corporate governance and merger activity in the United States: making sense of the 1980s and 1990s. Journal of Economic Perspectives, 15(2): 121-144. Holmstrom, B., Milgrom, P. 1987. Aggregation and linearity in the provision of intertemporal incentives. Econometrica, 55(2): 303-328. Holmstrom, B., Milgrom, P. 1991. Multitask principal-agent analyses: incentive contracts, asset ownership, and job design. Journal of Law, Economics, & Organization, 7(Special Issue): 2452. Hunter, J. E., Schmidt, F. L. 2004. Methods of meta-analysis: correcting error and bias in research findings. London: Sage Publications.
I
IfM Bonn. 2007. Die volkswirtschaftliche Bedeutung der Familienunternehmen. IfM Materialie Nr. 72.
Bibliography
223
J
Jacobs, M. T. 1991. Short-term America: the causes and cures of our business myopia. Boston: Harvard Business School Press. James, H. S. 1999. Owner as manager, extended horizons and the family firm. International Journal of the Economics of Business, 6(1): 41-55. James, H. 2006. Family capitalism: Wendels, Haniels, Falcks, and the Continental European model. Boston: Belknap Press. Jaskiewicz, P. 2006. Performance-Studie börsennotierter Familienunternehmen in Deutschland, Frankreich und Spanien. Dissertation: European Business School. Jell, F. 2008. Patent wars in oligopoly markets. Presentation at summer school on the economics and management of intellectual property rights, University of Bologna. Jensen, M. C. 1983. Organization theory and methodology. Accounting Review, 58(2): 319-338. Jensen, M. C. 1984. Self-interest, altruism, incentives and agency theory. Journal of Applied Corporate Finance, 7(2): 40-45. Jensen, M. C., Meckling, W. H. 1976. Theory of the firm: managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4): 305-360. Jensen, M. C., Murphy, K. J. 1990a. Performance pay and top management incentives. Journal of Political Economy, 98(2): 225-264. Jensen, M. C., Murphy, K. J. 1990b. CEO incentives: It’s not how much you pay, but how. Harvard Business Review, 68(3): 138-153. Johnson, H. T., Kaplan, R. S. 1987. Relevance lost. Boston: Harvard Business School Press. Johnson, S., La Porta, R., Lopez-de-Silanes, F., Shleifer, A. 2000. Tunneling. American Economic Review, 90(2): 22-27. Johnson, J. L. 1999. Strategic integration in industrial distribution channels: managing the interfirm relationships as a strategic asset. Journal of the Academy of Marketing Science, 27(1): 4-18. Jones, G. R. 1983. Transaction costs, property rights, and organizational culture: an exchange perspective. Administrative Science Quarterly, 28(3): 454-467.
K
Kandel, E., Lazear, E. P. 1992. Peer pressure and partnerships. Journal of Political Economy, 100(4): 801-817. Kao, K. C., Hockham, G. A. 1966. Dielectric-fibre surface waveguides for optical frequencies. Proceedings of the Institution of Electrical Engineers, London.
224
Bibliography
Kaplan, S. N. 1994. Top executives, turnover, and firm performance in Germany. Journal of Law, Economics, and Organization, 10(1):142-159. Kaplan, S. 2008. Cognition, capabilities, and incentives: assessing firm response to the fiber-optic revolution. Academy of Management Journal, 51(4): 672-695. Kaserer, C., Klingler, C. 2008. The accrual anomaly under different accounting standards – lessons learned from the German experiment. Journal of Business Finance and Accounting, 35(7-8): 837-859. Kaserer, C., Moldenhauer, B. 2008. Insider ownership and corporate performance – new evidence from Germany. Review of Managerial Sciences, 2(1):1-35. Kellermanns, F. W., Eddleston, K. A. 2004. Corporate entrepreneurship in family firms. Entrepreneurship Theory and Practice, 30(6): 809-830. Kepner, E. 1983. The family and the firm: a coevolutionary perspective. Organizational Dynamics, 12(1): 57-71. Kerr, S. 1975. On the folly of rewarding A, while hoping for B. Academy of Management Journal, 18(4): 769-783 Kerr, S. 1995. On the folly of rewarding A, while hoping for B. Academy of Management Executive, 9(1): 7-14. Kets de Vries, M. F. R. 1993. The dynamics of family controlled firms: the good and the bad news. Organizational Dynamics, 21(3): 59-71. Klein, S. B. 2000. Family businesses in Germany: significance and structure. Family Business Review, 13(3): 157-182. Klein, S. B., Astrachan, J. H., Smyrnios, K. X. 2005. The F-PEC scale of family influence: construction, validation, and further implication for theory. Entrepreneurship Theory and Practice, 29(3): 321-339. Klein, S. B., Bell, F. A. 2007. Non-family executives in family businesses – a literature review. Electronic Journal of Family Business Studies, 1(1): 19-37. Knight, F. 1921. Risk, uncertainty, and profit. Boston: Houghton Mifflin. Knight, G. A. 1997. Cross-cultural reliability and validity of a scale to measure firm entrepreneurial orientation. Journal of Business Venturing, 12(3): 213-225. Kochhar, R., David, P. 1996. Institutional investors and firm innovation: a test of competing hypotheses. Strategic Management Journal, 17(1): 73-84. Koenker, R., Bassett, G. 1978. Regression quantiles. Econometrica, 46(1): 33-50. Koenker, R., Hallock, K. F. 2000. Quantile regression: an introduction. Working paper. http://www.econ.uiuc.edu7~roger/research/intro/intro.html. Koenker, R., Hallock, K. F. 2001. Quantile regression. Journal of Economic Perspectives, 15(4): 143-156.
Bibliography
225
Koop, G. 2003. Bayesian econometrics. Chicester: Wiley. Kotlikoff, L. J. 1987. Justifying public provision of social security. Journal of Policy Analysis and Management, 6(4): 674-689. Kotter, J. P., Heskitt, J. L. 1992. Corporate culture and performance. New York: Free Press.
L
Lambert, R. A., Lanen, W. N., Larcker, D. F. 1989. Executive stock option plans and corporate dividend policy. Journal of Financial and Quantitative Analysis, 24(4): 409-425. Lambert, R., Larcker, D., Weigelt, K. 1993. The structure of organizational incentives. Administrative Science Quarterly, 38(3): 438-461. Lancaster, T. 2004. An introduction to modern Bayesian econometrics. Oxford: Blackwell Publishing. Landes, D. 1949. French entrepreneurship and industrial growth in the nineteenth century. Journal of Economic History, 9(1): 45-61. La Porta, R., Lopez-De-Silanes, F., Shleifer, A. 1999. Corporate ownership around the world. Journal of Finance, 54(2): 471-517. La Porta, R. Lopez-De-Silanes, F., Shleifer, A., Vishny, R. 1998. Law and finance. Journal of Political Economy, 106(6): 1113-1155. La Porta, R. Lopez-De-Silanes, F., Shleifer, A., Vishny, R. 2000. Investor protection and corporate governance. Journal of Financial Economics, 58(1-2): 3-27. Laverty, K. J. 1993. How valid are R&D measures in empirical tests of “short-termism”? Academy of Management Proceedings. Laverty, K. J. 1996. Economic “short-termism”: the debate, the unresolved issues, and the implications for management practice and research. Academy of Management Review, 21(3): 825860. Lawler, E. E. 1986. High involvement management. San Francisco: Jossey-Bass. Leamer, E. E. 1973. Multicollinearity: a Bayesian interpretation. Review of Economics and Statistics, LV(3): 371-380.973), 371-38 Le Breton-Miller, I., Miller, D., Steier, L. P. 2004. Toward an integrative model of effective FOB succession. Entrepreneurship Theory and Practice, 28(4): 305-328. Le Breton-Miller, I., Miller, D. 2006. Why do some family businesses out-compete? Governance, long-term orientation, and sustainable capability. Entrepreneurship Theory and Practice, 30(6): 731-746.
226
Bibliography
Lee, M. S., Rogoff, E. G. 1996. Research note: comparison of small businesses with family participation versus family businesses without family participation: an investigation of differences in goals, attitudes, and family/business conflict. Family Business Review, 9(4): 423-437. Lee, P. M., O’Neill, H. M. 2003. Ownership structures and R&D investments of U.S. and Japanese firms: agency and stewardship perspectives. Academy of Management Journal, 46(2): 212225. Levinson, H. 1971. Conflicts that plague family businesses. Harvard Business Review, 49(March): 90-98. LeMar, B. 2001. Generations- und Führungswechsel im Familienunternehmen. Mit Gefühl und Kalkül den Wandel gestalten. Heidelberg: Springer. Lindbeck, A. Weibull, J. W. 1988. Altruism and time consistency: the economics of a fait accompli. Journal of Political Economy, 96(6): 1165-1182. Liu, C. H., Yermack, D. 2007. Where are the shareholder’s mansions? CEOs home purchases, stock sales, and subsequent company performance. Working paper. http://ssrn.com/abstract=970413. Loewenstein, G., Prelec, D. 1992. Anomalies in intertemporal choice: evidence and an interpretation. Quarterly Journal of Economics, 107(2): 573-597. Loewenstein, G., Thaler, R. 1989. Anomalies: intertemporal choice. Journal of Economic Perspectives, 3(4): 181-193. Logue, D. E. 1985. Counter evidence on management’s investment myopia. California Management Review, 28(1): 169-178. Lorsch, J. L., MacIver, E. M. 1989. Pawns or potentates? The reality of America’s corporate boards. Boston, MA: Harvard Business School Press. Lubatkin, M. H. 2007. One more time: what is a realistic theory of corporate governance? Journal of Organizational Behaviour, 28(1): 59-67. Lubatkin, M., Lane, P. J., Collin, S., Very, P. 2006. An embeddedness framing of governance and opportunism: towards a cross nationally accommodating theory of agency. Journal of Organizational Behaviour, 28(1): 43-58. Lubatkin, M. H., Durand, R., Ling, Y. 2007a. The missing lens in family firm governance theory: a self-other typology of parental altruism. Journal of Business Research, 60(10): 1022-1029. Lubatkin, M. H., Ling, Y., Schulze, W. S. 2007b. An organizational justice-based view of selfcontrol and agency costs in family firms. Journal of Management Studies, 44(6): 955-971. Lubatkin, M. H., Schulze, W. S. Ling, Y., Dino, R. N. 2005. The effects of parental altruism on the governance of family-managed firms. Journal of Organizational Behavior, 26(3): 313-330. Lumpkin; G. T., Dess, G. G. 1996. Clarifying the entrepreneurial orientation construct and linking it to performance. Academy of Management Review, 21(1): 135-172.
Bibliography
227
Lyon, D. W., Lumpkin, G. T., Dess, G. T. 2000. Enhancing entrepreneurial orientation research: operationalizing and measuring a key strategic decision making process. Journal of Management, 26(5): 1055-1085.
M
Maassen, G., Van den Bosch, F. 1999. On the supposed independence of two-tier boards: formal structure and reality in the Netherlands. Corporate Governance: An International Review, 7(1): 31-37. Mace, M. L. 1971. Directors: myth and reality. Boston: Harvard Business School Press. MacLeod, W. B., Malcomson, J. M. 1989. Implicit contracts, incentive compatibility, and involuntary unemployment. Econometrica, 57(2): 447-480. MacLeod, W. B., Parent, D. 1999. Job characteristics, wages and the employment contract. In Polachek, S. W. (ed.), Research in Labor Economics 18. Stamford: JAI Press. Mael, F., Ashforth, . E. 1992. Alumni and their alma mater: a partial test of the reformulated model of organizational identification. Journal of Organizational Behavior, 13(2): 103-123.
Manager Magazin. 2008a. Familienbande. Spezialausgabe Oktober: 24-26. Manager Magazin. 2008b. Gutes altes Geld. Spezialausgabe Oktober: 70-74. Manz, C. C. 1986. Self-leadership: toward an expanded theory of self-influence processes in organizations. Academy of Management Review, 11(3): 585-600. Marginson, D., McAulay, L. 2008. Exploring the debate on short-termism: a theoretical and empirical analysis. Strategic Management Journal, 29(3): 273-292. Maslow, A. H. 1970. Motivation and personality. New York: Harper & Row. Maug, E. 1998. Large shareholders as monitors: is there a trade-off between liquidity and control? Journal of Finance, 53(1): 65-98. Maury, B. 2006. Family ownership and firm performance: empirical evidence from Western European corporations. Journal of Corporate Finance, 12(2): 321-341. Mayer, C. 1997. Corporate governance, competition and performance. Journal of Law and Society, 24(1): 152-176. May, P., Rieder, G., Brose, T. 2002. Empirische Untersuchung über Beiräte in deutschen Familienunternehmen. Bonn. McClelland, D. C. 1975. Power: the inner experience. New York: Irvington. McConaughy, D. L. 1994. Founding-family controlled corporations. Dissertation: University of Cincinnati.
228
Bibliography
McConaughy, D. L. 2000. Family CEOs vs. nonfamily CEOs in the family-controlled firm: an examination of the level and sensitivity of pay to performance. Family Business Review, 13(2): 121-131. McConaughy, D. L., Walker, M. C., Henderson Jr., G. V., Mishra, C. S. 1998. Founding family controlled firms: efficiency and value. Review of Financial Economics, 7(1): 1-19. McConnell, J., Servaes, H. 1990. Additional evidence on equity ownership and corporate value. Journal of Financial Economics, 27(2): 595-612. McGregor, D. 1966. Leadership and motivation. Cambridge: MIT Press. Mehran, H. 1995. Executive compensation structure, ownership, and firm performance. Journal of Financial Economics, 38(2): 163-184. Milgrom, P., Roberts, J. 1992. Economics, Organization, and Management. International edition. New Yersey: Prentice Hall. Miller, D. 1993. The architecture of simplicity. Academy of Management Review, 18(1): 116-139. Miller, D., Friesen, P. H. 1978. Archetypes of strategy formation. Management Science, 24(9): 921933. Miller, D., Le Breton-Miller, I. 2005. Managing for the long run: lessons in competitive advantage from great family businesses. Boston, MA: Harvard Business School Press. Miller, D., Le Breton-Miller, I. 2006. Priorities, practices and strategies in successful and failing family businesses: an elaboration and test of the configuration perspective. Strategic Organization, 4(4): 379-407. Miller, D., Steier, L., Le Breton-Miller, I. 2003. Lost in time: intergenerational succession, change and failure in family business. Journal of Business Venturing, 18(4): 513-531. Miller, D., Le Breton-Miller, I., Lester, R. H., Cannela, A. A. 2007. Are family firms really superior performers? Journal of Corporate Finance, 13(5): 829-858. Morck, R. K. 2003, Why some double taxation may make sense: the special case of inter-corporate dividends. NBER working paper no. 9651. Morck, R. K., Shleifer, A., Vishny, R. 1988. Management ownership and market valuation: an empirical analysis. Journal of Financial Economics, 20(1-2): 293-316. Morck, R. K., Yeung, B. 2003. Agency problems in large family business groups. Entrepreneurship Theory and Practice, 27(4): 367-382. Morck, R. K., Yeung, B. 2004. Family control and the rent-seeking society. Entrepreneurship Theory and Practice, 28(4): 391-409. Morris, J. R., Cascio, W. E., Young, W. F. 1999. Downsizing after all these years: questions and answers about who did it, how many did it, and who benefited from it. Organizational Dynamics, 27(3): 78-87.
Bibliography
229
Mulkay, B., Hall, B. H., Mairesse, J. 2001. Firm level investment and R&D in France and the United States. In Heinz, H., Strauch, R. (eds.), Investing today for the world of tomorrow. Heidelberg: Springer. Murphy, K. J. 1999. Executive compensation. In Ashenfelder, O., Cards, D (eds.), Handbook of labor economics, 3(2). Amsterdam: Elsevier.
N
Naldi, L., Nordqvist, M. Sjöberg, K., Wiklund, J. 2007. Entrepreneurial orientation, risk taking, and performance in family firms. Family Business Review, 20(1): 33-47. Narayanan, M. P. 1985. Managerial incentives for short-term results. Journal of Finance, 40(5): 1469-1484. Nordqvist, M. 2008. Unternehmerische Ausrichtung in Familienunternehmen. Zeitschrift für KMU und Entrepreneurship, 56(1/2): 62-78.
O
O’Donoghue, T., Rabin, M. 2000. The economics of immediate gratification. Journal of Behavioral Decision Making, 13(2): 233-250. O’Neill, H., Lenn, J. D. 1995. Voice of survivors: words that downsizing CEOs should hear. Academy of Management Executive, 9(4): 23-24.
P
Perez-Gonzales, F. 2006. Inherited control and firm performance. American Economic Review, 96(5): 1559-1588. Peterson, R. M., Dibrell, C. C., Pett, T. L. 2002. Long- vs. short-term performance of Western European, Japanese, and U.S. countries: where do they lie? Journal of World Business, 37(4): 245-255. Pieper, T. 2007. Mechanisms to assure long-term family business survival. Frankfurt: Peter Lang. Pieper, T., Klein, S. 2007. The bulleye: a systems approach to modeling family firms. Family Business Review, 20(4): 301-319. Pierce, J. L., Kostova, T., Dirks, K. T. 2001. Toward a theory of psychological ownership in organizations. Academy of Management Review, 26(2): 298-310. Pollack, A. 2008. Merck plans to cut 12% of work force as its sales and profit fall. The New York Times (online version), October 23rd, 2008.
230
Bibliography
Porter, L. W., Steers, R. M., Mowday, R. T., Boulian, P. V. 1974. Organizational commitment, job satisfaction, and turnover among psychiatric technicians. Journal of Applied Psychology, 5: 603-609. Porter, M. E. 1980. Competitive strategy: techniques for analyzing industries and competitors. New York: Free Press. Porter, M. E. 1992. Capital disadvantage: America’s failing capital investment system. Harvard Business Review, 70(5): 65-82. Porter, M. E. 1998. The competitive advantage of nations. New York: Free Press. Prendergast, C. 1999. The provision of incentives in firms. Journal of Economic Literature, 37(1): 7-63.
R
Reichelstein, S. 2000. Providing managerial incentives: cash flows versus accrual accounting. Journal of Accounting Research, 38(2): 243-269. Reitzig, M., Henkel, J., Heath. C. H. 2007. On sharks, trolls and their patent prey – unrealistic damage awards and firms’ strategies of being infringed. Research Policy, 36(1): 134-154. Reynolds, P. D., Camp, S. M., Bygrave, W. D., Autio, E., Hay, M. 2002. Global Entrepreneurship Monitor 2001 Executive Report. Babson College and London Business School. Rezaee, Z. 2001. Financial institutions, valuations, mergers, and acquisitions: the fair value approach. Wiley. Robert, C. P., Casella, G. 2004. Monte Carlo statistical methods. New York: Springer. Roe, M. J. 1993. Some differences in corporate structure in Germany, Japan, and the United States. The Yale Law Journal, 102(8): 1927-2003. Rogoff, E. G., Heck, R. K. Z. 2003. Evolving research in entrepreneurship and family business: recognizing family as the oxygen that feeds the fire of entrepreneurship. Journal of Business Venturing, 18(5): 559-566. Ross, S. 1973. The economic theory of agency: the principal’s problem. American Economic Review, 63(2): 134-139. Rossi, P. E., Allenby, G. M. 2003. Bayesian statistics and marketing. Marketing Science, 22(3): 304-328. Rothschild, M., Stiglitz, J. 1976. Equilibrium in competitive insurance markets: an essay on the economics of imperfect information. Quarterly Journal of Economics, 90(4): 629-649.
Bibliography
231
S
Salancik, G. R., Meindl, J. R. 1984. Corporate attributions as strategic illusions of management control. Administrative Science Quarterly, 29(2): 238-254. Salvato, C. 2004. Predictors of entrepreneurship in family firms. Journal of Private Equity, 7(3): 68-76. Samuelson, P. A. 1989. The judgment of economic science on rational portfolio management: indexing, timing, and long-horizon effects. Journal of Portfolio Management, 16(1): 4-12. Samuelson, P. A. 1994. The long-term case for equities. Journal of Portfolio Management, 21(1): 15-24. Sander, B. 2008. Börsenerfolg Familienunternehmen: Mehr Rendite mit GEX-Werten. München: FinanzBuch Verlag. Schawlow, A. L., Townes, C. H. 1958. Infrared and optical masers. Physical Review, 112(6): 19401949. Schein, E. 1968. Organizational socialization and the profession of management. Industrial Management Review, 9: 80-88. Schein, E. 1983. The role of the founder in creating organizational cultures. Organizational Dynamics, 12(1): 13-28. Schmidt, F. 1996. Statistical significance testing and cumulative knowledge in psychology. Psychological Methods, 1: 115-129. Schoemaker, P. J. 1982. The expected utility model: its variants, purposes, evidence and limitations. Journal of Economic Literature, 20(2): 529-563. Schotter, A., Weigelt, K. 1992. Behavioral consequences of corporate incentives and long-term bonuses. Management Science, 38(9): 1280-1298. Schulze, W. S., Lubatkin, M. H., Dino, R. N. 2002. Altruism, agency and the competitiveness of family firms. Managerial and Decision Economics, 24(4-5): 247-259. Schulze, W. S., Lubatkin, M. H., Dino, R. N. 2003. Toward a theory of agency and altruism in family firms. Journal of Business Venturing, 18(4): 473-490. Schulze, W. S., Lubatkin, M. H., Dino, R. N., Buchholtz, A. K. 2001. Agency relationships in family firms: theory and evidence. Organization Science, 12(2): 99-116. Sharma, P., Chrisman, J. J., Chua, J. H. 1997. Strategic management of the family business: Past research and future challenges. Family Business Review, 10(1): 1-35. Shane, S., Venkataraman, S. 2000. The promise on entrepreneurship as a field of research. Academy of Management Review, 25(1): 217-226. Shanker, M. C., Astrachan, J. H. 1996. Myths and realities: family businesses contribution to the U.S. economy: a framework for assessing family business statistics. Family Business Review, 9(2): 107-119.
232
Bibliography
Shearman, R. 1990. The meaning and ethics of sustainability. Environmental Management, 14(1): 1-8. Shleifer, A., Vishny, R. 1997. A survey of corporate governance. Journal of Finance, 52(2): 737783. Shivdasani, A. 1993. Board composition, ownership structure, and hostile takeovers. Journal of Accounting and Economics, 16(1-3): 167-198. Shivdasani, A., Yermack, D. 1999. CEO involvement in the selection of new board members: an empirical analysis. Journal of Finance, 54(5): 1829-1853. Simon, F. B., Wimmer, R., Groth, T. 2005. Mehrgenerationen-Familienunternehmen: Erfolgsgeheimnisse von Oetker, Merck, Haniel u.a.. Heidelberg: Carl-Auer Verlag. Simon, H. 2007. Hidden Champions des 21. Jahrhunderts – Die Erfolgsstrategien unbekannter Weltmarktführer. Frankfurt: Campus Verlag. Simon, H. 2009. Hidden Champions of the 21st century. New York: Springer. Sloan, R. G. 1996. Do stock prices fully reflect information in accruals and cash flows about future earnings? The Accounting Review, 71(3): 289-315. Smith, A. 1853. The theory of moral sentiments. London: Henry G. Bohn. Song, M., Podoynitsyna, K., Van der Bij, H., Halman, J. I. M. 2008. Success factors in new ventures: a meta-analysis. Journal of Product Innovation Management, 25(1): 7-27. Starbuck, W. H. 2006. The production of knowledge. The challenge of social science research. New York: Oxford University Press. Stavrou, E. T. 1999. Succession in family businesses: exploring the effects of demographic factors on offspring intentions to join and take over the business. Journal of Small Business Management, 37(3): 43-61. Stavrou, E. T., Kassinis, G., Filotheou, A. 2007. Downsizing and stakeholder orientation among the Fortune 500: does family ownership matter? Journal of Business Ethics, 72(2): 149-162. Staw, B. M., McKechnie, P. L., Puffer, S. M. 1983. The justification of organizational performance. Administrative Science Quarterly, 28(4): 582-600. Stein, J. 1988. Takeover threats and managerial myopia. Journal of Political Economy, 96(1): 6180. Stewart, A. 2003. Help one another, use one another: toward an anthropology of family business. Entrepreneurship Theory and Practice, 27(4): 383-396.
Stiftung Familienunternehmen. 2007. Die volkswirtschaftliche Bedeutung der Familienunternehmen. Stuttgart. Stiglitz, J., Weiss, A. 1981. Credit rationing in markets with imperfect information. American Economic Review, 71(3): 393-410.
Bibliography
233
Stulz, R. 1988. Managerial control of voting rights: financing policies and the market for corporate control. Journal of Financial Economics, 20(1-2): 26-54.
Stuttgarter Zeitung. 2008. Auch in 30 Jahren noch vernünftig leben – Familienunternehmer auf der Suche nach den eigenen Erfolgsfaktoren. July 10. Sywottek, C. 2008. Das Matroschka-Prinzip. Brand Eins, 10: 40-50.
T
Tagiuri, R., Davis, J. A. 1992. On the goals of successful family businesses. Family Business Review, 5(1): 43-62. Tagiuri, R., Davis, J. A. 1996. Bivalent attributes of the family firm. Family Business Review, 9(2): 199-208. Tajfel, H, Turner, J. C. 1986. The social identity theory of inter-group behavior. In Worchel, S., Austin, L. W. (eds.), Psychology of intergroup relations. Chicago: Nelson-Hall. Thakor, A. V. 1990. Investment myopia and the internal organization of capital allocation decisions. Journal of Law, Economics and Organization, 6(1): 129-154. Thaler, R. H. Shefrin, H. M. 1981. An economic theory of self-control. Journal of Political Economy, 89(2): 392-406.
The Economist. 2000. In praise of Bayes. September 28th. The Economist. 2006a. Bayes rules: a once-neglected statistical technique may help to explain how the mind works. Jan 5th. The Economist. 2006b. Taking stock of options. July 22nd. Thielemann, U. 2008. Die Gier ist schuld. www.tagesschau.de. September 16th. Thompson, B. 2007. Effect sizes, confidence intervals, and confidence intervals for effect sizes. Psychology in the Schools, 44(5): 423-432. Tirole, J. 2006. The theory of corporate finance. New Jersey: Princeton University Press. Tosi, H. L., Werner, S., Katz, J. P., Gomez-Mejia, L. R. 2000. How much does performance matter? A meta-analysis of CEO pay studies. Journal of Management, 26(2): 301-339.
U
Uchitelle, L. 2008. U.S. layoffs increase as businesses confront crisis. International Herald Tribune (online version), October 26th, 2008.
234
Bibliography
Uhlaner, L. M., Goor-Balk, H. J. M., Masurel, E. 2004. Family business and corporate social responsibility in a sample of Dutch firms. Journal of Small Business and Enterprise Development, 11(2): 186-194. Ulhoi, J. B. 2007. Revisiting the principal-agent theory of agency: comments on the firm-level and cross-national embeddedness theses. Journal of Organizational Behavior, 28(1): 75-80. Useem, J. 2003. Have they no shame? Fortune, 147: 56-64.
V
Vago, M. 2004. Integrating change management. Challenges for family business clients and consultants. Family Business Review, 17(1): 71-80. Villalonga, B., Amit, R. 2006. How do family ownership, control and management affect firm value? Journal of Financial Economics, 80(2): 385-417. Villalonga, B., Amit, R. 2008. How are U.S. family firms controlled? Review of Financial Studies, Forthcoming. Von Neumann, J., Morgenstern, O. 1953. Theory of games and economic behavior. New Jersey: Princeton University Press. Von Preen, A. 2008. Nur wenige Banker verlieren Geld. www.stern.de. October 20th, 2008. Von Thadden, E. L. 1995. Long-term contracts, short-term investment and monitoring. Review of Economic Studies, 62(4): 557-575.
W
Ward, J. L. 1987. Keeping the family business healthy. How to plan for continued growth, profitability, and family leadership. San Francisco: Jossey-Bass. Ward, J. L. 2004. Perpetuating the family business. Marietta: Family Enterprise Publishers. Weimer, J., Pape, J. 1999. A taxonomy of systems of corporate governance. Corporate Governance: An International Review, 7(2): 152-166. Weisbach, M. S. 1988. Outside directors and CEO turnover. Journal of Financial Economics, 20(12): 431-460. Wernerfelt, B. 1984. A resource based view of the firm. Strategic Management Journal, 5(2): 171180. Westhead, P., Howorth, C. 2007. Types of family firms: an exploratory conceptual and empirical analysis. Entrepreneurship and Regional Development, 19(5): 405-431.
Bibliography
235
Westhead, P., Howorth, C., Cowling, M. 2002. Ownership and management issues in firstgeneration and multi-generation firms. Entrepreneurship & Regional Development, 14(3): 247-269. Wiklund, J. 2006. Commentary: “family firms and social responsibility: preliminary evidence from the S&P 500”. Entrepreneurship Theory and Practice, 30(6): 803-808. Wiklund, J., Shepherd, D. 2003. Knowledge-based resources, entrepreneurial orientation, and the performance of small and medium-sized businesses. Strategic Management Journal, 24(13): 1307-1314. Woodworth, G. G. 2004. Biostatistics: a Bayesian introduction. Chicester: Wiley. Wooldridge, J. M. 2002. Econometric analysis of cross section and panel data. Cambridge: The MIT Press. World Economic Forum (2008). Globalization of alternative investments working papers volume 1 – the global economic impact of private equity report, Cologne/Geneva. Worrel, D. L., Davidson III, W. N., Sharma, V. N. 1991. Layoff announcements and stockholder wealth. Academy of Management Journal, 34(3): 662-678. Wortman, M. S. Jr. 1994. Theoretical foundations for family-owned businesses: a conceptual and research-based paradigm. Family Business Review, 7(1): 3-27.
Z
Zahra, S. A. 1996. Governance, ownership, and corporate entrepreneurship: the moderating impact of industry technological opportunities. Academy of Management Journal, 39(6): 1713-1735. Zahra, S. A., Hayton, J. C., Salvato, C. 2004. Entrepreneurship in family versus non-family firms: a resource-based analysis of the effect of corporate culture. Entrepreneurship, Theory and Practice, 28(4): 363-381. Zahra, S. A. 2005. Entrepreneurial risk taking in family firms. Family Business Review, 18(1): 2340. Zahra, S. A. 2007. An embeddedness framing of governance and opportunism: towards a crossnationally accommodating theory of agency: critique and extension. Journal of Organizational Behavior, 28(1): 69-73. Zellweger, T. 2007. Time horizon, costs of equity capital, and generic investment strategies of firms. Family Business Review, 20(1): 1-15.
Appendix List of firms in the dataset
American Express Co Autonation Inc* American Greetings Corp* Autozone Inc* ACE Ltd American International Group Inc Avaya Inc ADC Telecommunications Inc American Power Conversion Corp* Avery Dennison Corp Adobe Systems Inc* American Standard Cos Inc Avon Products Inc Advanced Micro Devices Inc* AmerisourceBergen Corp* B AES Corp* Amgen Inc Baker Hughes Inc Aetna Inc AmSouth Bancorp Ball Corp* Aflac Inc* Anadarko Petroleum Corp Bank of America Corp Agilent Technologies Inc* Analog Devices Inc* Bank of New York Co Inc Air Products & Chemicals Inc Andrew Corp Bank One Corp Alberto-Culver Co* Anheuser-Busch Cos Inc* Bausch & Lomb Inc Albertson’s Inc* AON Corp* Baxter International Inc Alcoa Inc Apache Corp* BB&T Corp Allegheny Energy Inc Apartment Investment & Management Co* Bear Stearns Cos Inc Allegheny Technologies Inc* Apollo Group* Becton Dickinson & Co* Allergan Inc* Apple Computer Inc* Bed Bath & Beyond Inc* Allied Waste Industries Inc Applera Corp - Applied Biosystems Group BellSouth Corp Allstate Corp Applied Materials Inc Bemis Co* Best Buy Co Inc* Alltel Corp Applied Micro Circuits Corp Big Lots Inc Altera Corp* Archer-Daniels-Midland Co Altria Group Inc Ashland Inc Biogen Inc* Biomet Inc* AMBAC Financial Group Inc AT&T Corp Ameren Corp Autodesk Inc BJ Services Co American Electric Power Co Inc Automatic Data Processing Inc* Black & Decker Corp
Note: * Family firm as identified by BusinessWeek (2003)
A
3M Co
CA Inc* Calpine Corp* Campbell Soup Co* Capital One Financial Corp Cardinal Health Inc* Carnival Corp* Caterpillar Inc CBS Corp Cendant Corp* Centerpoint Energy Inc Centex Corp CenturyTel Inc Charles Schwab Corp* Charter One Financial Inc* Chevron Corp
C
BMC Software Inc Boeing Co Boston Scientific Corp* Bristol-Myers Squibb Co Broadcom Corp* Brown-Forman Corp* Brunswick Corp Burlington Northern Santa Fe Corp Burlington Resources Inc
Table A-1: List of firms in the dataset (created from the S&P 500 as of as of July 31, 2003; firms are listed in alphabetical order) Chiron Corp* Chubb Corp* Ciena Corp Cigna Corp Cincinnati Financial Corp* Cinergy Corp Cintas Corp* Circuit City Stores Inc Cisco Systems Inc Citigroup Inc Citizens Communications Co Citrix Systems Inc Clear Channel Communications Inc* Clorox Co CMS Energy Corp Coca-Cola Co Coca-Cola Enterprises Inc* Colgate-Palmolive Co Comcast Corp* Comerica Inc Computer Sciences Corp Compuware Corp* Comverse Technology Inc* ConAgra Foods Inc Concord EFS Inc*
238 Appendix
F
Equifax Inc Equity Office Properties Trust* Equity Residential* Exelon Corp Exxon Mobil Corp
Family Dollar Stores Inc* Fannie Mae Federated Department Stores Inc E Eastman Chemical Co Federated Investors Inc* Eastman Kodak Co FedEx Corp* Eaton Corp Fifth Third Bancorp eBay Inc* First Data Corp* Ecolab Inc First Horizon National Corp Edison International FirstEnergy Corp EI Du Pont de Nemours & Co* Fiserv Inc* El Paso Corp FleetBoston Financial Corp Electronic Arts Inc Fluor Corp* Electronic Data Systems Corp Ford Motor Co* Eli Lilly & Co* Forest Laboratories Inc EMC Corp/Massachusetts* Fortune Brands Inc Emerson Electric Co FPL Group Inc Engelhard Corp Franklin Resources Inc* Entergy Corp Freddie Mac EOG Resources Inc Freeport-McMoRan Copper & Gold Inc*
Dollar General Corp* Dominion Resources Inc Dover Corp Dow Chemical Co Dow Jones & Co Inc* DTE Energy Co Duke Energy Corp Dynegy Inc
Note: * Family firm as identified by BusinessWeek (2003)
Dana Corp Danaher Corp* Darden Restaurants Inc Deere & Co Dell Inc* Delphi Corp Delta Air Lines Inc Deluxe Corp Devon Energy Corp* Dillard’s Inc*
D
ConocoPhillips Consolidated Edison Inc Constellation Energy Group Inc Convergys Corp Cooper Industries Ltd Cooper Tire & Rubber Co Corning Inc* Costco Wholesale Corp* Countrywide Financial Corp* CR Bard Inc Crane Co CSX Corp Cummins Inc* CVS Corp*
I
HCA Inc* Health Management Associates Inc Hercules Inc Hershey Co Hess Corp* Hewlett-Packard Co* Hilton Hotels Corp* HJ Heinz Co Home Depot Inc* Honeywell International Inc Humana Inc* Huntington Bancshares Inc/OH
Illinois Tool Works Inc* IMS Health Inc Ingersoll-Rand Co Ltd Intel Corp* International Business Machines Corp International Flavors & Fragrances Inc* H H&R Block Inc* International Game Technology Halliburton Co International Paper Co Harley-Davidson Inc Interpublic Group of Cos Inc Harrah's Entertainment Inc Intuit Inc* Hartford Fin. Services Grp Inc ITT Industries Inc Hasbro Inc*
Gannett Co Inc Gap Inc* Gateway Inc* General Dynamics Corp General Electric Co General Mills Inc General Motors Corp Genuine Parts Co Genzyme Corp* Georgia-Pacific Corp Gillette Co Golden West Financial Corp* Goldman Sachs Group Inc Goodrich Corp Goodyear Tire & Rubber Co Great Lakes Chemical Corp Guidant Corp*
G
Table A-1 (continued): List of firms in the dataset (created from the S&P 500 as of as of July 31, 2003; firms are listed in alphabetical order)
Appendix 239
Manor Care Inc* Marathon Oil Corp Marriott International Inc* Marsh & McLennan Cos Inc Marshall & Ilsley Corp Masco Corp* Mattel Inc Maxim Integrated Products Inc* May Department Stores Co Maytag Corp
Note: * Family firm as identified by BusinessWeek (2003)
KB Home Kellogg Co Kerr-McGee Corp Keycorp KeySpan Corp Kimberly-Clark Corp Kinder Morgan Inc* King Pharmaceuticals Inc* Kla-Tencor Corp* Knight Ridder Inc* Kohl's Corp Kroger Co
M
Jabil Circuit Inc* Janus Capital Group Inc JC Penney Co Inc JDS Uniphase Corp Jefferson-Pilot Corp John Hancock Financial Services Inc Johnson & Johnson Johnson Controls Inc Jones Apparel Group Inc* JPMorgan Chase & Co
K
L
Leggett & Platt Inc* Lehman Brothers Holdings Inc Lexmark International Inc Lincoln National Corp Linear Technology Corp* Liz Claiborne Inc Lockheed Martin Corp Loews Corp* Louisiana-Pacific Corp Lowe’s Cos Inc LSI Logic Corp* Ltd Brands* Lucent Technologies Inc
J
MBIA Inc MBNA Corp* McCormick & Co Inc McDermott International Inc McDonald's Corp McGraw-Hill Cos Inc* McKesson Corp MeadWestvaco Corp* MedImmune Inc* Medtronic Inc* Mellon Financial Corp* Merck & Co Inc Mercury Interactive Corp* Meredith Corp* Merrill Lynch & Co Inc Metlife Inc MGIC Investment Corp Micron Technology Inc Microsoft Corp* Millipore Corp Molex Inc* Molson Coors Brewing Co* Monsanto Co Monster Worldwide Inc* Moody's Corp
O Occidental Petroleum Corp Office Depot Inc* OfficeMax Inc Omnicom Group Inc Oracle Corp*
N Nabors Industries Ltd* National City Corp National Semiconductor Corp Navistar International Corp NCR Corp Network Appliance Inc* New Cingular Wireless Services Inc New York Times Co* Newell Rubbermaid Inc* Newmont Mining Corp Nextel Communications Inc* Nicor Inc Nike Inc* NiSource Inc Noble Corp Nordstrom Inc* Norfolk Southern Corp North Fork Bancorporation Inc Northern Trust Corp Northrop Grumman Corp Novell Inc Novellus Systems Inc
Paccar Inc* Pactiv Corp* Pall Corp Parametric Technology Corp Parker Hannifin Corp Paychex Inc* Peoples Energy Corp Peoplesoft Inc* Pepsi Bottling Group Inc PepsiCo Inc PerkinElmer Inc Pfizer Inc PG&E Corp Phelps Dodge Corp Pinnacle West Capital Corp Pitney Bowes Inc
P
Nucor Corp Nvidia Corp*
Morgan Stanley Motorola Inc*
Table A-1 (continued): List of firms in the dataset (created from the S&P 500 as of as of July 31, 2003; firms are listed in alphabetical order)
240 Appendix
Reebok International Ltd* Simon Property Group Inc* Regions Financial Corp SLM Corp Reynolds American Inc Snap-On Inc Robert Half International Inc* Solectron Corp Rockwell Automation Inc* Southern Co Rockwell Collins Inc SouthTrust Corp* Rohm & Haas Co* Southwest Airlines Co* Rowan Cos Inc Sprint Corp-PCS Group RR Donnelley & Sons Co* Sprint Nextel Corp Ryder System Inc St Jude Medical Inc St Paul Travelers Cos Inc S Sabre Holdings Corp Stanley Works Safeco Corp Staples Inc* Safeway Inc Starbucks Corp* Sanmina-SCI Corp* Starwood Hotels & Resorts Worldwide Inc* Sara Lee Corp State Street Corp Schering-Plough Corp Stryker Corp* Schlumberger Ltd* Sun Microsystems Inc* Scientific-Atlanta Inc Sungard Data Systems Inc Sealed Air Corp Sunoco Inc* Sears Roebuck and Co SunTrust Banks Inc Sempra Energy Supervalu Inc Sherwin-Williams Co Symantec Corp Siebel Systems Inc* Symbol Technologies Inc* Sigma-Aldrich Corp Synovus Financial Corp*
Note: * Family firm as identified by BusinessWeek (2003)
RadioShack Corp Raytheon Co
R
QLogic Corp Qualcomm Inc* Quest Diagnostics Inc Quintiles Transnational Corp* Qwest Communications Int. Inc*
Q
Plum Creek Timber Co Inc (REIT) PMC - Sierra Inc* PNC Financial Services Group Inc Power-One Inc PPG Industries Inc PPL Corp Praxair Inc Principal Financial Group Procter & Gamble Co Progress Energy Inc Progressive Corp* Prologis Providian Financial Corp Prudential Financial Inc Public Service Enterprise Group Inc Pulte Homes Inc*
Wachovia Corp Wal-Mart Stores Inc* Walgreen Co* Walt Disney Co* Washington Mutual Inc
W
Veritas Software Corp* Verizon Communications Inc VF Corp* Visteon Corp Vulcan Materials Co
V
Union Pacific Corp Union Planters Corp Unisys Corp United Parcel Service Inc United States Steel Corp United Technologies Corp UnitedHealth Group Inc Univision Communications Inc* Unocal Corp UnumProvident Corp* US Bancorp UST Inc
T Rowe Price Group Inc Target Corp TECO Energy Inc Tektronix Inc Tellabs Inc* Temple-Inland Inc* Tenet Healthcare Corp Teradyne Inc Texas Instruments Inc Textron Inc Thermo Electron Corp Thomas & Betts Corp Tiffany & Co Time Warner Inc TJX Cos Inc Torchmark Corp Toys R US Inc Transocean Inc Travelers Property Casualty Corp Tribune Co Tupperware Brands Corp TXU Corp Tyco International Ltd
U
Sysco Corp
T
Table A-1 (continued): List of firms in the dataset (created from the S&P 500 as of as of July 31, 2003; firms are listed in alphabetical order)
Appendix 241
Zions Bancorporation
Note: * Family firm as identified by BusinessWeek (2003)
Zimmer Holdings Inc
Z
Yahoo! Inc* Yum! Brands Inc
Y
Xcel Energy Inc Xerox Corp Xilinx Inc XL Capital Ltd
X
Waste Management Inc Waters Corp Watson Pharmaceuticals Inc* WellPoint Health Networks WellPoint Inc Wells Fargo & Co Wendy’s International Inc Weyerhaeuser Co Whirlpool Corp Williams Cos Inc* Winn-Dixie Stores Inc* WM Wrigley Jr Co* Worthington Industries* WW Grainger Inc* Wyeth
Table A-1 (continued): List of firms in the dataset (created from the S&P 500 as of as of July 31, 2003; firms are listed in alphabetical order)
242 Appendix
Appendix
243
List of variables
Table A-2: List of variables used in the empirical analyses
Variables (in alphabetic order)
Description
Assets
Sum of current assets and net property, plant, and equipment and other non-current assets (in mn $) (source: Compustat North America; data item: AT)
CAPEX/assets
Cash outflow or the funds used for additions to the company’s property, plant, and equipment (in mn $) divided by total assets (in mn $) (source: Compustat North America; data item: AT, CAPX) Cash outflow or the funds used for additions to the company’s property, plant, and equipment (in mn $) divided by total assets (in mn $) (source: Compustat North America; data item: AT, SALE)
CAPEX/sales
Cash flow
Sum of after-tax income, depreciation and after-tax R&D (in mn $) (source: Compustat North America; data items: DP, PI, TXT, XRD)
Cash flow/assets
Sum of after-tax income, depreciation and after-tax R&D (in mn $) divided by total assets (in mn $) (source: Compustat North America; data items: AT, DP, PI, TXT, XRD)
CEO age
The age of the CEO in the respective year (source: Compustat ExecuComp; data item: P_AGE_2)
CEO gender
The gender of the CEO (source: Compustat ExecuComp; data item: PGENDER)
CEO duration
Number of years the individual has served as CEO (source: Compustat ExecuComp; data item: BECAMECEO)
CEO duality
Dummy=1 if CEO is also chairman of the board of directors (source: manual collection)
Change in property, plant, and equipment (PPE)
Change in property, plant, and equipment (PPE) (that is, PPE t – PPE t-1 (in mn $) (source: Compustat North America; data item: PPENT)
Change in sales
(Sales t – sales t-1) divided by sales t-1 (source: Compustat North America; data item: SALE)
Dividend yield
Annual dividend per share divided by the company’s close price for the fiscal year. This figure is then multiplied by 100. (source: Compustat North America; data item: DVYDF)
Employees
The number of company workers as reported to shareholders. This is reported by some firms as an average number of employees and by some as the number of employees at year-end (source: Compustat North America; data item: EMP)
Family CEO
Dummy=1 if CEO is from family (source: manual collection)
244
Appendix
Table A-2 (continued): List of variables used in the empirical analyses Variables (in alphabetic order)
Description
Family firm1
Combined stock ownership of the family > 5% or member of the family is either CEO or chairman (source: manual collection)
Family firm2
Family owns more stock than all institutional investors together do (minimum: 5%) or member of the family is either CEO or chairman (source: manual collection)
Family firm3
Combined stock ownership of the family > 5% and member of the family is either CEO or chairman (source: manual collection)
Family firm4
Family owns more stock than all institutional investors together do (minimum: 5%) and member of the family is either CEO or chairman (source: manual collection)
Family firm5a
Family owns more stock than all institutional investors together do (minimum: 5%) and member of the family is either CEO or chairman and the firm is younger than 51 years (source: manual collection)
Family firm5b
Family owns more stock than all institutional investors together do (minimum of 5%) and member of the family is either CEO or chairman and the firm is older or equal than 51 years
Family management
Dummy=1 if CEO or chairman is from family (source: manual collection)
Firm age
Log (number of years since the firm was founded) (source: manual collection)
Firm size
Log (total assets) (source: Compustat North America; data item: AT)
Industry dummies
Dummy variables created from 2-digit SIC codes indicating industry membership (55 different industries) (source: Compustat ExecuComp; data item: SIC)
Leverage
Long-term debt divided by total assets (source: Compustat North America; data items: AT, DT)
Market-to-book value
Sum of market value of equity and book value of debt divided by book value of total assets (source: Compustat North America; data items: AT, DT, MKVALF)
Ownership share of family
Percentage of stock owned by family (source: manual collection)
Ownership share of financial investors
Percentage of stock owned by financial institutions (large banks, insurance companies, investment funds, etc.) (source: manual collection); financial investor is used as a synonym for institutional investor
Percentage change in workforce
(Number of employees t – number of employees t-1) divided by number of employees t-1 (source: Compustat North America; data item: EMP)
Appendix
245
Table A-2 (continued): List of variables used in the empirical analyses Variables (in alphabetic order)
Description
Percentage decrease in workforce
Inverse of variable percentage change in workforce; observations in which the workforce is increased are indicated as missing values
Personnel intensity
Number of employees divided by total assets (source: Compustat North America; data items: AT, EMP)
Property, plant, and equipment (PPE)
Cost of tangible fixed property used in the production of revenue less accumulated depreciation (in mn $) (source: Compustat North America; data item: PPENT)
R&D/assets
R&D expenditures (in mn $) divided by total assets (in mn $) (source: Compustat North America; data items: AT, XRD)
R&D/sales
R&D expenditures (in mn $) divided by total assets (in mn $) (source: Compustat North America; data items: SALE, XRD)
Risk diversified
Part of the variance in the firm’s returns that is explained by changes in the market (that is, firm beta multiplied by variance in market returns) (source: Compustat North America; data item: MKRTXM)
Risk undiversified
Part of the variance in the firm’s returns that is not explained by changes in the market (that is, variance in the firm’s returns minus diversified risk) (source: Compustat North America; data item: MKRTXM)
ROA
Return on assets, which is defined as income before extraordinary items divided by total assets (source: Compustat North America; data item: ROA)
ROE
Return on equity, which is defined as income before extraordinary items, divided by common equity as reported (source: Compustat North America; data item: ROE)
Sales
Gross sales (amount of actual billings to customers for regular sales completed during the period) reduced by cash discounts, trade discounts, and returned sales and allowances for which credit is given to customers (in mn $) (source: Compustat North America; data item: SALE)
Sales growth
3-year (5-year) least squares annual growth rate of sales (source: Compustat ExecuComp; data item: SALE3LS)
Share of base salary (in %)
The dollar value of the base salary earned by the CEO during the fiscal year divided by total pay (source: Compustat ExecuComp; data item: SALARY)
Share of annual bonus (in %)
The dollar value of a bonus (cash and non-cash) earned by the named executive officer during the fiscal year divided by total pay (source: Compustat ExecuComp; data item: BONUS)
246
Appendix
Table A-2 (continued): List of variables used in the empirical analyses Variables (in alphabetic order)
Description
Share of long-term incentive plan (in %)
This is the amount paid out to the executive under the company's long-term incentive plan. These plans measure company performance over a period of more than one year (generally three years) divided by total pay (source: Compustat ExecuComp; data item: LTIP)
Share of stock pay (in %)
The value of restricted stock granted during the year (determined as of the date of the grant) divided by total pay (source: Compustat ExecuComp; data item: RSTKGRNT)
Share of stock option pay (in %)
The aggregate value of stock options granted to the executive during the year as valued using S&P's Black Scholes methodology divided by total pay (source: Compustat ExecuComp; data item: BLK_VALU)
Total pay
The sum of the ExecuComp data items SALARY (base salary), BONUS (annual bonus), LTIP (long-term incentive plan), RSTKGRNT (restricted stock), BLK_VALU (stock options), ALLOTHPD (all other paid), and OTHANN (other annual) (in mn $)
Time dummies
Dummy variables indicating year of observation (1992-2003) (source manual collection)
Workforce decreased
Dummy=1 if workforce decreased in current period compared to the previous period (source: Compustat North America; data item: EMP)
Workforce decreased > 5% (6%, 8%, 10%)
Dummy=1 if workforce decreased in current period compared to the previous period by more than 5% (6%, 8%, 10%) (source: Compustat North America; data item: EMP)
Note: for the exact Compustat data item definitions, see http://www.ecom.unimelb.edu.au/research/databases/Compustat/dataguide/index.htm (accessed October 1st, 2008).
Appendix
MatlabTM code for Bayesian random-effects model Figure A-1: MatlabTM code (adjusted from http://www.sinc.stonybrook.edu/stu/dokim (accessed October 1st, 2008)
% Data clear all; load TOBQ_FAM1.txt; data = TOBQ_FAM1.txt; [nobs nvar] = size(data); ids = data(:, 1); year = data(:,2); ystart = min(year); y = data(:, 3); x = data(:,4:60); X = [ones(nobs,1) x]; [nobs nvar] = size(X); t = zeros(nobs ,1); at = zeros(nobs,1); pid = zeros(nobs,1); n_obs = zeros(10, 1); ids = [ids ; 0]; id = ids(1); i = 1; k = 0; while i <= nobs; j = 0; k = k + 1; while (ids(i) == id) & (i <= nobs); j = j + 1; t(i) = j; at(i) = year(i) - ystart + 1; pid(i) = k; i = i + 1; end; n_obs(k) = j; id = ids(i); end; % Prior mean_prior = 0; % insert values for prior b00 = mean_prior*ones(nvar,1); M0 = mean_prior * eye(nvar); v0 = 100; s0 = 0.01; p0 = 100; m0 = 0.01; % Gibbs sampling ndraw = 10000; nburn = 1000; betasave = zeros(ndraw - nburn, nvar); sigmae2save = zeros(ndraw - nburn, 1); sigmau2save = zeros(ndraw - nburn, 1); usave = zeros(ndraw - nburn, 10);
247
248
Appendix Figure A-1: MatlabTM code (continued) % Initial values sigmae2 = 100; sigmau2 = 100; u = normrnd(0, 10, max(pid), 1); %Random-effects estimator %u = rand(max(pid), 1)-0.5; %Fixed-effects estimator for i=1:ndraw; % Beta b0 = b00+randn(1,1); M1 = M0 + X'*X; ystar = zeros(nobs, 1); for k = 1:nobs; ystar(k) = y(k) - u(pid(k)); end; beta1 = inv(M1)*(M0*b0 + X'*ystar); vm = chol(inv(M1)); beta = sqrt(sigmae2)*vm'*randn(nvar,1) + beta1; % Sigmae2 s1 = s0 + (ystar - X*beta)'*(ystar-X*beta)+ (beta-b0)'*M0*(beta-b0); v1 = v0 + nobs; chie = chis_rnd(1,v1); taue = chie/s1; sigmae2 = 1/taue; resid(:,i) = (ystar - X*beta); % Sigmau2 p1 = u'*u + p0; m1 = m0 + 10; chiu = chis_rnd(1,m1); tauu = chiu/p1; sigmau2 = 1/tauu; % Individual effect u u = zeros( max(pid), 1); for j = 1:max(pid) VarU = sigmae2*sigmau2/(sigmae2 + (sum(pid==j)*sigmau2)); MeanU = (sigmau2* sum(y(pid==j) - X(pid==j,:)*beta))/(sigmae2 + (sum(pid==j)*sigmau2)); u(j) = normrnd(MeanU, sqrt(VarU)) ; %Random-effects model end; end; % Excluding draws that are needed for convergence if i > nburn betasave(i-nburn,:,:) = beta(:,1); sigmae2save(i-nburn,:,:) = sigmae2; sigmau2save(i-nburn,:,:) = sigmau2; usave1(i-nburn,:,:) = u(1,:); usave2(i-nburn,:,:) = u(2,:); end; end; % Computing means and variances of the estimated coefficients and the residuals mean_beta = mean(betasave); std_beta = sqrt(var(betasave)); mean_sigmae = mean(sqrt(sigmae2save)); std_sigmae = sqrt(var(sqrt(sigmae2save))); mean_sigmau = mean(sqrt(sigmau2save)); std_sigmau = sqrt(var(sqrt(sigmau2save))); mean_u1 = mean(usave1); mean_u2 = mean(usave2); mean_resid = mean(resid(:));
Appendix Figure A-1: MatlabTM code (continued) % Computing the fraction of positive draws in the posterior distribution of the coefficients for jj=1:nvar for ii=1:i-nburn if betasave(ii,jj)>0 betapositive(ii,jj)=1; else betapositive(ii,jj)=0; end; end; beta_pos_fraction(jj)=sum(betapositive(:,jj))/(ndraw-nburn); end; % Display of results in tables disp('Results:'); disp(' '); disp('Mean coefficients'); disp([mean_beta]); disp(' '); disp('Std. error coefficients'); disp([std_beta]); disp(' '); disp('Quantiles of coefficients in percent: 1,5, 25, 50, 75, 95,99'); disp('Note: Constant is not displayed!'); for ii=2:nvar disp([ prctile(betasave(:,ii),1), prctile(betasave(:,ii),5), prctile(betasave(:,ii),25), prctile(betasave(:,ii),50), prctile(betasave(:,ii),75), prctile(betasave(:,ii),95), prctile(betasave(:,ii),99)]); end; disp(' '); disp('Positive fraction of draws by coefficient'); disp('Note: Constant is not displayed!'); for jj=2:nvar disp([ beta_pos_fraction(jj)]); end; disp(' '); disp('Note: constant + individual effect gives overall individual effect'); disp('Mean of overall individual effect (average of alpha_i)'); disp([mean(u+mean_beta(1,1))]); disp(' '); disp('Mean variance'); disp([mean_sigmae]); disp(' '); disp('Std. error variance'); disp([std_sigmae]); disp(' '); disp('Mean error'); disp([mean_resid]); disp(' '); % Display of results in graphs for j=1:nvar figure(j) hist(betasave(:,j),100); end;
249
Market-to-book value ROA Family firm1 t-1 Family firm2 t-1 Family firm3 t-1 Family firm4 t-1 Family firm5a t-1 Family management t-1 Ownership share of family t-1 Ownership share of financial investor t-1 Firm age Firm size t-1 Leverage t-1 CEO duality t-1 CEO duration t-1 Share of option payment t-1 Share of stock payment t-1 Diversified risk Undiversified risk
0.16 0.21 0.21 0.18 0.12 0.15 0.23 0.11 0.03 -0.37 -0.29 -0.31 -0.18 0.06 -0.06 0.15 0.11 0.32 0.02 0.01 0.06 0.06 0.05 -0.00 0.03 0.02 0.02 -0.12 -0.02 -0.05 0.06 -0.00 -0.09 0.01 -0.13
2
0.96 0.47 0.40 0.33 0.85 0.48 -0.17 -0.36 -0.19 -0.21 -0.22 0.30 -0.10 0.01 0.08 0.22
3
0.50 0.42 0.35 0.89 0.46 -0.20 -0.35 -0.18 -0.19 -0.23 0.32 -0.10 0.00 0.07 0.21 0.85 0.69 0.56 0.64 -0.17 -0.21 -0.08 -0.16 -0.16 0.19 -0.10 -0.06 0.13 0.14
5
0.82 0.47 0.67 -0.26 -0.17 -0.02 -0.16 -0.16 0.20 -0.09 -0.06 0.14 0.11
6
0.39 0.57 -0.23 -0.26 0.02 -0.19 -0.19 0.11 -0.07 -0.04 0.15 0.18
7
0.30 -0.10 -0.39 -0.19 -0.20 -0.25 0.36 -0.11 0.02 0.12 0.24
8
9
-0.24 -0.12 -0.05 -0.09 -0.12 0.09 -0.06 -0.09 0.08 0.07
Table A-3: Correlation table 4
-0.07 -0.25 0.04 0.06 -0.06 0.00 0.06 -0.12 0.08
10
0.42 0.45 0.20 -0.10 0.16 -0.21 -0.15 -0.61
11
0.37 0.05 -0.11 0.14 -0.01 0.10 -0.38
12
0.11 -0.14 0.15 -0.15 -0.25 -0.39
13
15
16
0.20 0.05 -0.10 -0.03 -0.04 -0.25 -0.04 0.10 -0.07 -0.12 0.01 -0.10
14
0.04 0.19
17
0.15 0.44
18
Notes: Correlations with an absolute value greater than 0.04 have a p-value 0.05. The Pearson correlation coefficient is used for metric variables, the point-biserial correlation coefficient is used in case one variable is dichotomous, and Cramer’s V is used if both variables are dummy variables.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 19 20
1
0.26
19
250 Appendix
Correlation table