Cluster Genesis: Technology-Based Industrial Development

Cluster Genesis This page intentionally left blank Cluster Genesis Technology-Based Industrial Development Edited ...

Author: Pontus Braunerhjelm | Maryann P. Feldman

35 downloads 836 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

DOWNLOAD PDF

Cluster Genesis

This page intentionally left blank

Cluster Genesis Technology-Based Industrial Development

Edited by Pontus Braunerhjelm and Maryann Feldman

1

3

Great Clarendon Street, Oxford ox2 6dp Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York ß Oxford University Press 2006 The moral rights of the authors have been asserted Database right Oxford University Press (maker) First published 2006 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover and you must impose the same condition on any acquirer British Library Cataloguing in Publication Data Data available Library of Congress Cataloging in Publication Data Data available Typeset by SPI Publisher Services, Pondicherry, India Printed in Great Britain on acid-free paper by Biddles Ltd., King’s Lynn, Norfolk ISBN 0–19–920718–6 1 3 5 7 9 10 8 6 4 2

978–0–19–920718–3

Dedicated To Our Families


Preface

The common thread in this volume was that these individuals were connected in some way to the coeditors and had serendipitous discussions over a number of years about the need for greater focus on the early origins and the evolutionary trajectories of regional technology based clusters. Funding from Handelsbanken’s Research Foundation allowed this group to come together twice to form the type of temporary spatial agglomeration that proved fruitful and energetic and resulted in an exchange of ideas. It is fair to say that our thinking on the subject evolved as a result of these meetings. Often in a group it is difficult to attribute an idea to one individual or another but certainly our inquiry has been enhanced by the flow of ideas and exchange. The process of working on this book has been an intellectually exciting and rewarding endeavor. We would like to acknowledge support from the Handelsbanken’s Research Foundation and the Rotman School of Management at the University of Toronto. We would like to thank Belinda Lobo, Benny Borgman, and Christian Helgesson for their assistance with the preparation of this manuscript.

vii


Contents

List of Figures List of Tables List of Contributors

1. The Genesis of Industrial Clusters Maryann Feldman and Pontus Braunerhjelm

xi xiii xv

1

Part I. Creation Myths Revisited 2. Origins and Growth of the Hollywood Motion-Picture Industry: The First Three Decades Allen J. Scott

17

3. The Coevolution of Technologies and Institutions: Silicon Valley as the Iconic High-Technology Cluster Martin Kenney and Donald Patton

38

4. Accounting for Emergence and Novelty in Boston and Bay Area Biotechnology Jason Owen-Smith and Walter W. Powell

61

Part II. Considering the Developing Cluster Context 5. Anatomy of Cluster Development: Emergence and Convergence in the US Human Biotherapeutics, 1976–2003 Elaine Romanelli and Maryann Feldman 6. Policy-Induced Clusters: The Genesis of Biotechnology Clustering on the East Coast of China Martha Prevezer and Han Tang

87

113

ix

Contents

7. The Emergence of a European Biotechnology Cluster: The Case of Medicon Valley Pontus Braunerhjelm and Christian Helgesson

133

8. The Emergence of Ireland’s ICT Clusters: The Role of Foreign Direct Investment Frank Barry

148

9. The Emergence of Israel’s Venture Capital Industry: How Policy Can Influence High-Tech Cluster Dynamics Gil Avnimelech and Morris Teubal

172

Part III. Crafting Cluster and Economic Development Policy 10. Clusters and Clustering: Stylized Facts, Issues, and Theories Luigi Orsenigo

195

11. Mors tua, Vita mea? The Rise and Fall of Innovative Industrial Clusters Mario A. Maggioni

219

12. Local Antecedents and Trigger Events: Policy Implications of Path Dependence for Cluster Formation David A. Wolfe and Meric S. Gertler

243

13. The Role of Public Policy in Emerging Clusters Bo Carlsson

264

Bibliography Index

279 315

x

List of Figures

2.1: Locations of motion-picture production companies in Los Angeles, 1915

25

2.2: Locations of motion-picture production companies in Los Angeles, 1930

31

2.3: Average size of establishment in the motion-picture industry in California and the rest of the USA, 1921–37

33

3.1: Genealogy of Silicon Valley technologies

43

3.2: Employment in four Bay Area counties, 1959–2001

44

3.3: Establishments in Four Bay Area counties, 1959–2001

45

4.1: Boston main components ties by dyads and year, 1988–99

66

4.2: Bay Area main components ties by dyads and year, 1988–99

66

4.3: Boston and Bay Area networks: 1988, 1994, and 1999

69

6.1: The geographic distribution of biotech research organizations in China

119

6.2: The geographic distribution of biotech firms in China

120

¨ resund region 7.1: Map defining the O

135

7.2: Interlinks between firms and academia in Medicon Valley

139

7.3: Number of biotech start-ups in Medicon Valley, 1995–2002

142

8.1: Employment levels in computer hardware and software in Ireland

149

8.2: Irish GNP per head as percentage of EU-15 average, 1960–2002

151

11.1: Agglomeration costs and benefits for incumbents, with critical sizes of a cluster

222

11.2: The development of an industrial cluster (in isolation)

225

xi

List of Figures 11.3: 11.4:

The development of the old industrial cluster, stock–flows relation

225

The development of the new industrial cluster, stock–flows relation

226

11.5:

Learning effects and clusters leapfrogging

228

11.6:

The growth and depletion of an industrial cluster

230

11.7:

The growth and depletion of an industrial cluster

230

11.8:

r-type policy

231

11.9:

K-type policy

232

11.10: Long-run equilibrium: two clusters coexist

234

11.11: Long-run equilibrium: cluster j drives cluster i to extinction

234

xii

List of Tables

2.1: Employment in motion-picture production, United States, California, and New York, 1921–37

30

4.1: R&D outputs by region, 1988–99

73

4.2: Innovation data for Betaseron and Avonex, 1988–99

78

5.1: US biotherapeutics firms by types of entry and organizational origins

96

5.2: Biotherapeutics firms in MSAs and combined MSAs

97

5.3: Entrepreneur relocations across combined MSA for top 12 regions

100

5.4: Firm relocations across combined MSA for top 12 regions

101

5.5: Distribution of biotherapeutics firms in San Diego

103

5.6: Distribution of biotherapeutics firms in Los Angeles

104

5.7: Distribution of biotherapeutics firms in New York

107

6.1: Summary of science and technology and biotechnology policies

115

6.2: Returnees as percentage of top scientists in China, 2003

117

6.3: Top 20 country rankings of the most-cited 150 countries in all fields, 2003

118

6.4: Type of sector specialization of biotechnology firms: Beijing, Shanghai, and Shenzhen, 2003

120

6.5: Ownership type for firms, Beijing, Shanghai, and Shenzhen

121

6.6: Biotechnology research projects supported by the National Natural Science Fund by location in 1999

124

6.7: Comparison of universities and student numbers: Beijing, Shanghai, and Shenzhen

124

6.8: R&D staff in relation to total employees

124

Appendix 6.1: Details of interviews used in the analysis

131

xiii

List of Tables 7.1: Comparison of volume of biotechnology-related articles from three regions

141

7.2: Number of citations in biotechnology-related articles

141

7.3: Number of firms and employees in Medicon Valley, 2002

142

7.4: Medicon Valley labor force with tertiary education, 1999

143

8.1: Country shares in world computer hardware exports

155

8.2: The relative importance of computer sector employment in EU countries

155

8.3: The relative importance of computer software employment in EU countries

159

8.4: EU employment in mass market packaged software

160

9.1: Israel’s high-tech cluster: selected structural elements, 1970s–1990s

175

9.2: R&D support from OSC, million US$

178

9.3: Yozma Funds and associated VC companies

180

9.4: Venture capital raised and invested

180

9.5: Foreign partners of Yozma Fund and privatization

181

9.6: Capital raised by private equity organizations in Israel, 1991–2003

182

9.7: Start-ups, VC funding, OCS grants, exits, and closures, 1991–2003

183

9.8: ICT and software manufacturing: sales, exports, and employees

186

11.1: Simulation results

xiv

235

List of Contributors

Gil Avnimelech is Assistant Professor at the School of Management, Ben Gurion University, Beersheva, Israel. Frank Barry is Professor at the School of Economics, University College Dublin, Ireland. Pontus Braunerhjelm holds the Leif Lundblad Chair in International Business and Entrepreneurship at The Royal Institute of Technology, Stockholm, Sweden. Bo Carlsson is the E. Mandell deWindt Professor of Industrial Economics and the Director of Ph.D. Programs and Research at Weatherhead School of Management, Case Western Reserve University in Ohio, USA. Maryann Feldman holds the Jeffery S. Skoll Chair in Technical Innovation and Entrepreneurship and is Professor of Business Economics at Rotman School of Management University of Toronto, Canada. Meric S. Gertler holds the Goldring Chair in Canadian Studies at the Department of Geography and Program in Planning. He is also CoDirector of the Program on Globalization and Regional Innovation Systems at University of Toronto, Canada. Christian Helgesson has a master degree in political science from Stockholm University. He is a research project leader at the Swedish Center for Business and Policy Studies (SNS) in Stockholm. Martin Kenney is affiliated with the Department of Human and Community Development, University of California, Davis, in the USA. He is also a Senior Project Director at the Berkeley Roundtable on International Economy.

xv

List of Contributors

Mario A. Maggioni is Professor of Economics at the Department of International Economics, Institutions and Development, Faculty of Political Science, Catholic University, Milan, Italy. Luigi Orsenigo is affiliated to the University of Brescia, and Cespri at the Bocconi University, Milan, Italy. Jason Owen-Smith is Associate Professor of Sociology and Organizational Studies at the University of Michigan in Ann Arbor, USA. Donald Patton is a Research Associate at the Department of Human and Community Development, University of California, Davis, USA. Walter W. Powell is Professor of Education and affiliated Professor of Organizational Behavior, Sociology, and Communications at Stanford University, California. He is also an External faculty member at the Santa Fe Institute, California, USA. Martha Prevezer is Professor at the School of Business and Management, Queen Mary College, University of London, UK. Elaine Romanelli is Associate Professor and Director at the Entrepreneurial Studies Program McDonough School of Business, Georgetown University, USA. Allen J. Scott is Professor at the Department of Policy Studies and Department of Geography, University of California, USA. Han Tang is affiliated with Queen Mary College, positioned in Shanghai, China. Morris Teubal is Professor at Department of Economics, Hebrew University, Jerusalem, Israel. David A. Wolfe is Professor of Political Science at the Centre for International Studies, University of Toronto, Canada.

xvi

1 The Genesis of Industrial Clusters Maryann Feldman and Pontus Braunerhjelm

Every culture and civilization has creation stories or myths about how the society started and took shape. These stories help define identity and order a chaotic, complex universe. Unfortunately, an appreciation of history and context is missing in the contemporary discussion of clusters. While increasingly clusters—regional concentrations of related firms and organizations—are perceived to be the locus of economic growth there is little understanding of how successful clusters come into existence. Moreover, little guidance is provided on the role of policies that are conducive to the formation of clusters—both what policies to promote and, equally important, what policies to avoid. Many places attempt to emulate the world’s most famous industrial cluster Silicon Valley, with its rich institutional landscape of engaged and leveraged research universities, high-flying local venture capitalists, world class supporting business and legal consultants, and rich collaborative networks. Yet these attempts ignore the historic development of these institutions and the way that they have coevolved. Importantly, the ingredients associated with Silicon Valley’s success were not in place initially and Silicon Valley was not the obvious location for the computer industry. Indeed, it was the genesis of Silicon Valley—the dynamic process of creating the industry that created the concomitant location of the institutional ingredients and the social relationships which makes them effective. Our understanding of the origins of industrial clusters needs to move beyond suggestions of a list of ingredients that, once in place, result in economic development. It is as if in the current conceptualization clusters emerge full grown, like Athena from the head of Zeus, without passage through defining developmental stages. Lists of attributes of successful

1

The Genesis of Industrial Clusters

clusters tell us little about how these clusters get started and what differentiates successful clusters from places where investments yield no significant benefits for the local economy. While the literature tends to highlight success stories there are plenty of counterfactual examples in which the ingredients appear to be in place yet the local area did not become the site of vibrant innovative activity and competitive industries. Moreover, even Silicon Valley, the archetype, has failed to develop some emerging sectors, suggesting that there is a more complex, underlying process that links location to industrial success. The purpose of this book is to examine the origins and emergence of technology-based industrial clusters in order to understand the forces that promoted economic development. Our view is that while mature clusters may look similar, what really matters is the process by which clusters come into existence. In examining the creation myths of cultures, Joseph Campbell notes that despite the unique idiosyncratic details modern scholarship has found commonalities and universal themes (Campbell 1972: 8–10). It is in the spirit of identifying similarity in the early formation and genesis of industrial clusters that we undertake this edited volume. Borrowing from Nelson and Winter’s notion (1982) of recipes, it appears that unique regional recipes develop for applying knowledge, commercializing discoveries, and creating new industries. Most technology entrepreneurs operate on thin margins and their focus is on short term survival. For the most part, their actions are reactions to immediate problems. Instead of wisely considered, far-sighted solutions, entrepreneurial activity is by necessity messy, complex, and adaptive. When entrepreneurs confront new technological opportunities they fashion solutions that adapt what they have on hand from what is easily accessible. The solutions they adopt are more likely to come from local sources—either through tapping networks of people working on similar things or through serendipitous encounters. Most importantly, these regional recipes use local ingredients in creative and adaptive ways. Solutions that appeared to work are diffused, repeated, and fine-tuned, gradually evolving into accepted routines and operating procedures. These routines are adopted by institutions to define common practices. Over time, a repertoire of actions develops, orchestrated by a common vision of the industry. This encourages further experimentation and adaptation. Knowledge of what does not work, what approaches have previously been tried and led to dead ends are part of this local knowledge. Some histories of clusters stress the brilliance of one or a handful of dedicated individuals. Certainly, these are good stories but the legacy

2


of individuals like Fred Terman, Eugene Kleiner, and Robert Noyce lies in the social organization of their model of innovation and its evolution over time. Certainly having a rich supply of entrepreneurs is important, but in Silicon Valley it is the institutions that nurture firms which provides a powerful advantage over other regions. Technology firms located in Silicon Valley are capable of extremely rapid growth because there is an understanding and an appreciation of how things are done and how people may work together. Social scientists call this by various names, an ecosystem, an innovation system, a social structure of innovation, or an incubator region. A critical element is that this set of interrelated institutions formed over time in tandem with the firms that make up the industries in the region. In our view it is the genesis of Silicon Valley or any cluster that defines social relationships and creates a shared vision of the industry’s economic activity. It is this process that ultimately determines the success and sustainability of the cluster. Our conceptualization, which is a common feature throughout this volume, is that cluster formation is a sequential process with an evolutionary logic (Feldman 2001). Rather than provide a mechanical catalog of growth phases or sequences of events, our intention is to provide an appreciation of the endogenous process that leads to local agglomeration, technological change, and economic growth. The conceptualization presented here is a variant and extension of the classic product life cycle model which uses a biological metaphor to describe stages of development (Abernathy and Utterback 1978). Certainly others will come after to dissect the stages and the transition mechanisms. Indeed, we hope that they do. There is simply too much at stake to be uncritical when our focus is long-term economic growth and when significant public funds are used to promote cluster formation activities. In the examples provided here, some triggering events coupled with an entrepreneurial spark seem necessary in order for industry clusters to emerge and enter a sustainable growth trajectory. In our narrative, any agglomeration can, in principle, be accounted for in terms of some initial seeding event—what matters most is what happens next. Our approach recognizes that alternatively there may well be pre-existing social or natural conditions or resources that account for the appearance of the first seedlings of growth in any given place. Developments in the early phase condition the later phase—path dependence sets in and new business models and institutions develop together. The actual shift of a given location into a leading position may have less to do with these pre-existing conditioning influences and more to do with the increasing returns

3


and competitive advantages that are, at least in part, derived from the agglomeration’s own internal functional evolution. What distinguishes clusters that grow rapidly from the less successful ones that stagnate or fail to develop is vigorous entrepreneurial activity and the active building of institutions aided by the forces of agglomeration economies. A second cluster development phase begins when one particular location starts to pull ahead of other locations. While there may be multiple nascent agglomerations for an industry, one region may pull ahead and develop some internal logic that provides an advantage to firms located there. This turn of events again can be the result of purely random processes, or it may stem from some unique resources in the location’s developmental logic. What appears to differentiate clusters from agglomerations is the internal social dynamics. A third phase can usually be identified in which a location and industry intensifies its competitive advantages, extends and consolidates its market reach, while other locations enter a period of comparative stagnation or decay—the location simply becomes the place to be. Note that we cannot predict which of the locations identified in the first phase of this process will become the dominant location in the third phase, except to say that it will have achieved this status by reason of its superior command of localized increasing returns effects. In this manner, a historic process begins as an open window of locational opportunity. Our argument is that there is necessarily another and more fundamental set of issues to be settled before we can really understand how industrial agglomerations arise. The history of every industrial agglomeration can be told in terms of a succession of locational events, and of course, prior natural and social conditions predispose regions in some developmental directions rather than others, sometimes decisively so. There may be strong locational factors underlying the initiating events of this process, or there may be none (i.e. the initiating events are random); it may even be that the initiating events are a result of a misperception on the part of relevant decision-makers. The important analytical issue, however, is not so much how the seed of the agglomeration was planted, as to how, once its locational coordinates have somehow or other been established, the emerging local economic system is subject to a structured and selfreinforcing process of growth and development. In conceptual terms, there is a difference between planting of the seed in a cluster and the subsequent achievement of industrial competitiveness. In a policy context, clusters are born and develop on the basis of specific combinations of capabilities, incentives, and opportunities. The three

4


elements are inseparable and linked to each other in intricate ways. Competencies obviously contribute to create and define opportunities as well as afford the ability to take advantage of existing opportunities. The latter feeds back on the processes of accumulation and development of new competencies. Competencies without incentives remain unused. But incentives without sufficient capabilities are sterile and might even be destructive. More interestingly, particular sets of capabilities identify sets of appropriate incentives, which in turn and once again influence the speed and directions of the processes of accumulation of competencies. Understanding how different types of opportunities, incentives, and capabilities match with each other would require sophisticated taxonomical exercises and hard theorizing.

Organization of The Book This book is organized into three parts. Part I presents interpretative histories of three well-known specific clusters. Part II focuses on emerging clusters to illustrate the divergent processes, the different stages of development, and the different institutional contexts from which clusters have sprung. Part III considers the implications for economic development policy.

Creation Myths Revisited While standard accounts credit the emergence of Hollywood on the basis of its physical geographic characteristics, Allen J. Scott proposes an alternative view in Chapter 2 which suggests that it was the successful implementation of a new business model that anchored the industry in Southern California. Three decisive factors are presented that allowed Hollywood to emerge as a leading movie production hub, even though the main creative and commercial activities of the motion-picture industry remained in New York. First, even though there were many other conceivable locations, Southern California was a suitable location for film production in the winter months due to the climate and the scenic landscape. Second, an unfavorable institutional set-up—particularly for independent motion producers—prompted a growing accumulation of movie production facilities in Hollywood. That led to various experiments, entry and exit, and created a sufficient agglomeration of activity (‘critical mass’) for take off. The third reason was the invention of a new

5


business model. Thomas Ince, who arrived in 1911 in Southern California, established a new studio and reorganized the production process. In Chapter 3, Martin Kenney and Donald Patton consider the iconic cluster: Silicon Valley. The emergence of Silicon Valley shares some features with the motion-picture industry. For instance, both places experienced an inflow of entrepreneurs from other parts of the country and also saw a novel business model emerge. The environment that existed in the late 1950s, later known as Silicon Valley, was not unique. Indeed, similar conditions existed in Boston and New York, for example. Both places were also influenced by policies—institutions in the motion-picture case and governmental spending on defense in the Silicon Valley case—but neither was specifically targeted to support these industries or places. A few individuals also seem to have been instrumental in the evolution of new clusters such as Fred Terman in Silicon Valley and motion picture moguls Cecil B. de Mille and Thomas Ince in Hollywood. The partially random nature of the process is evidenced by the fact that William Shockley, one of the coinventors of the transistor at Bell Labs in New Jersey and the founder of the first semiconductor firm, wanted to be near his elderly and sick mother (Moore and Davis 2001). In Chapter 4, Jason Owen-Smith and Woody Powell contrast the development of the biotech industry in Boston and the San Francisco Bay Area, the two most prominent US clusters. Their analysis rests on three presumptions: high-tech clusters in particular require both the presence of networks and spatial density, interorganizational networks serve the dual purposes of being locus of innovation and the underlying support structure that host the institutional and social context necessary for innovation, and, finally, the form and substance of innovation in successful clusters vary over time and with patterns of emergence. Even though differences in the innovation process in the two clusters can be traced to their origins, a process of convergence has been going on over the last twelve years, where the previous strong dependence on public research organizations (PROs) and venture capital (VC) has been replaced by strong firm alliances. Consequently, even though path dependence and variety in culture and norms determine the trajectory, clusters may still be quite comparable in terms of performance.

Considering the Development Clustering Context To further enhance the insights regarding the forces prompting cluster emergence, Part II provides an empirical comparison of five highly

6


innovative clusters in high-technology industries. As evidenced from these analyses, there are many forces that may spark clusters—serendipity does impact how and where clusters emerge. Still, irrespective of the differences in institutions, traditions, and level of development among the clusters analyzed, there are also some apparent similarities that can be identified. One prominent feature of cluster formation seems to be the importance of entrepreneurship as an endogenous process, and the emergence and sustainability of clusters seem to critically interact with entrepreneurial activities. Elaine Romanelli and Maryann Feldman pursue this angle of cluster emergence in Chapter 5 by examining the spatial and temporal dimension of a variety of forms of entrepreneurship across cities in the USA in the human biotherapeutics industry. The first finding is that clusters grow predominantly through the investments of local entrepreneurs, local firms, and local venture capitalists. Second, for three of the regions with the largest clusters—San Diego, Boston, and San Francisco— the critical spur to growth appears to be a tendency of entrepreneurs to leave local, established firms to found their own ventures. Moreover, only those regions that exhibited this secondary, or second-generation, growth, developed grew to substantial sizes relative to other clusters. Cluster locations and growth are, however, not solely a consequence of local investments. More than 32 percent of entrepreneurs relocated from one metropolitan region to another to found new firms. Hence, the attraction of entrepreneurs and firms to a region is a tertiary influence on growth, occurring late in the history of the industry and the clusters. That was also a striking feature of the development of both the Hollywood and the Silicon Valley clusters. This is followed by Chapter 6, in which Martha Prevezer and Han Tang analyze the genesis of three major biotechnology clusters along the east coast of China. Biotechnology is a young but targeted activity for the Chinese government, with the ambition to develop China into a globally leading nation in this area. The evolution of the leading three Chinese clusters are compared and related to the preceding conditions in the respective region to host a biotechnological cluster. Two main phases are distinguished in the evolution of government policy with respect to these clusters. A preceding phase mainly concerned with institutional reform, creating new forms of property rights, setting out strategic programmes for the development of the biotechnology and building up the appropriate knowledge base. Partly as a result of these policies, the entry of companies into the biotech clusters in each region more than doubled during the

7


period 1995–2003 as compared to the period before 1995. While most of the entries in the pre-1995 period were government-initiated, many new companies formed in the latter period were founded by scientists returning from abroad. The previous history of each region has influenced the structure and the specialization of the respective cluster. Pontus Braunerhjelm and Christian Helgesson examine the forces that sparked the emergence of a leading European biotechnology cluster— Medicon Valley—in Chapter 7. Medicon Valley displays one characteristic feature which separates it from most other biotechnology clusters: its geographic location stretches over two countries, that is, the southern part of Sweden and the northern part of Denmark. This makes Medicon Valley an especially interesting case since there is a possibility to compare the performance and dynamics of the cluster in a region where institutional differences do appear, simultaneously as the two countries share a number of other characteristics. To some extent that resembles the comparison of the Boston and the Bay areas in the biotech study, and the institutional differences which influenced the location of the film industry. The emergence of the Medicon Valley biotechnology cluster seems to have followed a somewhat different path than the biotech clusters in the USA. Whereas entrepreneurs orchestrated the emergence of clusters in the USA, universities played a crucial role in the Medicon Valley case and entrepreneurs entered—and reinforced—the process in a later stage. The more dynamic and fast growing biotechnology sector in Denmark is argued to be partly policy driven, partly associated with cultural differences. In Chapter 8, Frank Barry presents an analysis of the Irish cluster, proposing that a decisive factor in the emergence of the Irish ICT cluster was the presence of foreign anchor firms, such as Intel. These firms also acted as incubators for domestic entrepreneurs. Half of the Irish manufacturing labor force is now employed in foreign-owned firms, with the bulk in information and communication technology (ICT) firms. The ICT cluster (both hardware and software) are agglomerated around Dublin. Already in the mid-1990s Ireland had become host to the most important ICT cluster in Europe. Equally transformative was institutional reform and fiscal policy that also paved the way for future tax reductions, a newly developed ‘social partnership model’ of wage determination and Ireland’s EU membership in 1973 which guaranteed access to the European markets. Of the factors typically cited as important for development and genesis of innovative clusters is the availability of venture capital. In Chapter 9, Gil Avnimelech and Morris Teubal document the emergence of the

8


Israeli VC industry. The process is divided into three phases that spread out over three decades, which also mirrors Israel’s innovation and technology policy. There was a deliberate policy that targeted VC—besides the importance of bringing in foreign expertise at an early stage—once the existence of a sufficient knowledge base had been established. A truly innovative part in Israel’s policy was the inclusion of a mechanism from the very start for the government to withdraw and privatize the VC funds.

Crafting Clusters and Economic Policies To design economic policy prescriptions, a better understanding of the mutual impact and interdependencies underlying cluster formation, entrepreneurship, and the role of government is required. This section contains four chapters, all preoccupied with the development of capabilities to foster the emergence of clusters and the ensuing policy implications. In Chapter 10, Luigi Orsenigo presents a survey of factors previously associated with the crafting of new innovative high-technology clusters, emphasizing the biotechnology industry. There is perhaps no other industry that has been as carefully studied with regard to geographic clustering. The evidence strongly supports the view that there are strong agglomeration forces in high-technology sectors, mainly related to the concentration of scientific knowledge. Adequate incentive structures and entrepreneurial activity are also important, although their nature and their effects may differ substantially across space and time. Still, the evidence seems to support a picture whereby the spatial concentration of innovative activities derives mainly from processes of spin-off from highly capable universities and research centres. A conclusion drawn is that in the case of biotechnology, the emergence of clusters is not a purely random phenomenon. Initial conditions and ‘endowments’ play a crucial role in defining the geography of innovation. However, these do not suffice to account either for the genesis of clusters, neither for the failures. The literature forcefully points to the observation that—as much as agglomeration forces are influenced by ‘structural’ initial conditions— processes are the essence of what clusters are made of. Mario Maggioni develops a formal mathematical model to analyze the rise and fall of clusters in Chapter 11. It is shown how a major technological innovation sets a process of creative destruction into motion, where new clusters appear and replace clusters based on obsolete technologies. In the last stage when the cluster matures, it either achieves

9


a national or international leadership in a given sector or technology. Conversely, the cluster may decline both economically and socially, generating migration outflows, mass unemployment and regional stagnation. The decisive element appears to be different institutional frameworks. Technological and regional dynamics go hand in hand and are mutually determined in a complex web of circular cumulative causation with both positive and negative feedbacks, where institutions play a major role. In a simulation exercise the author stresses the type of policy necessary at different stages of cluster development. Basically, to promote endogenous cluster growth after the initial phases, policies that support firm-based micro-level incentives seem to be critical rather than policies aimed at strengthening the ‘carrying capacities’. Most European policymakers overemphasize the latter type of policies as a means to initiate cluster emergence and growth. In Chapter 12, David Wolfe and Meric Gertler examine the degree of path dependency in cluster origin and development. One issue that continues to bedevil the analysis of clusters is the question of their origins and the relative importance of chance events, or serendipity, as opposed to planned policy actions. Particularly controversial is the role of path dependencies created by small, initial—often chance—events as opposed to the role of conscious direct policy initiatives. The presence, or absence, of key institutional elements of the local or regional innovation system may affect both their innovative capacity and their potential to serve as nodes for cluster development. Much of the literature on path dependency, and a certain stream of analysis in the literature on clusters, suggests that they are frequently seeded by chance events or by policy initiatives that frequently had a different intention than the support and development of a cluster. The authors present a more nuanced view where more weight is allocated to the role of policies in generating and developing clusters. In the final chapter, ‘Public Policies in Emerging Clusters’, Bo Carlsson addresses the question of how to design a public policy in a nondeterministic, evolutionary, and highly complex world. That is, a world where the most desirable outcomes are unknown but there may be many possible acceptable outcomes, and where change is characterized by both path dependence and unpredictability. Based on the case studies previously presented, as well as former carefully examined cases in the literature, the cornerstone of a viable policy—defined in terms of functions and requirements of policy instruments—is presented. One conclusion is that policy interventions exaggerate the system characteristics in

10


cluster policies at the expense of incentives on the level of individuals, potential entrepreneurs, and firms. Without a proper balance between institutional design and incentive design the probability of a successful emergence and evolution of cluster will diminish. Particularly in Europe, with its tradition of governmentally administrated and targeted policies, there is a risk of downplaying the role of individual initiatives in the creation of clusters. While there are many attempts to develop cluster, these may compete away all the eventual welfare effects that result from clusters.

Synthesizing the Results Cluster formation is a process that relies on the coevolution of technology, business models, and local supporting institutions. Serendipity in cluster emergence is a conspicuous feature in the cases presented here, and has also been confirmed in the previous literature. Seeding events of clusters may either stick and a cluster forms, or slip away and the nascent cluster fails to develop. Louis Pasteur said that fortune favors the prepared mind and cluster formation appears to favor the prepared region. Path dependence and resource accumulation are part, but only part of the story. Within successful clusters there is a degree of self-organization that, while seemingly unique, reflects an underlying complex social process. If policies fail to understand the dynamics of clusters emergence, the risk that a nascent cluster will decline may be evident, particularly since high-technology clusters are involved in a fierce competition for talent, finance, and firms. Some characteristics seem to cut through most of the case studies and also appear in the policy chapters. We like to list and briefly comment upon these. First, a prominent feature of cluster formation is the importance of entrepreneurship. The emergence and sustainability of clusters seem to critically interact with entrepreneurial activities. All the case studies— from the cluster ‘icons’ in Silicon Valley and Hollywood, to the hightechnology cluster cases studies in the USA, Europe, and Israel—point at the importance of entrepreneurs. Entrepreneurs are continuously experimenting with regard to products, markets, and business models. These experiments within regions, as well as nations, augment the knowledge base within the clusters. In China, and to some extent in Ireland and Sweden, the cluster emergence has partly been prompted by other forces and the role of entrepreneurs is not that clear-cut. However, it was also

11


shown that dynamics and growth were lower in less entrepreneurial Sweden, and to some extent in the less entrepreneurial clusters in China, simultaneously as entrepreneurial activity has increased in Ireland. In the case of China, it is, however, too early to speculate about the future viability of their biotechnological clusters. A second observation that seems robust is that entrepreneurs with business experience stand the highest chances of survival and growth. Spin-offs from old firms, or individuals with previous entrepreneurial experience, become crucial assets in the development of clusters. Note also that in all cases, possibly with the exception of the later stages in the Israeli experience, VC firms and support organizations appear subsequent to an influx of entrepreneurs. Third, these observations have immediate policy interpretations. Policies are by no means redundant in the case of cluster emergence, despite a high element of serendipity. Basically, in the early stages policy is decisive in creating and maintaining a knowledge base, which is also illustrated in the case studies. Foremost, the challenge for policymakers is to stimulate entrepreneurship, an area where Europe is lagging behind the USA in particular, but many Asian states as well. That comes out as an unusually robust result and contrasts with the conventional wisdom where focus is on building ‘carrying’ capacities or targeting certain areas, for example the objective within the EU to raise R&D spending to 3 percent of gross domestic product (GDP). In particular, focus has to be turned to the diffusion and exploitation of knowledge through entrepreneurs identifying economic opportunities, rather than static accumulation of knowledge resources. That means looking at incentives and risk–reward ratios to engage the pool of potential entrepreneurs within a region or nation. Finally, and perhaps most important from a policy perspective, is the ability to adopt flexible and coevolving policies designed to foster cluster emergence, as shown in the cases of China, Ireland, Israel, but also in the comparison between Denmark and Sweden. Cluster formation is thus not a deterministic process. The emergence of clusters can be seen as an evolutionary process where the initial seed may be fostered by what look like idiosyncratic forces, but which reflect the development of institutions, technologies, and firms in a dynamic and self-organizing process. An inferior capability to define and adjust policies in a way that reflects the needs of an emerging cluster is probably the most efficient way to halt a process of cluster emergence. By that we do not advocate subsidies in various forms. Rather, the critical ingredient is to find a proper balance between carrying capacities, the institutional set-up, and the individual incentive design.

12


Certainly, there are no guarantees in economic development policy; however, given that economic growth and prosperity are at stake, understanding how clusters begin and transform local economies is an important policy concern that should be informed by analysis and actively debated and evaluated.

13


Part I Creation Myths Revisited


2 Origins and Growth of the Hollywood Motion-Picture Industry: The First Three Decades Allen J. Scott

This chapter provides an analytical account of the origins and early development of the motion-picture industry in Hollywood and shows how it then evolved in the period from about 1915 to 1930 into a dense agglomerated complex of production activities. This task entails two broad research thrusts. On the one side, it calls in general for the identification of a conceptual lens through which issues of the genesis and growth of localized industrial systems can be viewed in some theoretically coherent manner. On the other side, it also revolves around effective marshalling of the empirical data on the emergence of the industry in this corner of Southern California so as to achieve an interpretative reconstruction above and beyond the simple descriptive ordering of empirical outcomes in time and space. Needless to say, the task is fraught with much difficulty. For one thing, the conceptual framework offered is at best provisional and speculative, and must be taken as a work in progress rather than a set of finished affirmations. For another, relevant information on the business activities of early film-production companies is hard to find, and most especially, the kind of information that sheds light on the specifics of the industry as a productive agglomeration. Inevitably, then, there are troubling gaps in the empirical record and the explanatory reading that follows falls occasionally onto fragile ground. With these provisos in mind, this chapter attempts to show how Hollywood evolved over the first three decades of the twentieth century from a loose and rather chaotic collection of motion-picture shooting activities

17

Origins of Hollywood Motion-Picture Industry

into a dense interlocking system of production companies, anchored in geographic space by its own virtuous circle of endogenous development. Out of its initial stirrings from about 1907 to 1915, Hollywood rapidly established itself as the preeminent center of motion-picture production not only in the USA but also in the world. By the late 1920s, the classic studio system of production was firmly established, and Hollywood was now poised at the threshold of a golden age that would continue through World War II until the late 1940s (Schatz 1988). This remarkable series of developments has been subject to investigation time and again from the perspective of both the business analyst and the film historian. In spite of a wealth of literature there remains an important set of unresolved questions about just why Hollywood emerged as the core location of the American film industry in the first instance, and how it then established its early supremacy over a number of other equally likely or unlikely production locales. At the heart of these questions, there lies the further issue of how it was that Hollywood came to be constituted as a peculiar kind of industrial district generating streams of localized competitive advantages that have sustained it as the world’s premier center of motionpicture production for close to a century.

Why Hollywood? Observations on the Early Motion-Picture Industry The very earliest stirrings of the motion-picture industry in Europe and North America at the end of the nineteenth century have been documented at length elsewhere (e.g. Bordwell, Staiger, and Thompson 1985; Jacobs 1939; Musser 1994). Most of the details need not detain us here. At the turn of the century production was concentrated in the New York– New Jersey metropolitan area, the two leading firms being Edison Manufacturing Company and the American Mutoscope and Biograph Company. Films at this time comprised diverse short subjects projected in store-front theaters or nickelodeons. In the first decade of the twentieth century, audiences grew rapidly, and the number of production companies expanded greatly in the New York–New Jersey area as well as in other selected parts of northeast. In these early days of the industry, open-air stages lit naturally by the sun were widely utilized for photography, though indoor shooting of films using artificial light was practiced. It was also common for production companies to shoot films in urban or natural settings, and filmmakers were constantly on the search for suitable

18


locations. The cold, harsh winters of the northeast presented obvious problems to early filmmakers in this respect, and so every year, teams of camera operators and actors would fan out from the main studios across the southern and western states, and even further afield, in order to enlarge their sphere of filming opportunities. One location that attracted filmmakers from the northeast early on was Southern California; and without any question, the climatic and other physical attributes of the region made it a favored venue at this time. Certainly, in the period from about 1907 to 1915, knowledgeable people in and around the industry were increasingly ready to proclaim the virtues of Southern California as a location for the motion-picture industry, not only because of the region’s warm sunny climate and mild winters but also because of the diversity of landscapes that it offered (cf. Spencer 1911). This reasoning, indeed, has become enshrined in a conventionalized explanation of the development of the motion-picture industry in Hollywood that is now more or less repeated without much further elaboration from one study to another (see e.g. Jacobs 1939; Palmer 1937; Ramsaye 1926; Sklar 1975; Torrence 1982; Zierer 1947). Some analysts have thrown into the ring a further argument related to the formation of the New York– based Motion Picture Patents Company and the strong desire on the part of independent motion-picture producers to put as much distance as possible between themselves and the Company given its aggressive enforcement of the patent rights that it held on critical cinematographic equipment (Hampton 1931; Jacobs 1939). Others propose a much rehearsed but surely improbable gloss to the effect that a further factor in the choice of Los Angeles as a shooting locale was that it was close to the Mexican border, and hence producers who were violating the Company’s patent rights could easily make their escape whenever the authorities appeared on the scene. Sklar (1975) has added that the open-shop labor market arrangements of Los Angeles in the period up to about 1935 were an important attraction to motion-picture producers from the northeast of the country. By contrast, Hampton (1931) claims—not unreasonably— that the puritanical atmosphere of Los Angeles in the early years of the twentieth century should have given San Francisco a clear locational advantage over its southern neighbor as a center of motion-picture production. To what extent, we may ask, do explanatory sketches of these sorts really account for the emergence of the motion-picture industry in Hollywood? Do the factors typically advanced constitute a set of necessary and sufficient conditions, or are they merely contingencies in a much more

19


complex set of historic processes that might just as easily have led to altogether different outcomes? Of course, we must always take seriously the reasons that people give for their own decisions. Many representatives of the motion-picture industry at the beginning of the twentieth century were unquestionably persuaded as to the superiority of Southern California as a production locale, and clearly acted on their convictions. Does this truth about the matter provide us with an adequate basis for explanation, as certain idealist historians and geographers might suggest? Alternatively, is the genesis of the motion-picture industry in Hollywood merely a matter of historic and geographic accident? Before any effort to address questions of these sorts directly, this chapter makes a brief excursion into the broad issue of the formation and logic of industrial districts or regional productive agglomerations in general. Armed with whatever insights this exercise might offer us we can then come back to the specific problem of Hollywood and the American motion-picture industry at the beginning of the twentieth century.

The Dynamics of Locational Agglomeration Most statements about the origins of industrial agglomerations rely ultimately on one or other of two approaches that in effect circumvent a genuinely explanatory result. The first is the principle of post hoc propter hoc in which firms are described as following one another in a locational sequence leading formally to an eventual accumulation or cluster of producers in a given place. In this sort of narrative, any agglomeration can in principle be accounted for in terms of an initial seed whose planting constitutes the critical moment out of which all else flows. The second can be characterized as the locational factors approach. In this case, various empirical features of a locale are measured, and a statement then constructed to the effect that in view of these factors there is a certain likelihood for a given kind of industrial agglomeration to occur there (Scott and Storper 1987). Of course, the history of every industrial agglomeration can be told in terms of a succession of locational events; and of course, prior natural and social conditions predispose regions in some developmental directions rather than others, sometimes decisively so. My argument here is that there is necessarily another and more fundamental set of issues to be settled before we can really understand industrial agglomerations in genetic terms. This argument turns on the intertwined problems of (a) when and how a simple accumulation of production units at any given

20


place begins to manifest signs of an endogenous development dynamic, and (b) how this place then pulls ahead of actual or latent competitors, and how it then sometimes becomes by far the most advanced production center for its type of output. There may be strong locational factors underlying the initiating events of this process; or there may be none (i.e. the initiating events are random); it may even be that the initiating events are a result of a misperception on the part of relevant decision-makers. The important analytical issue, however, is not so much how the seed of the agglomeration was planted, as to how, once its locational coordinates have somehow or other been established, the emerging local economic system is subject to a structured and self-reinforcing process of growth and development. In conceptual terms, we must distinguish between the planting of the seed and blooming of the organism, that is the first occurrence of a locational signal from the achievement of competitive supremacy. Despite its rather abstract logic, this account of the formation of industrial agglomerations offers us a number of important clues about the early historic geography of the motion-picture industry in the USA. The problem we now face is how to exploit this account in order to identify critical turning points in the industry’s development while at the same time respecting the full complexity of events as they actually occurred on the ground. In order to achieve this difficult synthesis we first of all need to review the main empirical record.

The Early Motion-Picture Industry in the United States: From New York to Los Angeles As we have seen, the most important motion-picture production companies in the USA in the first decade of the twentieth century were located in the New York–New Jersey metropolitan area, with subsidiary centers in Chicago (primarily, Essanay Studios and the Selig Polyscope Company) and Philadelphia (the Lubin Film Company). In 1908, a core group of these firms, under the leadership of the Edison Manufacturing Company and the American Mutoscope and Biograph Company, established a cartel officially designated the Motion Picture Patents Company, but more popularly known as the Trust (Allen 1976; Balio 1976a; Bowser 1990; Brownlow 1979; Cassady 1982; Izod 1988). The other production companies involved in this action were Essanay Studios, Kalem Company, Lubin Film Company, Selig Polyscope Company, and Vitagraph Company, together with three distributors, Kleine Optical

21


Company, Me´lie`s, and Pathe´ Fre`res. The Trust functioned as a holding company for the patents owned by its members, and this effectively gave it monopoly control over the then most efficient equipment for both shooting and projecting films. The Trust was accordingly able, through its licensing operations, to exert massive control over the production, distribution, and exhibition of motion pictures in the USA. Its level of control was augmented in 1910 when it established the General Film Company to distribute its members’ products throughout the USA. General Film was notorious for the sanctions it exerted over its licensed exhibitors by threatening to cut off their supply of films should they attempt to deal with independent companies. For a time, the Trust even managed to establish itself as a monopsony for Eastman Kodak film. These arrangements enabled the Motion Picture Patents Company to earn unusually high revenues while at the same time imposing shackles on the business activities of independent competitors. One outcome of this state of affairs was that members of the Trust were encouraged to make the ultimately self-defeating calculation to the effect that their competitive advantage lay more in cultivating their monopoly powers than in improving the quality and appeal of their films. This calculation not only opened the Trust to the scrutiny of regulators in the Department of Justice but also induced its independent competitors to pursue what turned out in the end to be a superior competitive model (see below). In 1915 the Trust ceased to be a major force, and in 1918 it was disbanded entirely. Meanwhile, members of the Trust and independent companies alike were dispatching teams every winter to carry out location shooting in the southern and western parts of the country where the weather was more propitious for filming in this season. For a time, Jacksonville, Florida, was a favored location, and in the early teens Kalem, Lubin, Selig, Thanhouser, and Vim Comedy, among others, were active in the area (Ponti 1992). Southern California was initially chosen as a shooting locale when the Selig Polyscope Company came to Los Angeles in the winter of 1907– 08 to film The Count of Monte Cristo. This was the first of many visits over the next few years by production companies based in the northeast in search of temporary winter havens for filming activities. The first permanent studio in Southern California was constructed by Selig in 1909 in the community of Edendale (now Glendale) just to the east of Hollywood. A number of other companies also built studios in the region; these include the New York Motion Picture Company in 1909, Biograph in 1910, Nestor, Vitagraph, and the Independent Motion Picture Company in 1911, and Mack Sennett’s Keystone Studio and Lubin in 1912 (Bowser 1990; Clarke

22


1976; Florey 1923, 1948; Jessen 1915; Slide 1994). In view of the fact that some of these firms were members of the Motion Picture Patents Company, it makes little sense to claim that they came to California in search of refuge from the Company’s agents. By 1912, according to Clarke (1976), there were seventeen production companies at work in Los Angeles. However, most of these companies were headquartered in the northeast, and the biggest studios by far were all still located in the New York area, which remained at this time the main center of motion-picture production in the USA. It is one of the ironies of modern Hollywood, given the recurrent complaints that have been made over the last couple of decades about ‘runaway production’, that its own earliest stirrings came about as a result of a process of locational decentralization.

Hollywood Emergent An alert observer scanning the motion-picture production industry in the USA in the year 1912 or 1913 would have noted the continued dominance of New York, the minor recent clustering of decentralized production units in Jacksonville and Los Angeles, and a scattering of studios in many other parts of the country. In California, apart from Los Angeles, the American Film Company was flourishing in Santa Barbara, an offshoot of Essanay was making Broncho Billy films in Niles in the Bay Area, and the Balboa Studios were just getting off the ground in Long Beach. On the basis of this information, and supposing that premonitions about the theory of locational agglomeration were already in the air, our fictitious observer would no doubt hazard a best guess to the effect that the main creative and commercial center of the motion-picture industry would probably continue to be New York (which, furthermore, contained the country’s largest concentration of writers, actors, producers, scene decorators, stage hands, and so on), and that other parts of the country would at best function as subsidiary satellite locations. Alternatively, even if our observer invested a high degree of faith in the virtues of mild sunny climates as a decisive factor in the success of the motion-picture industry, he or she would surely be as likely to suggest Jacksonville, Palm Beach, Niles, Santa Barbara, San Diego, and a dozen other similar places as being just as attractive, if not more so, as Hollywood. The initial accumulation of motion-picture production companies in Los Angeles up to about 1912, then, can perhaps best be thought of as a rather arbitrary outcome, one that could as easily have occurred at a great many other locations. I would add the speculation that the same outcome

23


was no doubt encouraged by self-reinforcing gossip about the merits of the region for camera work as itinerant film crews reported back to their peers in the northeast. There was certainly no lack of promotional propaganda on this matter at the time. Whatever the case may be, the important point to be stressed here is that there is little or no evidence to suggest that the motion-picture industry in the region up to this stage was anything but a distant satellite of New York. Not until the turbulent years from about 1912 to 1915 does the industry in Southern California really begin to show signs that a process of internal transformation and developmental change was occurring. This is a period marked by considerable business effervescence as many new production facilities were established and as others disappeared through bankruptcy or merger. More than anything, Hollywood was now starting to function as an incipient agglomeration with its own distinctive production system and labor market characteristics, and with innovative capacities (in terms of both commercial practices and film content) that seemed to set it strongly apart from the more established firms of the northeast. By 1915, as Figure 2.1 shows, Hollywood and its surrounding area was beginning to assume the geographic form of a prototypical industrial district. This trend was underpinned by the arrival of a number of critical figures from New York, each of whom helped to establish a dynamic Hollywood cinema as such. One of these figures was Cecil B. de Mille who came to Los Angeles in 1913 on behalf of the Lasky Feature Play Company, and proceeded to film The Squaw Man in a barn at the intersection of Selma and Vine Steets. The Squaw Man became the first Hollywood film to enjoy major international success. A year later, Zukor’s Famous Players Company also established production facilities in Hollywood and then in 1916 merged with the Lasky Company. The resulting Famous Players-Lasky Corporation was the forerunner of Paramount, Hollywood’s first major. In addition, in 1915, the Fox Film Corporation set up operations in Hollywood, and Carl Laemmle launched Universal Pictures which he established on a large lot at Universal City in the San Fernando Valley. Lasky, Zukor, Fox, and Laemmle were among the first Hollywood moguls. All four were Jewish immigrant entrepreneurs with modest backgrounds in sales and showmanship, in contradistinction to the more patrician figures (with the exception of Lubin) who were associated with the Motion Picture Patents Company. While the business practices of the members of the Trust were centrally focused on equipment licensing revenues, the new production companies of Hollywood were much more concerned with film content and audience appeal (Jones 2001). One index of the difference between the two groups is

24

Origins of Hollywood Motion-Picture Industry one mile one km

New York Motion Picture Co.

B U R B A N K

MALIBU SANTA MONICA

ay od W ywo Holl

STUDIO CITY Ventura

MAP AREA

PASADENA

LOS AN GELES C OU N T Y LONG BEACH ORANGE COUNTY

Universal Pictures GLENDALE

S H I L L

nd

at

e

Keystone Studio le

Vine Gower

ga en hu Ca

O D W O L Y L H O

G

Mutual Film Corp. Kalem Co. Vitagraph Co. Fine Arts Studio Essanay Studio Fox Lasky Co. Film Lubin Film Co. Masterpiece Corp Majestic and Film Co. Reliance Studios

Christie Film Co. Hollywood

ica

Highland

on

M

La Cienega

a Bre

E & R Jungle Film

Bosworth Studio Su

Wilshire

ns

et

Selig Polyscope Co.

LOS ANGELES CIVIC CTR

Pico

La

Hal Roach Studio CULVER CITY Triangle Studios

Famous Players

Washington Vermont

ta

n Sa

to

Melrose

HILLS

So

t

nse

Su

BEVERLY

Figure 2.1. Locations of motion-picture production companies in Los Angeles, 1915 Source: Addresses were obtained from a great diversity of publications, directories, and web pages. The information shown is probably not complete.

that members of the Trust tended to resist the production of feature films in favor of less demanding short films, whereas the leading independents were much more aggressive in developing feature films—though we should probably not try to press the point of contrast too far.1 Another significant difference is that the Trust was much more opposed than the independents to the emerging star system in the motion-picture industry, and many of its members refused to divulge personal information about or even the names of their main performers. Yet as early as the mid-teens, stars were becoming an important device for branding films and for helping to stabilize markets. At the same time, firms belonging to the Trust rented out their films by the foot irrespective of content, a practice that scarcely gave them much incentive to raise product quality. The Hollywood independents, by contrast, started at an early stage to concentrate

25


on feature films and to promote individual stars, while simultaneously building strong narrative dynamics into the films that they made (Allen 1976; Hampton 1931). One of the decisive moments of this trend was the production of Tess of the Storm Country by the Famous Players Film Company in Southern California in 1914, a film that propelled Mary Pickford into superstar status. The vibrancy of the nascent Hollywood cluster was further fortified when Charles Chaplin went to work for Mack Sennett’s Keystone Studio in 1913. To be sure, Chaplin moved on in 1914 to the Essanay studio at Niles in the Bay Area, only to return to Hollywood in 1916 with the Mutual Film Corporation, and then to set up his own studio (at La Brea and Hollywood Boulevards) in 1918. Other major stars who came to prominence at this time were Douglas Fairbanks, William S. Hart, and Roscoe Arbuckle, representing, respectively, the swashbuckling adventure films, westerns, and comedies that were now (among other genre films) pouring out of Hollywood. Above all, the figures of Thomas Ince and D. W. Griffith tower over this historic moment, with an influence on both the business and aesthetic practices of Hollywood production companies that alone was probably sufficient to push them to the leading edge of the industry. Ince can be seen in a sense as the harbinger of the full-blown studio system. He arrived in Southern California in 1911 to produce cowboy films for the New York Motion Picture Company, establishing a studio known familiarly as Inceville in Santa Monica at the point where Sunset Boulevard meets the Pacific Ocean. Ince was the first producer to attempt to industrialize the whole filmmaking process and to push it beyond the rather simple set of craft practices that had largely constituted it up to that point. Instead of the improvization that was widely characteristic of film production at that time, Ince developed much more methodical procedures based on his perfection of the continuity script. Above all, he radically separated conception from production, and broke the shooting process down into disconnected segments that were then reassembled into the final film at the postproduction stage. The continuity script could thus be used much like an industrial blueprint. On these foundations, an advanced division of labor started to make its appearance in the motion-picture industry, and the elements of a modern managerial model of production were installed (Bordwell, Staiger, and Thompson 1985; Staiger 1982). The net result was to endow production companies with a greatly expanded capacity to control the entire filmmaking process, and, above all, to exert discipline over the conduct of talent workers like writers, directors, and actors.

26


If Ince is notable for his managerial innovations, D. W. Griffith is celebrated above all for his influence on the whole conception of cinematic entertainment. He is credited among other things with developing the close-up, the flashback, and fade-out techniques in cinematography (Florey 1923; O’Dell 1970). Griffith’s career is a long and complicated one. After an already active professional life in motion pictures on both the east and west coasts, Griffith achieved his most important directorial success in 1915 with the production of Birth of a Nation for the Epoch Producing Corporation. The film was shot at the Fine Arts Studio located at the intersection of Hollywood and Sunset Boulevards, and is reputed to have cost $85,000, making it five times more expensive than any other film heretofore produced (Clarke 1976; Stern 1945). In addition, the film subsequently earned gross revenues of over $18 million, far more than the earnings of any other motion picture of the silent era so that it stands on record by a very long margin as the first blockbuster film. Birth of a Nation also helped to make cinema acceptable to the respectable middle class in America, which had hitherto largely considered it to be an offensively inferior medium of entertainment (Izod 1988; Slide 1994). This dramatic surge in Hollywood’s business fortunes and reputation was the added spark needed to push the entire complex to the undisputed forefront of motion-picture production. The surge was magnified, moreover, by Griffith’s development of an even more grandiose project in 1916 into which he poured all of his profits from Birth of a Nation. This was the film Intolerance, produced on the enormous budget for the time of $2 million. Although the film failed commercially, the economic energies that it set in motion via its huge cast and lavish sets (built on a lot just across the street from where Birth of a Nation had been produced) no doubt sent ripples through the local community. These two films together stand not only as monuments to early cinematic art but also as major business experiments that contributed significantly to Hollywood’s forward momentum. If there is any breakthrough moment to be discovered in the development of Hollywood as a productive agglomeration, it can surely be best identified with these two films of D. W. Griffith. More generally, Hollywood’s critical takeoff as an industrial complex hinges around the year 1915 when an extraordinarily potent combination of commercial and cultural forces came together, and when Hollywood was finally transformed from its status as a simple branch-plant extension of New York’s motion-picture industry to a composite system with a strong endogenous dynamic of development. After 1915, the number of workers and establishments in the Hollywood motion-picture industry shot up

27


both absolutely and relative to the rest of the country. The Motion Picture Studio Directory2 of 1918 records a total of thirty-seven studios in Los Angeles, forty-five in New York, and a scattering in other places, including seven in Jacksonville, and six in Chicago. By 1919, 80 percent of the world’s motion pictures were being made in California (Davis 1993). Two years later, in 1921, according to the first Bienniel Census of Manufactures in the USA, the motion-picture industry in California as a whole employed 5,329 workers in 68 establishments as compared with New York State’s 3,922 workers in 20 establishments. It would certainly be difficult to account for this growth by reference to the physical characteristics of Southern California, except to say that they played a contingent and less than necessary role. Rather, the rise of Hollywood can best be seen as a consequence of the vigorous system of productive organization that evolved out of the disparate collection of branch plants that had drifted into the area in the six or seven years before 1915. The physical geography of Southern California is in fact irrelevant to the question of how the dominating economic and cultural power of the Hollywood motion-picture industry came to be, and it cannot be sufficiently reiterated that a mature motion-picture industrial complex might have sprung up virtually anywhere in the USA in the early years of the twentieth century. However, once the forces of agglomeration are set in motion at any location, increasing returns effects sustain an upward spiral of growth and development while making it increasingly difficult for other locations to compete. In many respects, New York, with its decisive first-mover advantages, might well have been expected to forge ahead as the main production center in the USA, just as other primary metropolitan areas in other countries (e.g. London, Paris, Berlin, and Moscow) maintained a leading role as centers of motion-picture production despite their unfavorable climates. If climatic conditions were so critical to success at this time, we would also need to explain not only why the French film ˆ te d’Azur, but industry did not migrate in any important degree to the Co more significantly, how the industry, given its location in Paris, came to dominate world trade in motion pictures up to World War I (Jarvie 1992; Thompson 1985). New York producers enjoyed the benefits of a head start, but squandered their initial advantages by pursuing a business strategy that undermined their long-term competitive advantages. By contrast, the Hollywood independents, most of whom were aggressively opposed to the Trust, pioneered an alternative and vastly more successful strategy based on the cultivation of mass audiences by means of decisive stylistic and content innovations. The superior dynamic of individual and collective

28


development that came to function in Hollywood, allowed it to catch up with and then greatly to surpass New York. These remarks can be taken in a certain sense as a translation into placespecific terms of the developmental model of the motion-picture industry proposed by Jones (2001). In Jones’ model, the origins of the industry as a going concern are analyzed in terms of the coevolution of entrepreneurial careers, institutional rules, and competitive dynamics, that is the main developmental axes of any viable community of producers. As Mezias and Kuperman (2000) have argued, in addition, the multiple overspill effects that run systematically through such communities help to maintain high levels of business vitality. An essential point that lies at the core of the present investigation is that the same effects will also always give rise to a certain propensity for distinctive agglomerations to appear in geographic space.

Consolidation: Toward the Golden Age of Hollywood The final years of World War I saw a slowing down in the pace of motionpicture production in USA, and even after the war was over, the industry remained morose for some time. With the dawning of the 1920s, however, the industry entered a new period of growth, most of which was now predictably concentrated in Hollywood. Among the important events of this period were the formation of the early majors, in addition to Fox, Paramount, and Universal. In 1918, the Warner Brothers Company opened its first west coast studio on Sunset Boulevard. The following year, the United Artists Corporation was founded by Charles Chaplin, Douglas Fairbanks, D. W. Griffith, and Mary Pickford to distribute their independently made films (Balio 1976b). CBC Sales Film Corporation was established in 1920 and then renamed Columbia Pictures in 1924. Metro–Goldwyn–Mayer was born out of a complex merger in the same year, and RKO, which similarly had its origins in a series of mergers, was formed in 1928. By this time, the main outlines of classic Hollywood were in place, and its rise to international preeminence assured. Already, in the mid-1920s, 30 percent of the industry’s total revenues were being generated by exports (North 1926). It was in the mid-1920s, too, according to Koszarski (1990) that the term ‘Hollywood’ started to be used in its generic sense to designate the motion-picture industry at large. The data laid out in Table 2.1 help to summarize the main economic trends in the motion-picture industry in the 1920s and on into the 1930s

29

Origins of Hollywood Motion-Picture Industry Table 2.1. Employment in motion-picture production, USA, California, and New York, 1921–37 Year

USA

California

New York

Employment Establishments Employment Establishments Employment Establishments 1921 1923 1925 1927 1929 1931 1933 1935 1937

10,659 9,904 11,518 16,013 19,602 14,839 19,037 27,592 34,624

127 97 132 142 142 140 92 129 83

5,329 7,137 NA 12,852 NA 11,182 16,417 23,278 30,408

68 48 72 78 NA 71 39 75 35

3,922 1,734 NA 1,907 NA 2,594 1,748 3,240 2,883

20 16 18 21 NA 26 24 24 21

Source : US Department of Commerce, Bureau of the Census, Biennial Census of Manufactures (NA ¼ not available).

in the USA. The data are consistent with the notion that California had by this time shifted into a sort of second phase of development, characterized by rapid absolute growth and increasing shares of both employment and productive capacity. Note in particular that in 1921 California had just under 50 percent of all employment in the US motion-picture industry and that by 1937 its share had grown to 87.8 percent. The industry correspondingly stagnated in New York, with its share of employment falling from 36.8 percent to 8.3 percent over the same period. Not only was Hollywood consolidating its technical, commercial, and cultural domination of the entire motion-picture industry at this time, it was also widening and deepening its roots in the local area. It was in the 1920s that the basic distinction—still in use—between majors and independents came into being, and by 1930 the Hollywood production system could be succinctly described as comprising eight majors together with a surrounding constellation of smaller independents. The majors themselves could be further subdivided into the so-called big five (Fox, MGM, Paramount, RKO, and Warner Brothers), and the little three (Columbia, United Artists, and Universal Pictures), the distinction between them deriving from the fact that the former group owned extensive first-run exhibition facilities while the latter did not. The spatial distribution of both majors and independents in and around Hollywood in 1930 is displayed in Figure 2.2. In comparison with Figure 2.1, the spatial pattern revealed here indicates that a modest westward shift into central Hollywood had occurred in the intervening fifteen years combined with considerable locational intensification. Observe, in

30

Origins of Hollywood Motion-Picture Industry one mile one km

MAP AREA

B U R B A N K o ywo Holl

MALIBU SANTA MONICA

LOS AN GELES C OU N T Y

dW

STUDIO CITY

PASADENA

ORANGE COUNTY

GLENDALE

S H I L L

ga en hu Ca

D O O Y W L L O H

LONG BEACH

e

at

nd

le

Vine Gower

Universal

ay

Ventura

G

Hollywood

ns

et

Bre

a

Pico

La

RKO CULVER CITY MGM

Su

Wilshire

to

La Cienega

Fox

M

Paramount Highland

a nic

o

a

nt

Sa

Columbia

United Artists

Melrose

LOS ANGELES CIVIC CTR

So

S

BEVERLY HILLS

Warner Brothers

Washington Vermont

t

se un

Figure 2.2. Locations of motion-picture production companies in Los Angeles, 1930 Source: Majors are represented by a square. Addresses were obtained from a great diversity of publications, directories, and web pages. The information shown is probably not complete.

particular, the conspicuous clustering of production companies in the vicinity of Gower Street between Sunset and Santa Monica Boulevards. Many production companies making cheap B movies thronged together in this area, most of them small independents, though Columbia was also distinctly one of their numbers. As a consequence, the area was widely referred to as Poverty Row, or alternatively, Gower Gulch in reference to the large number of cowboy films made locally and the crowds of suitably costumed actors on the street seeking work as extras (Fernett 1973; Mezias and Mezias 2000; Pitts 1997). Figure 2.2 also indicates that some of the majors had by 1930 located in the geographic periphery of the main Hollywood complex where they could find cheap land for their spaceintensive sound stages and back lots.

31


As the majors grew by acquisition, merger and internal expansion over the 1920s, they were involved in a trend to vertical integration. In the first place, the big five studios were aggressively engaged in vertical integration of all three main phases of the motion-picture business, namely, production, distribution, and exhibition. By the late 1920s the big five controlled significant segments of the entire industry and exercised something close to monopoly power over first-run theaters. In cases where they did not own the theaters, they were nevertheless able to circumvent open market competition by means of blind-booking and block-booking strategies (Huettig 1944). In the second place, the production phase itself was also subject to vertical integration. Thus, in the major studios all the main tasks of filmmaking—writing, directing, acting, sound-stage operation, musical composition and performance, film editing, and so on—were brought together under one structure of ownership and employment. The most talented workers were signed up to long-term contracts, usually of seven years’ duration. The effect of this second form of vertical integration on plant size is diagnostically revealed in Figure 2.3, which shows that as the 1920s came to an end, the average number of employees per establishment in California’s motion-picture industry began to increase dramatically, rising from 165 employees in 1927 to just over 850 in 1937. These two types of vertical integration in the motion-picture industry in the 1920s and 1930s complemented and reinforced one another. On the one hand, the majors were in control of the entire motion-picture commodity chain and could find outlets for all their films irrespective of quality, hence allowing for stability in the organization of production. On the other hand, their contractual control of the most popular performers ensured that quality was sufficiently high on a sufficient number of occasions that mass defections of audiences did not occur. The Paramount Decree of 1948 undercut this model of productive organization and ushered in a new Hollywood in which vertical disintegration was now the watchword (Scott 2002; Storper and Christopherson 1987). Still, it would be an error to think of the major studios in the 1920s and 1930s as being totally integrated. On the contrary, they retained a degree of flexibility in regard to many types of production tasks that they could not efficiently internalize on a permanent basis. Numerous individuals at this time, even talent workers and skilled craftspeople were employed as temporary or freelance workers by the studios. Koszarski (1990) points specifically to the frequency of this practice in regard to cinematographers. Other types of worker that the studios had subcontract or freelance agreements were writers, dancers, character

32

Origins of Hollywood Motion-Picture Industry California

Employees per establishment

900

700

500

300

Rest of U.S

100

1920

1925

1930

1935

1940

Year

Figure 2.3. Average size of establishment in the motion-picture industry in California and the rest of the USA, 1921–37 Source: US Department of Commerce, Bureau of the Census, Biennial Census of Manufactures.

actors, extras, and so on. The studios were also subject to the normal ebb and flow of labor turnover among their rank-and-file employees, and their location at the heart of a dense and suitably socialized pool of workers certainly helped to keep the costs they had to bear as a result of this activity at a relatively low level. It would seem that a wide array of specialized services and supplies to the motion-picture industry was to be found in Los Angeles at this time, and one source published in 1928, lists multiple addresses under headings like film editing, film laboratories, orchestras, agents, cowboy equipment, costumes and props, animals, and so on, though we have no way of knowing from the information provided what was purchased by the majors and what by independent production companies. The majors also lent out directors, stars, and other elite workers to one another from time to time, and they jointly established the Central Casting Corporation in 1925 so as to bring order into the deployment of temporary labor in the industry. They sought, as well, to improve the

33


return on their fixed capital investments by renting out studio space to independents (Fernett 1988). These observations suggest that classic Hollywood represented a much more complicated and diverse production system than is set forth in those accounts that insist on its character as a monolithic structure of vertically integrated majors and mass production. That said, much further historic research is clearly needed on the interplay between organizational forms on the one side and external and internal economies on the other side in the Hollywood of the 1920s and 1930s. This remark applies not only to the majors but also to the much-neglected independents of this period. Over the 1920s and 1930s, then, Hollywood can be described as a distinctive industrial district imbued with multiple spillover effects flowing from its internal transactional order and the dense, many-sided local labor market that had developed in the urban community around it. This local labor market grew greatly in size as migrants from all over the USA and the rest of the world moved into Los Angeles in search of work in the motion-picture industry. Among them, we must count the great number of talented foreigners drawn to Hollywood by its rewards and gratifications, and whose number included, if we consider only directors in the 1920s, Ernst Lubitsch, Friedrich Wilhelm Murnau, Michael Curtiz, Alex¨ stro ¨ m (Robinson 1968, 1977). Like industrial ander Korda, and Victor Sjo districts everywhere, Hollywood also acquired an idiosyncratic superstructure of institutions in response to diverse needs for coordination and collective order. In 1922 the main production companies collaborated to establish the Motion Picture Producers and Distributors of America (MPPDA) as an instrument of joint representation.3 One of the principal early functions of the MPPDA was to regulate what threatened to become an explosive public relations situation as general perceptions of Hollywood’s moral values (hot on the heels of a series of scandals) became evermore negative (cf. Anger 1975; Shindler 1996). In 1930 the MPPDA formally adopted the so-called Hays code, a set of self-imposed ideological and moral injunctions, as a preemptive strike against extra-industry regulation. The MPPDA was also greatly exercised by labor relation issues, and in 1926, it helped to engineer the Studio Basic Agreement, representing one of the first efforts on the part of the industry to ensure smooth management–labor interactions. The Agreement was a simple two-page document signed by nine production companies and five unions in Hollywood that laid out a framework for adjudicating grievances and other disputes4 (Ross 1941, 1947). In the aftermath of the Studio Basic

34


Agreement, and in an attempt to head off full-blown labor organizing activities in Hollywood, the studios created the Academy of Motion Picture Arts and Sciences in 1927 to function essentially as an overarching company union (Nielson and Mailes 1995). The Academy, as initially constituted, had five branches representing producers, writers, directors, actors, and technicians. It failed, however, to establish any credibility as a labor organization, and in the wake of the Great Depression, when the studios demanded that workers across the board take deep pay cuts, it steadily gave ground to independent unionization movements5 (Ross 1947; Scott 2005). In spite of this concentrated development of productive activity and institutional order in Hollywood, the executive offices of all the major production companies remained in New York, from which base, they coordinated operations across production, distribution and management, and dealt with vital financial issues. This locational split in the industry’s functions was to last until well into the 1970s. On these foundations, Hollywood entered its golden age, a period of ‘mature oligopoly’ (Balio 1976c) that lasted from the late 1920s to the late 1940s when both the Paramount Decree and the development of television heralded an extended period of crisis and a reversal of earlier trends to vertical integration. The beginning of the golden age can be marked symbolically by Warners’ release of the first sound feature film, The Jazz Singer, in 1927 (Gomery 1976), though this in and of itself was only one expression of a continuing series of technical and cultural innovations (from technicolor to Busby Berkeley’s choreography) that reached into every corner of the motion-picture industry. Hollywood was now manifestly dominated by the major studios, each of which came to acquire distinctive stylistic features, ranging from the lavish mise en sce`ne typical of MGM films to Paramount’s urbane comedies. The motion pictures that flowed copiously out of Hollywood’s production system at this time evinced an unequalled command of narrative economy and visual storytelling that captivated popular audiences all over the world, though they were also roundly condemned by many European intellectuals, such as Adorno, Horkheimer, Duhamel, and Gide (Benjamin by contrast took a more tolerant attitude), for the cultural and political stupefaction that they were alleged to induce. Hollywood’s revenge, so to speak, can be regularly observed every Friday and Saturday evening outside the art house cinemas of Berlin, Paris, and London, as the avatars of the same European intellectuals line up to see the identical films today.

35


Conclusions This chapter has tried in both theoretical and empirical terms to understand something of the origins and growth of the Hollywood motionpicture industry. More broadly, the argument can be taken as an attempt to add historic and geographic substance to the kind of location theory that comes out of the work of Arthur (1987) and David (1985) with its synthetic description of agglomeration economies, path dependency, and lock-in. I have tried to outline a brief historic geography of Hollywood showing how it shifted in the mid-teens of the twentieth century from being a branch-plant extension of New York’s motion-picture industry to an economically sustainable agglomeration in its own right. Much of this shift can evidently be explained in terms of the development of a highly successful business model in Hollywood combined with a powerful and complex endogenous dynamic of growth based on its transactional, labor market, and innovative capacities. It is no doubt the case that the early movie pioneers in California really did locate there because they were in search of sunshine and mild winters, and certainly once they arrived there they made ample opportunistic use of the local landscapes. But Southern California does not have a monopoly on sunshine and mild winters and varied landscapes, and numerous other locations in the US south and west might just as easily have acquired (and some did acquire for a time) a reputation among early production companies as the place to be in the winter season. In brief, there might well have been another alternative geography of the motion-picture industry in the USA in the twentieth century in comparison with the one that actually came to pass. Had the early business history of the motion-picture industry played itself out in a different key, there is every reason to suppose that New York might have capitalized on its first mover advantages and retained its preeminence throughout the succeeding century. After about 1915, however, the die was cast in another direction. The Hollywood production complex now surged rapidly ahead, and over the next decade it decisively consolidated its ascendancy. These developments laid the foundations for a golden age that lasted until the late 1940s. But even as the old Hollywood fell into crisis after 1948, a new Hollywood rose up again on its ashes. The new Hollywood gradually and painfully reorganized itself around a highly disintegrated network model of production that actually intensified the play of agglomeration and has been one of the principal foundations of Hollywood’s continued competitive success down to the present day. The interesting question at this point is

36


whether the changes now occurring, as globalization begins to run its course, will usher in yet another incarnation of Hollywood, or whether they will open up a new window of locational opportunity, thereby encouraging the rise of new agglomerations of motion-picture producers in other parts of the world.

Acknowledgments This research was supported by the National Science Foundation under grant number BCS-0091921 with supplementary funding from the Haynes Foundation. I wish to thank Yixiu Ye for her able research assistance. This essay is an edited version of Chapter 2 of the author’s book On Hollywood: The Place, The Industry (Princeton: Princeton University Press, 2005). The author and editors are grateful to Princeton University Press for permission to reproduce copyrighted material here.

Notes 1. In 1915, four of the leading independent companies, Famous Players, Fox, Lasky, and Universal produced on average thirty-seven feature films each. The members of the Trust produced on average thirteen feature films each. See Hanson (1988). 2. Published in Los Angeles by the Motion Picture News. 3. The MPPDA is the antecedent of today’s Motion Picture Association of America. 4. Signatories (on the management side) were Universal Pictures, MGM, First National Pictures, Famous Players-Lasky, FBO Studios, Producers’ Distributing Corp., Warner Brothers, Educational Film Exchanges, and Fox Film Corporation, and (on the union side) the International Alliance of Theatrical and Stage Employees, the United Brotherhood of Carpenters and Joiners, the International Brotherhood of Electrical Workers, the International Brotherhood of Painters and Paperhangers, and the American Federation of Musicians. 5. The Academy, which still exists, has long since abandoned its original goals, and is now mainly known for its patronage of the annual Oscar awards.

37

3 The Coevolution of Technologies and Institutions: Silicon Valley as the Iconic High-Technology Cluster Martin Kenney and Donald Patton

In the 1990s Silicon Valley achieved iconic status for economic development planners globally. But how did Silicon Valley come into being? We demonstrate that its rise was a social process of bricolage in which actor’s fashioned solutions for various problems that they confronted with what they found at hand (Garud and Karnoe 2003). Frequently, these solutions were adapted from existing forms and then applied to new purposes. For the most part, they were responses to immediate problems, rather than wisely considered, far-sighted solutions by prescient economic actors maximizing their utility functions. Those solutions that appeared to work were diffused, repeated, and adjusted, gradually evolving into routines and institutions (Nelson and Winter 1982). These routines and institutions enabled further experimentation even as a stable repertoire of actions came into being. Borrowing from Spender’s notion (1989) of industrial recipes, we argue that through an unplanned iterated learning process Silicon Valley actors developed regional recipe for creating and nurturing start-ups. This chapter examines how technology and institutions coevolved to create an ecosystem within which entrepreneurs were able to encapsulate many of the new innovations in separate firms, as opposed to all of the innovations being commercialized by existing firms. The information/computer/electronics (ICE) and, to a far lesser degree, biomedical technologies have formed the core of the venture capitalfinanced start-up economy. For the last five decades, the ICE technologies have experienced exponential rates of improvement in cost and

38

The Coevolution of Technologies and Institutions

functionality combined with extremely large intellectual property content. Also, the ICE technologies have frequently experienced (or, alternatively, entrepreneurs have created) moments when entry barriers have been lowered sufficiently to allow nimble well-placed entrepreneurs to enter new market niches. Thus, on one dimension, understanding Silicon Valley is predicated upon tracing the evolution of technologies and the industries based on them, and, on the other dimension, the evolution of the institutions, practices, and cultural understandings that orient action. The chapter examines the building of the entrepreneurial support infrastructure and its co-evolution with the local high-tech industries. In our discussion we highlight the way Silicon Valley entrepreneurs developed new business models and often combined different technologies to create new business opportunities. We also consider the importance of culture as an explanatory variable arguing that culture particularly is as much a dependent variable as it is an independent variable. In our estimation, it is better to consider culture as having coevolved with the regional business activity—and in the case of Silicon Valley might better be seen as a learned set of guides to action, rather than as something emanating from a Gold Rush mentality or a set of personal attributes. In the conclusion, we reflect upon the implications of our findings for an evolutionary theory of the development of entrepreneurial, high-technology regions.

The Formation of Silicon Valley Entrepreneurship in Silicon Valley involves two separate sets of organizations (Kenney and Burg 1999). The first set of organizations are the ones from which the entrepreneurs emerge and the second are the organizations that specialize in supporting the entrepreneurial process. The primary source of entrepreneurs for Silicon Valley start-ups has been other firms (Gompers, Lerner, and Scharfstein 2003). Though Gordon Moore, a founder of Intel, believes that university research institutions contributed little to the evolution of the semiconductor industry (Moore and Davis 2001), generalizing from the semiconductor industry he underestimates the role of universities and corporate research laboratories in providing the support and intellectual space within which the seeds of new industries could develop (National Research Council 1999a, 1999b). A number of the defining firms in individual industries can be attributed to universities and corporate research laboratories. For example, 3Com,

39


Cisco, Yahoo!, Seagate, Google, Sun Microsystems, and Cadence are directly linked to Bay Area corporate research institutes and universities (Kenney and Goe 2004). Having a rich source of entrepreneurs is very important, but it is the institutions that nurture the firms they create which give Silicon Valley a powerful advantage over other regions. Silicon Valley hosts a set of interdependent institutions specialized in supporting firms, particularly technology firms capable of extremely rapid growth. These institutions form what observers termed an ecosystem, a social structure of innovation, or an incubator region (Bahrami and Evans 2000; Eisenhardt and Schoonhoven 1990; Florida and Kenney 1990; Lee et al. 2000; Todtling 1994). However, this ecosystem was not sui generic, but rather formed over time in tandem with the industries that were formed in the region. Establishing when Silicon Valley was formed is not simple. Conventionally, it might be dated to the decision by William Shockley to establish Shockley Semiconductor in Silicon Valley. This was a defining moment and there was an element of serendipity in the choice of Palo Alto for the firm. Shockley also had considered the Boston area where there were already a number of transistor firms using germanium as their substrate, MIT was producing numerous technically capable personnel, and a group of early technology adopters (the minicomputer firms) were on the verge of being established. On reflection, there seems little doubt that a number of other regions, such as Los Angeles, Long Island, or northern New Jersey would have had sufficient technical personnel, lead customers, and other institutional supports to allow an industry to take root. For example, the germanium-based transistor firms on the East Coast particularly Boston might have switched to silicon, which ultimately became the substrate of choice for the most important technology of the late twentieth century as the silicon semiconductor became the ubiquitous enabling technology for digitization. Or, alternatively, Texas Instruments (TI) in Texas might have begun to spin-off firms. Fortunately or unfortunately, TI was not as badly managed as Shockley Semiconductors and it never experienced the mass resignations that led to the creation of Fairchild Semiconductor. It was Fairchild Semiconductor that began leaking people and contributed the start-ups that eventually transformed the region into what the editor of the Electronic News first described in 1973 as the Silicon Valley. If Shockley Semiconductor was the bad seed that soon failed, Fairchild was the most fecund of all. However, in some sense, survival is a function of the environment and the environment into which Fairchild was born was munificent. To illustrate, it is convenient to explore what existed in

40


the prehistory of Silicon Valley not as a theological exercise, but as a partial explanation of why the semiconductor industry would find the region such a conducive environment. There was an existing electronics industry that could be traced back as far as Lee De Forest and the invention of the vacuum tube (Sturgeon 2000). Also, other entrepreneur-based, high-tech electronics firms, the most salient of which were HewlettPackard and Varian, were located in the region. Though these firms were not semiconductor firms, they were electronics firms and HP was especially important because it produced the basic equipment that all electronic firms needed. Shockley was not the first person or firm to decide to locate in the South Bay. In 1952 IBM decided to establish a branch of its Yorktown Heights research laboratory in San Jose to tap the skilled personnel in the area.1 Approximately eighteen years later Xerox established its West Coast research facility in Palo Alto to access the skilled labor pool. Already in 1958 there were real business reasons to establish a new high-technology electronics firm in the region. Shockley’s relocation to Palo Alto was serendipitous and, at least, partially motivated by his desire to live close to his mother. However, the history of Silicon Valley also includes the role of Frederick Terman initially Dean of Engineering and then Provost at Stanford (Sturgeon 2000). A professor of electrical engineering, Terman admired the MIT model of university interaction with business. He was also a fervent believer in the economic potential of electronics, championing the establishment of electronics firms in the region and assiduously working to both attract them and support entrepreneurs including those on his faculty wishing to establish firms. Terman urged Shockley to establish his firm in Palo Alto. Serendipity brought Shockley to Silicon Valley, but Fredrick Terman was a centrally located actor actively trying to manipulate fate. The entrepreneurial support system would also evolve with the development of new firms. Yet it would be incorrect to attribute the creation of the VC industry directly to Fairchild. History shows that the capital required to start Fairchild was raised by Arthur Rock, who, at the time, was a manager at the New York investment bank Hayden Stone. And yet there were already a number of informal investors in the region that were willing to invest in new electronics start-ups. Thus, even though the capital to establish Fairchild was raised in New York, there was already a group of proto-venture capitalists in formation at the time of Fairchild’s establishment. In fact, there was a history of Stanford professors and administrators investing in start-ups that can be traced back to the early

41


investors in Federal Telegraph (Sturgeon 2000). Though it is difficult to be certain, there is anecdotal evidence that the Bay Area was already one of the national centers for angel investors in electronics.2 The Bay Area VC industry grew with the semiconductor industry, but it is important to note that semiconductors were only one of a number of industries funded. Whether Fairchild should be considered irreplaceable in the formation of the Silicon Valley high-technology cluster is probably unanswerable. Leslie (2000) argues that a microwave technology cluster was established in the region at roughly the same time (1950s) on the basis of defense department research. From 1955 through the early 1960s, there was a high-technology electronics boom and many firms were formed in the region with funding from informal investors. There is reason to believe that a high-technology cluster of some sort was evolving and would have continued to evolve. However, the region is termed Silicon Valley with good reason. To return to our earlier statement, the semiconductor was the most important technology of the twentieth century, and it was a critical input that made new industries such as work stations, personal computers, and computer networking possible. There can be little doubt that the semiconductor is at the heart of the dominance that Silicon Valley has shown for the last three decades.

Entrepreneurs, Technologies, Firms, and Industries in Silicon Valley The evolution of Silicon Valley is based on its entrepreneurs, the technologies they commercialize, and the firms they create. Figure 3.1 is a chronology of the most significant technologies that have fueled the region’s growth. In some cases, a technology was developed in Silicon Valley, but eventually shifted out of the region entirely. Of course, the sources are not always local as the semiconductor technology was imported from Bell Laboratories in New Jersey. The level of employment is a good measure of which industries were most important. Figure 3.2 shows that software currently employs the greatest numbers in the region having overtaken components (semiconductors) in 1996. It continued to grow very rapidly in the late 1990s fueled by the dot-com boom. The aerospace sector represented by guided missiles has been largely steady at 20,000 through the entire period with the exception of the 1980s when it doubled due to Reagan’s Star Wars plan. The growth of computer and peripherals’ employment is interesting

42

Stanford, HP, UC Berkeley, Xerox PARC, IBM San Jose, UCSF SEMICONDUCTORS

1960

Memory COMPUTERS 1965 MAGNETIC STORAGE

Minis

1970 SOFTWARE Floppy disk drives

1990

RISC ASIC

COMPUTER NETWORKING

Semiconductor equipment

1985

Relational Hard disk drives PC software databases Games software

Microprocessor Semiconductor design software

1980

Disk drive part suppliers

1975

Local area network

Biotech Game machines

Wide area networks Internet Optics

1995 (c) MartinKenney

Figure 3.1. Genealogy of Silicon Valley technologies

Apple

Micros Workstations

The Coevolution of Technologies and Institutions 200,000 180,000

Computers & peripherals

160,000 140,000

Communications equipment

120,000

Components

100,000 80,000

Guided missiles

60,000

Instruments

40,000 Software

20,000 2001

1998

1995

1992

1989

1986

1983

1980

1977

1974

*1971

*1968

*1965

*1962

*1959

-

Figure 3.2. Employment in four Bay Area counties, 1959–2001 Note: Data before 1998 was collected by SIC code. For 1998–2001 data was not available in SIC codes. Therefore data was collected in NAIC codes that approximate SIC codes.

because it peaked in the early 1980s and then decreased to approximately 25,000 in 2001. Benefiting from the dot-com boom, communication equipment employment grew from approximately 20,000 in the 1980s to over 50,000 in the late 1990s. Scientific instruments also have been an important contributor to Bay Area employment since the 1970s. However, the overwhelming growth has been in software, an industry that did not even merit a separate category until the early 1970s; a few years after IBM signed the consent decree unbundling software and hardware. Employment provides one perspective on the structure of the Bay Area high-technology industries. Figure 3.3 indicates the number of establishments in each industry, and this provides a different perspective. The number of establishments in each industry differs so radically that a logarithmic scale was required to present the data. Notice that during the entire period there were no more than six establishments in guided missiles. In the case of components, instruments, communication equipment, and computer and peripherals, the number of firms was in the hundreds, though obviously there was much churn during the entire period. In absolute terms, the number of computer establishments declined since its high in the 1980s. This corresponds with the proliferation and later shakeout of microcomputer and workstation manufacturers. The one industry showing a continuing high rate of entry is software, which, with a few exceptions of which the most notable was the collapse of the

44

The Coevolution of Technologies and Institutions 10,000 Computers & peripherals Communications equipment Components

1,000

100

Guided missiles 10

Instruments

2001

1998

1995

1992

1989

1986

1983

1980

1977

1974

1971

1968

1965

1962

1959

Software 1

Figure 3.3. Establishments in four Bay Area counties, 1959–2001 Note: Data before 1998 was collected by SIC code. For 1998–2001 data was not available in SIC codes. Therefore data was collected in NAIC codes that approximate SIC codes.

dot-com bubble in 2001, has continually grown in the number of establishments and employment since the Census of Manufacturing first began collecting data in 1974. The quantitative indicators and the figure provide an overview of the development of the region. In the following sections, we examine the development of the most salient industries and firms providing a richer description of the coevolution of the region, technologies, and industries.

Semiconductors and Ancillary Industries Over the last four decades, semiconductor technology has been characterized by one overwhelming dynamic, namely Moore’s law, which has correctly predicted that the areal density of transistors would double every eighteen months, and since the cost of a semiconductor device is roughly comparable to the chip’s dimensions, either performance increases for the same price or price drops accordingly. What this means is that each new generation of semiconductor devices is able to process more information than the previous generation, providing the opportunity to increase the speed and capability of any artifact containing semiconductors. As a result, products containing ICs experience constant improvements in functionality, and functionality that formerly was too expensive or even impossible to undertake continually becomes less expensive and enters the realm of the possible.

45


Semiconductor technology was so fecund in opening new economic spaces that new business opportunities repeatedly emerged, and the cognoscenti had opportunities to create their own firms. This fecundity is illustrated by the fact that Fairchild and its successor firms experienced 134 spin-offs by 1986 (SEMI 1986), and there have been more since then. In tandem with the increase in areal density-making integrated circuits less expensive per transistor, the cost of a fabrication facility doubled every four years (Leachman and Leachman 2004). When Fairchild began producing chips, converted pizza ovens were used for the baking process. By 1975, a fabrication facility cost approximately $50 million (OhUallachain 1997: 220) and in keeping with what product life-cycle theory would predict, entry costs increased to the point at which there were far fewer entrants.3 According to SEMI’s genealogy, from 1974 to 1980 inclusive there were 21 entrants in Silicon Valley (or an average of three per year). In the seven prior years 1967–73 inclusive, there were forty-three start-ups (an average of 6.1 per year). Yet in the following six years from 1981 through 1986 inclusive, forty-six firms were established (an average of 7.7 ¼ 46/6 firms per year). The increased rate of entry after 1981 was the result of a collective action solution to increasing cost of fabrication. Beginning in the early 1980s, a number of start-ups were established to design and market new ICs. However, they contracted for manufacturing from the integrated producers that had spare capacity. This circumvented the entry barrier created by the capital cost of fabrication. The integrated producers benefited, because their expensive fabrication facilities could be fully utilized. The difficulty with this solution was that during semiconductor market booms, the integrated manufacturers reclaimed their capacity, and the fabless firms often lost their access to the fabrication capacity. A market-based solution was the silicon foundry specialized in contract manufacturing. These foundries, which were established in Taiwan, were pure contractors that did not compete with their customers. As specialists they had to be willing to invest, provide excellent service, and rapid turnaround (Leachman and Leachman 2004). Soon, a number of Taiwanese firms were established to produce chips designed by other firms. This developed into a symbiotic relationship eliminating the high cost of manufacturing as an entry barrier and unleashing a plethora of new Silicon Valley semiconductor start-ups that specialized in design and marketing. Semiconductor devices were the foundational industry for the region, and Fairchild was the ideal typical Silicon Valley start-up. However, the key to continuing entrepreneurship in the semiconductor industry has

46


been the ability to create new business models. When the cost of a fab became an entry barrier in the early 1970s, a collective action solution was fashioned that reopened the industry to start-ups and the pace of start-up formation, once again, accelerated. Developing this solution was only possible because the actors in the support infrastructure were willing to back start-ups pioneering new business models. Frequently, an industrial cluster will both attract and spawn supplier firms for the core industry (Porter 1998). The roots of the semiconductor equipment industry can also be traced to Fairchild. Fairchild initially built its equipment internally, but soon decided to divest these activities and assisted the spin-off of firms like Electroglass, Kasper, and Micro Tech (Moore and Davis 2001; von Hippel 1988: 173). The most significant surviving Fairchild-linked firms are Applied Materials (established in 1967), which is the largest semiconductor equipment maker in the world; KLA (established in 1976); Tencor (established in 1976); Lam Research (established in 1980); and Novellus (established in 1984); all of which are located in Silicon Valley. Though very few ICs are made in Silicon Valley, it shares with Japan the distinction of being the global center for semiconductor production equipment design and manufacturing. In fact, the headquarters of the Semiconductor Equipment and Materials Industry Association is in San Jose. In the last three decades, a merchant IC design automation software industry emerged. This software was a response to the fact that the increasingly complex IC designs could no longer be done on paper without an unacceptable number of errors. Thus, in the late 1960s the integrated semiconductor firms began developing software tools for design automation. Fairchild was an early leader as its engineers developed Computer-Aided Design (CAD) software (Walker 1998). At the beginning of the 1980s, a number of IC design software start-ups were established. Many of the advances were made at UC Berkeley and certain UCB professors participated in forming start-ups. For example, in 1982 Solomon Design Associates was established by Jim Solomon who was assisted by a number of UCB professors. SDA merged with ECAD, a start-up that was publicly traded, to form Cadence Design Systems (Solomon 1988). Today, Cadence is the world’s largest supplier of electronic design technologies, methodology services, and design services. In 1986, Synopsys, a major competitor, was founded in North Carolina, as a spinout from a General Electric acquisition, Calma. However, at the suggestion of its VC investors moved it to Silicon Valley (de Geus 1988). As the software improved, an ever-greater number of the IC firms abandoned their

47


in-house software and purchased software from the design software vendors. The standardization of the design software facilitated the rise of the fabless semiconductor firms as they could purchase their design tools, releasing them from the onerous task of creating their own software. The software also allowed the foundries to stipulate their manufacturing parameters in the software to be used by the designers. In other words, the design software became the interface between the designers and the manufacturers. The development of a supplier industry enriched the semiconductor industry ecosystem. A number of these firms were very successful and thus also enriched the venture capitalists investing in them, thereby contributing to VC industry growth. Each further development of the division of labor reinforced not only the semiconductor industry, but also the institutions supporting the entrepreneurial infrastructure. The semiconductor industry was important for a number of reasons beyond its technological fecundity. First, the large number of spin-offs in the 1960s encouraged the already existent entrepreneurial culture. Second, the semiconductor industry provided significant investment opportunities for venture capital. Third, it attracted attention to the region and many of the region’s entrepreneurs including Robert Noyce, Gordon Moore, and Jerry Sanders, became iconic figures, even as the region became Silicon Valley.

Computers Silicon Valley has been the birthplace of computer firms serving a wide variety of product classes (i.e. IBM-compatible mainframes, minicomputers, work stations, personal computers, etc.), though interestingly enough, Silicon Valley only became dominant in workstations. Rather than discuss the entire history of computer producers in Silicon Valley, the greatest attention is paid to the computers based on microprocessors, a new category of integrated circuits that were pioneered by Silicon Valley semiconductor firms in the early 1970s. These small computers dedicated to individuals were central to the establishment of the networked, distributed computing paradigm that dominates contemporary computing. This period is also interesting, because in personal computing today, Silicon Valley firms produce many of the crucial components, even though Silicon Valley is no longer the center of the PC industry. Initially, of course, the computing industry was dominated by IBM and the various other mainframe producers. It was with the minicomputer,

48


which was so important for the building of Route 128, that Silicon Valley firms began to experience success in computing. The greatest success was HP, but many other VC-financed start-ups entered the market; some of them, such as Tandem Computer established in 1975 to offer fail-safe computing, captured unique niches while others were me-too firms. Another important firm was Amdahl, which was founded in 1970 by Eugene Amdahl, a key IBM computer designer, and offered an IBM plugcompatible computer. A number of these computing firms were successful, but they did not spawn waves of new firm creation and entirely new industries, rather they occupied niches and created large capital gains for investors. For Silicon Valley, the great wave of new firm creation in computing would begin in the late 1970s when two technological trajectories combined to create personal computing. The first trajectory was the work at Xerox PARC, which developed an expensive workstation that was a personal computer, that is not a time-shared computer. The Xerox effort, in fact, created a workstation designed by engineers for engineers. Xerox failed to capture the market, but many start-ups entered the market to try where Xerox was failing. Very quickly, a market for workstations developed and an industry emerged led by Sun (Stanford University Network) Microsystems based in Silicon Valley, and Apollo Computers based in Route 128. Sun became the dominant workstation provider, though in the late 1980s and early 1990s, it was challenged by another Stanford spinout, Silicon Graphics, Inc., which specialized in graphics computing. Eventually, the workstation firms would morph into the computer server providers. The other personal computing trajectory was what was then called microcomputers, and it led directly to the PC. Beginning in the mid1970s, many hobbyists and engineers including Apple’s Steve Jobs and Steve Wozniak began building computers using the newly introduced microprocessors from Silicon Valley firms, such as Intel and Zilog, and the non–Silicon Valley firm Motorola. Silicon Valley soon became a hotbed of hobbyist computer start-ups with their locus in the now famous Homebrew Computer Club that met at Stanford University (Freiberger and Swaine 1984; Langlois 1990).4 Of all the start-ups, Apple Computer was the most strategic as Steve Jobs actively tapped the Silicon Valley entrepreneurial support structure (Young 1988: 151). By utilizing this infrastructure and conforming to its requirements, Apple was transformed as investors required the appointment of experienced management and made other changes necessary to establish a real business. This support helped tip the scales for Apple’s survival and growth.

49


During the early 1980s microcomputer start-ups proliferated. By all measures, the region was on its way to becoming the industry center. New firms were being established to provide software (e.g. Visicalc was developed by Bay Area entrepreneurs) and components. But then in August 1981, IBM introduced its PC, which rapidly became the dominant design, and nearly all the non-IBM compatible microcomputer firms in Silicon Valley and other places disappeared. Within three short years, most Silicon Valley microcomputer firms, with the notable exception of Apple, left the business (Angel and Engstrom 1995). After the mid-1980s, Silicon Valley would not host any new PC companies with the exception of HP, which entered during this period. The demise of the PC industry did not mean that Silicon Valley would not benefit from the diffusion of PCs. Numerous start-ups found opportunities in supplying components including microprocessors (Intel and AMD), BIOS chips (AMI, Phoenix Technologies, and Chips and Technologies), graphics chips (S3, Nvidia, and Cirrus Logic), hard disk drives (HDDs) (Seagate, Quantum, and Conner Peripherals), and even computer mice (Logitech and Kensington). The loss of the personal computer industry to IBM and then the cloners created new markets for peripherals and components that Silicon Valley firms could supply. With the introduction of the IBM PC, with its simple architecture and the ability of low-cost cloners to enter the market, Silicon Valley’s technological prowess no longer provided any particular advantage for PC assembly. Apple survived, in an ever-narrowing niche, on the basis of marketing and some desirable software features. Despite this, Silicon Valley’s position as a center for computing systems firms deteriorated as the PC turned computing hardware into a commodity, and eroded the workstation market. In historic terms, with each new computing category Silicon Valley firms were early leaders, and yet, in some cases, the industry evolved in ways that prevented them from remaining in that industry.

Peripherals—Magnetic Storage The origins of the magnetic data storage industry can be traced to research conducted in IBM’s San Jose Laboratories. Beginning in the 1970s, after the IBM consent decree, which unbundled IBM’s software and allowed plug-compatibility (see Amdahl above), entrepreneurs began to leave IBM’s San Jose operation to establish firms to exploit the new market opportunity of supplying storage devices for the new entrants. Soon,

50


Silicon Valley experienced a wave of spin-offs in HDDs similar to the earlier one in semiconductors. During the 1980s, in tandem with the rapid growth of the PC industry, there were many VC-financed entrants in the HDD industry (McKendrick et al. 2000). Since the integrated HDD manufacturers like IBM and DEC would not sell components to the new entrants, there was enormous demand (Christensen 1992: 95). These small independent HDD manufacturers were eager to buy components, creating opportunities for component supplier start-ups. This prompted a massive wave of start-ups as the venture capitalists initially experienced enormous returns through public stock offerings and acquisitions. However, by the mid-1980s, the HDD industry experienced a powerful shakeout of both HDD manufacturers and component suppliers (Bygrave and Timmons 1992). The collapse of the HDD Bubble did not mean that there were no new opportunities in fields related to magnetic storage. For example, in the 1990s, a new area of VC funding was HDD arrays, which are groups of HDDs using sophisticated software working together so as to provide redundancy and back up (McKendrick, Doner, and Haggard 2000; McKendrick 2001). Still later there was a wave of start-ups commercializing storage area networks, which combined networking technology with storage technology to optimize a firm’s usage of its various data storage systems. In summary, magnetic storage exhibited similar technical and organizational characteristics to semiconductors. The technology was rapidly improving; there was a global class source of ideas and technical personnel located in Silicon Valley, and a similar entrepreneurship and spin-off dynamic. As with semiconductors, as the entry barriers in HDD manufacturing increased, the entrepreneurs found new opportunities in disk arrays and storage area networks. Despite earlier shakeouts, VC proved willing to fund storage-related start-ups with new business models.

Computer Networking5 The first computer networking firms in Silicon Valley were established in the early 1970s (Burg 2001). Time-sharing of minicomputer capacity was one of the earliest forms of computer networking, and a number of startups were established in Silicon Valley and other regions to exploit it. As a greater number of computers were installed on corporate campuses, an opportunity arose to provide technologies that would allow for faster data transfer rates through local area networks (LANs). The initial opportunity was in exchanging data between mainframes.

51


The proximate cause for dramatically increased interest in computer networking was an effort that began in the early 1970s to automate the office. This office of the future required a network to share files between computers and expensive peripherals such as printers and data storage devices. A pioneer in this quest was Xerox PARC, which in the mid-1970s created a system of small computers, laser printers, and data storage devices networked by what would be called Ethernet. PARC was not alone in this effort; minicomputer firms such as Wang Computers were also trying to create the future office. At the end of the 1980s, computers were proliferating and entrepreneurs began forming firms to design and produce networking equipment, which interestingly enough was dependent on semiconductors capable of signal processing. At the time, the market was still small and there were no standards to ensure computer interoperability. The critical event that catalyzed the formation of an industry was the 1980 decision by Xerox to offer low-cost licenses for the Ethernet standard. In 1978 Robert Metcalfe left Xerox PARC and in 1979 started 3Com. In rapid succession, Zilog lost three groups of LAN entrepreneurs. As Ethernet became the de facto standard, a positive feedback loop ensued as the increasing number of users created a growing market for yet other innovations (Burg and Kenney 2003), and venture capitalists became more confident in funding firms (Burg and Kenney 2000). The proliferation of LANs, many running different protocols, created an opportunity for an interconnection solution. A number of firms were created to solve this problem. The most successful would be Cisco Systems, a Stanford University spin-off that commercialized a multiprotocol router. In the early 1990s, data communications traffic exploded as LANs proliferated and wide area networks were created. File-sharing and e-mail became standard business applications, and corporations began interconnecting their global operations. The increasing standardization of the datastream meant that a simpler, cheaper, and faster solution, the switch, could be deployed. In typical fashion entrepreneurs began leaving existing firms to establish switching firms with VC financing. To ensure they did not miss this new technology, the established networking firms, such as Cisco, Synoptics, and 3Com, acquired many of these switching start-ups for large premiums, encouraging greater investment and yet more spinouts. In the 1990s, the networking firms, and especially Cisco, developed a strategy of scanning their ecosystem to identify firms developing important new technologies and markets. Start-ups that were experiencing the greatest success were then acquired. In effect firms, especially Cisco, were

52


using the VC financing system as an integral component of their R&D strategies (Mayer and Kenney 2004). This encouraged a further proliferation of networking start-ups established in the hope that they too would be acquired. The opportunities in networking were not limited to simply increasing speed and bandwidth. The network also became more complicated, thereby providing entrepreneurial opportunities for network management, security, and other software and hardware such as specialized ICs; many of these opportunities were exploited by start-ups. By the mid-1990s, computer networking had become one of the core Silicon Valley industries. A business model emerged in which venture capitalists-funded start-ups that were established with acquisition as an exit strategy. Cisco pioneered a new corporate strategy of using the Silicon Valley start-up ecosystem to identify the new technologies that would affect its business. As firms competed and grew and yet others were formed, Silicon Valley increasingly became the knowledge center for computer networking. This deep knowledge meant that Silicon Valley firms, entrepreneurs, and venture capitalists would be uniquely positioned to see the next big thing.

The World Wide Web The World Wide Web (WWW) protocols were not a product of Silicon Valley; in 1991–2 they were developed at CERN in Geneva (Abbate 1999; Kenney 2003). At the time, there were few start-ups aiming to exploit the Internet, which was still largely an academic operation funded and controlled by the US federal government. In 1993 entrepreneurs had not yet comprehended the opportunities that the Internet represented. There was also a delay in convincing venture capitalists that the WWW presented an investment opportunity (Ferguson 1999). However, the lag in comprehension did not last long, especially in Silicon Valley, and by early 1994, venture capitalists were receiving business plans from entrepreneurs with ideas for the commercial exploitation of the WWW. The first easy-to-use Web browser Mosaic was developed at the University of Illinois and given away for free. Mosaic formed the basis of one of the earliest Internet startups, Netscape, which was established in April 1994 by Jim Clark, an exStanford professor and founder of Silicon Graphics Inc. He went to the University of Illinois and hired most of the key persons who had designed Mosaic and moved them to Silicon Valley. Less than one and one-half years later, Netscape had an initial stock offering in August 1995 at a

53


valuation of nearly $1 billion. Netscape’s remarkable increase in value alerted every venture capitalist and entrepreneur that the WWW was a new opportunity. Given the greater VC resources and large number of entrepreneurs, the Bay Area quickly became the center for WWW startups (Kenney 2003; Zook 2002). As the number of WWW users exploded, new business ideas and opportunities proliferated. This expansion provided opportunities for yet other start-ups to develop new software and Web-based services. Businesses were built around searching and cataloging other sites, providing instant messaging, selling products online, software tools, and Web-hosting services among others. Investors were willing to fund entrepreneurs experimenting with an amazing proliferation of business models. As these firms went public or were acquired at large premiums, and as the user base grew, the high stock market valuations for Internet-related firms unleashed a frenzy of investing encouraging even greater speculation. By mid-1999 there was what might be termed a full-scale investment panic as public investors drove the price of new issues skyward. By the time the Bubble ended in 2000, more than 370 self-identified Internet-related firms had gone public and their total valuation had reached $1.5 trillion, though they had only $40 billion in sales (Perkins 2000). Approximately, 50 percent of all the new Internet firms were headquartered in the Bay Area. In 1999, the average return for early stage VC funds was 91.2 percent, the highest in history (NVCA 2000a).6 The returns for the most successful funds were astronomical—many had annual returns of 100 percent and one even had a 400 percent annualized return. The amount of VC invested in Internet-related firms grew from a nearly negligible $12 million in the first quarter of 1995 to $31 billion in 1999 (NVCA 2000b). In percentage terms, the increase was equally dramatic, growing from a negligible percentage in 1995 to over 60 percent of total investment in the fourth quarter of 1999 (NVCA 2000b: 31). Faster than anywhere else, Silicon Valley entrepreneurs glimpsed the potential of the WWW as a commercial opportunity and then mobilized the resources necessary to try to enact that future.

Software The richness and diversity of software firms in Silicon Valley are remarkable. As mentioned earlier, the highly specialized field of semiconductor design software is almost entirely located in Silicon Valley. In 2003,

54


software was the largest employer in Silicon Valley, and despite recent setbacks in the longer term it is expected to grow (see Figure 3.2). In software, as has been the case with other industries, Silicon Valley has pioneered certain software sectors and then lost them. For example, it is no longer a significant producer of PC software, with certain exceptions, such as Intuit for PC financial applications and various PC game software firms. Microsoft’s PC software monopoly resulted in the demise of Silicon Valley firms such as Visicalc and Borland Computer. Even when new PC software such as the Netscape browser is commercialized in Silicon Valley, Microsoft has been able to use its monopoly power to destroy them. The only major survivors have been the tax software producer, Intuit, and the utilities software firm, Symantec—and both of these are threatened by Microsoft. Silicon Valley firms have been far more successful in business productivity software. The most significant of these are relational database software, which was pioneered roughly contemporaneously at IBM’s San Jose Laboratories and UC Berkeley. All of the key independent relational database firms (with the exception of Microsoft, a late entrant) are located in Silicon Valley. The largest of these is Oracle which is the second largest independent software firm in the world. Other important firms include Sybase, Informix (purchased by IBM), and IBM. Oracle, in particular, has spawned other important business software firms including Peoplesoft and Seibel, which pioneered other niches in the business software field. In entertainment software, Silicon Valley also experienced success. Here, the Silicon Valley pioneer was Atari, which later collapsed. Atari’s demise in the 1970s permitted the control over the game boxes to move to Japan, and today Japan is the major competitor for the US game software makers. The largest entertainment software firm in Silicon Valley is Electronic Arts, which is located in Redwood City. Electronic Arts, which used to be a developer, today not only produces games but also distributes them for other producers. They are intimately connected to the cutting-edge PC graphics chipmakers also located in the region, because these graphics capabilities determine software usability. Drawing upon a similar expertise base, the Bay Area is also host to a number of leading computer film animation firms including Pixar and Lucas Arts. Producing special effects, these firms are critical for contemporary cinema and computer games.7 Though Silicon Valley has not proved to be as dominant in software as in some other industries, it is one of the key global software centers. Today’s Silicon Valley start-ups use Linux operating system and programs such as Java as the basis of their products. In fact, the Finnish developer of

55


Linux moved to Silicon Valley. Open source offers them a way to outflank Microsoft’s grip on software innovation. Moreover, even as this is written, Silicon Valley start-ups are integrating Indian software production capabilities into their business plans, in the same way, as the semiconductor design firms have integrated Taiwanese fabs into their business plans. In other words, new business models are still being created.

Support Infrastructures and Culture The development of a rich and complicated support infrastructure for entrepreneurs provides important advantages to Bay Area entrepreneurs. The goal of the actors in the support infrastructure is to participate in the capital gains that accrue when one of the start-ups is successfully sold, either to the public in an initial public stock offering or through an acquisition. The entrepreneurial support network has become so resource laden that the various actors in the network are willing to fund emerging ideas in new fields as has been the case in biotechnology (see Kenney 1986; Romanelli and Feldman in this volume), superconducting, and, most recently, nanotechnology. If these investments fail, as was the case in superconducting, only a relatively small proportion of the total VC resources and, perhaps, a few venture capitalists will be lost. If the investments succeed, as was the case with biotechnology, a new investment field will be created. Ultimately, the actors are agnostic as to what constitutes a suitable field for investment the market for the firms they support informs them by providing them with capital gains. The Silicon Valley culture benefits from interaction in many venues that contributes to cross-disciplinary information sharing and synthesis. With so many technologists, investors, and others interacting, there are ample opportunities for combining existing technologies to create new products (Hargadon 2003). One often cited example of this is the bioinformatics start-ups that combine the technologies of computing and gene-mapping. Many of these were formed in the Bay Area. The repeated successes in establishing new firms and then being able to garner large capital gains on a significant number of them created a culture of entrepreneurship. Interestingly, this culture differs remarkably from other entrepreneurial cultures that are based on the idea of establishing and then managing and controlling one’s own firm. The Silicon Valley culture is based on establishing and then selling the company to either the public or a corporate acquirer. In either case the entrepreneur loses control

56


of the firm. The objective, then, is capital gains, which under normal conditions can only be secured by creating a viable firm (though during the last high-technology bubble many unviable firms were created and foisted on the public). It also bears mentioning that other regions such as Boston and Israel appear to have cultures that are similar. Though we focus on identifiable institutions in this chapter, it is important to note that an entrepreneurial culture developed in Silicon Valley which, though not unique, can be characterized as extreme entrepreneurism. As Saxenian (1994) observes, during the economic boom periods changing jobs is a given part of the labor market in Silicon Valley. Over time, participating in a start-up has become a career path. This acceptance of start-ups as normal has reduced the career risk of becoming an entrepreneur. Moreover, whereas thirty years ago the entrepreneur was expected to use credit card debt and even mortgage their home as part of the process, in the last twenty years such measures are no longer necessary prior to receiving VC. It is not in the venture capitalist’s interest to raise the barriers to entrepreneurship and increase the concerns of the entrepreneur. This lowering of entry barriers has culminated in the mythology that failure will not necessarily prevent an entrepreneur from receiving funding for another start-up. Given that the Silicon Valley economy is based on capital gains, a culture and ideology encouraging entrepreneurship is a prerequisite and a natural outcome. In keeping with the capital gains-driven economy, one of the primary cultural and economic goals is to secure stock options or equity. This has led to an environment within which equity is extended to a large number of persons in the corporate hierarchy. The ownership of options elicits extraordinary effort from the employees and, if the firm is successful, creates many wealthy managers and engineers. A certain number of these experienced and now wealthy individuals will in turn be willing to invest in other entrepreneurs or even launch their own start-up, thereby perpetuating the entrepreneurial cycle. Another aspect of the Silicon Valley culture was memorialized in Michael Lewis’s book (2000) entitled The New New Thing, which described Jim Clark’s involvement in the creation of Netscape. In this case the hero is Jim Clark who ruthlessly capitalizes on the new WWW browser technology and reaps enormous capital gains. The region has developed a corporate environment within which new technologies, a great hack, and huge capital gains are the reigning myths. In this environment a hot new firm or technology attracts attention and floods of resumes. The ability to become involved in the hottest new technologies attracts many of the best

57


engineers in that field who desire to be involved in the newest technology. The economic incentives and culture are aligned to encourage high-risk entrepreneurship.

Reflections An evolutionary and systemic perspective provides an appropriate means for understanding Silicon Valley, and, by extension, other entrepreneurial high-technology districts (Avnimelech, Kenney, and Teubal 2003). Often discussions omit or elide the technological trajectories that underpin such industrial districts and overemphasize cultural aspects; we explicitly argue against this. The basis of much of this romanticization of the entrepreneur is a belief that the culture is sui generis. A more appropriate model would treat culture as a constructed and evolving social artifact. The entrepreneurs that have benefited from the system and the actors in the support infrastructure have every reason to support a specific set of cultural beliefs. The environment evolved, though not in a conscious directed manner, as a result of individuals pursuing various goals, one of the most important of which was the capture of capital gains. Viewed from a longer-term historic perspective, what is striking is how a number of the technologies exploited in Silicon Valley, such as semiconductors, magnetic storage, and computer networking, have had trajectories that have unfolded in such a way so as to enable yet further opportunities to establish new firms. In a number of sectors when potential for future start-ups appeared stymied by requirements such as enormous capital investments to create semiconductor fabrication facilities, new business models were developed to circumvent the entry barrier. An evolutionary perspective highlights the region’s remarkable success in repurposing its intellectual assets and attracting new talent from around the world. The constituents of the support infrastructure created their own niches and then were able to draw resources from the environment. They became actors trying to improve their processes, which by definition, included supporting and assisting the entrepreneurs. They also changed the environment by creating demand for entrepreneurs, reinforcing the cultural valuation of the entrepreneur, and routinizing the start-up process. These actions explicitly recognized that the entry barriers for entrepreneurship are not only financial but also social and psychological. The literature has treated the willingness-to-take-a-chance attitude in Silicon Valley as an

58


innate characteristic; we see it as a communally created social norm. The support infrastructure assisted in this in a wide variety of ways from developing an understanding that it was not necessary to require that the entrepreneur invest their entire net worth into the firm, to allowing the entrepreneurs to receive the greatest attention. In other words, the agents in the support infrastructure changed the environment to be more favorable to their practice. Institutions and routines developed in the Bay Area ensure that the region can attract the entrepreneurs of the future. For example, its global-class universities and corporate research laboratories continue to attract many of the best and brightest students, researchers, and faculty members. The many very successful high-technology firms, nearly all of which still tout their entrepreneurial origins, attract thousands more engineers and managers some of whom will become entrepreneurs and still more are willing to join start-ups. These individuals are then placed into a munificent environment that values and even glorifies entrepreneurship and, very importantly, places the resources for attempting a start-up within reach. It is a little wonder that new things emerge and attract seed funding in an environment where venture capitalists and a large community of angel investors are willing to invest to explore their business potential. The ability of this process to discover the New Thing is remarkable. Promising technologies receive resources, both managerial and financial, for experimentation. Those technologies that show evidence of yielding significant capital gains encourage other entrepreneurs to launch firms that attract yet more VC. Successful exits can precipitate full-scale investment manias. In contrast, some technologies do not lend themselves to large gains and they are soon dropped as having no promise for this particular methodology of supporting innovation. For example, Thomas Murtha, Lenway, and Hart (2001) show Silicon Valley was the home to a number of flat panel display screen firms funded by venture capitalists, but quickly the venture capitalists came to understand that the industry provided few suitable investment opportunities and abandoned the field. Other industries, such as personal computers, superconductivity, and soon, perhaps, nanotechnology, received VC investment initially, but were later abandoned. An entrepreneur with a new business model need only convince a few venture capitalists to gamble. Moreover, in contrast to personal investors, by the nature of the limited partnership format, the venture capitalists must invest or they cannot continue in the business. This means that they

59


are at least willing to listen to high-risk proposals and also willing to invest in high-risk situations. For example, venture capitalists-funded firms established to exploit Linux, such as Redhat and VA Linux, despite the fact that the business models were predicated upon unseating the Microsoft monopoly through the use of a free-operating system. In other words, the infrastructural agents can support very high-risk projects as long as they have a commensurate potential for enormous rewards. Our evolutionary treatment of organizations and technologies presents Silicon Valley as a complex tapestry replete with commensurate coevolution within which both have shaped each other and created routines and a cultural gestalt that is self-reinforcing. The organizations in the support infrastructure function as an initial selection mechanism. Firms without the perceived requisite potential for outside capital gains are not funded, while ideas that appear to be sound—by the standards of the support network—receive funding, thereby perpetuating their survival. In this ecosystem, actor incentives, technological trajectories, and business models have coevolved and become mutually reinforcing.

Notes 1. The San Jose Laboratory would pioneer magnetic data storage media and became IBM’s global center of excellence for magnetic media. Much later many key Silicon Valley disk drive entrepreneurs spun out of IBM to create new disk drive firms (McKendrick et al. 2000a). Later, IBM’s laboratory was an important source of the relational database technology that firms such as Oracle and Sybase commercialized in the mid-1980s. 2. The historic record suggests that in Boston there were very few angels, and this, in fact, is cited as one of the reasons for forming the first formal VC firm American Research and Development (see Hsu and Kenney 2005). 3. The classic citations on industry or product life-cycle theories are Abernathy (1978), Abernathy and Utterback (1978), or, most recently, Klepper (1996). 4. Bill Gates also was far more closely related to this hobbyist stream than he was to PARC. 5. This section is largely drawn from von Burg (2001). 6. The three-year compounded average annual return was a more modest 47.9 percent! 7. This may be changing as they use PC-like computers that are harnessed by sophisticated graphics software.

60

4 Accounting for Emergence and Novelty in Boston and Bay Area Biotechnology* Jason Owen-Smith and Walter W. Powell

All happy families resemble one another, each unhappy family is unhappy in its own way. Leo Tolstoy

Existing studies of innovation and clusters all too often begin with Tolstoy’s insight that successful regions resemble one another, while those that falter do so idiosyncratically. This conjecture, however, is rarely an a priori proposition. Instead, we suggest it is a substantive artifact of reliance on methods that emphasize comparative statics over analyses that focus on emergence and dynamics. The San Francisco Bay Area and Cambridge/Boston are the world’s largest and most commercially successful biotechnology regions. The attributes and successes of these regions are widely studied and their efforts broadly emulated (Powell et al. 2007). Despite similarities in scale and outcomes, however, each region emerged through a distinctive process that continues to influence its outputs. These variations, in turn, suggest that there are multiple pathways to similar outcomes and offer a corrective to efforts to transpose a ‘standard’ model of regional innovative success that may never have existed. Drawing upon a data-set that tracks strategic alliance networks in human therapeutic and diagnostic biotechnology over a twelve-year period (1988–99), we examine patterns in the development of two canonically successful biotechnology clusters in the Boston/Cambridge Massachusetts metropolitan region and the San Francisco Bay Area. We emphasize the extent to which interesting variations in the form and substance of

61

Accounting for Emergence and Novelty

innovative activity are apparent when viewed with a dynamic lens. Broad similarities in ascendant clusters, we contend, can be outcomes of divergent patterns of development. Moreover, we suggest that these patterned variations can shape the nature of innovations produced by firms.

Regional Advantage and Industrial Development in Biotechnology1 We focus on the commercial field of biotechnology, which developed scientifically in university labs in the 1970s, saw the founding of hundreds of small science-based firms in the 1980s, and matured in the 1990s with the release of dozens of new therapeutics. The field is notable for both its scientific and commercial advances as well as for the diverse cast of organizational players—universities and other Public Research Organizations (PROs), government laboratories, Venture Capital (VC) firms, large multinational pharmaceutical corporations, and smaller dedicated biotechnology firms (DBFs)—involved in its development. In this field, where the sources of scientific and technical leadership are widely dispersed and rapidly developing, and where the relevant skills and resources necessary to produce new medicines are scattered, collaboration among organizations became a necessary component of success. An elaborate system of private governance emerged to orchestrate the interorganizational networks such collaborations constituted (Powell 1990, 1996) and the internal structures and practices of DBFs changed accordingly as firms co-evolved with the networks that characterize the industry. During the very early years of the industry, from the early 1970s to the late 1980s, most biotech firms were very small start-ups that relied, of necessity, on external support. Lacking the skills and resources needed to bring new innovations to market, they became involved in elaborate lattices of relationships with universities and large pharmaceutical firms (Kenney 1986; Powell and Brantley 1992). Lacking a knowledge base in the new scientific field of molecular biology, large companies were drawn to start-ups by the latter’s capabilities in basic and translational science (Galambos and Sturchio 1996; Gambardella 1995). Asymmetries in technological, regulatory, and financial muscle drove early collaborative patterns in the industry (Hagedoorn and Roijakkers 2002; McKelvey 1996; Orsenigo 1989; Orsenigo, Pammolli, and Riccaboni 2001). Despite arguments that the new field would undergo a shakeout as large pharmaceutical firms developed the technical competencies that would

62


allow them to assert dominance over weaker small firm partners (Sharpe 1991; Teece 1986), the founding of new firms accelerated. Established firms’ efforts to cherry-pick promising new ventures faced significant obstacles imposed by deeply collaborative R&D efforts and a mobile scientific labor force. Instead of consolidation and shakeout, the industry’s later years witnessed the give-and-take and mutual forbearance characteristic of relational contracting (Macneil 1978), which became the dominant practice in the field. By the late 1980s, several biotech firms (e.g. Biogen, Genzyme, Chiron, Genentech, Amgen, and Immunex) had become large organizations and numerous pharmaceutical firms had created in-house molecular biology research programs. Even when mutual need declined as a spur to collaboration, the pattern of dense interconnection deepened, suggesting that the original motivation of exchanging complementary resources had shifted to a broader focus on mining innovation networks to explore new forms of collaboration and product development (Powell et al. 2005). An analytic story that places networks alone at the heart of biotechnology’s development misses an important component of the analysis, however. Despite the evolution of dense and expansive networks, geography played an essential role in the industry’s evolution and remains an important feature even today. The networks that now characterize this complex commercial field emerged from distinct geographic roots. Beginning in the Bay Area and Boston, then spreading to other areas, such as San Diego, Seattle, and Bethesda, MD, clusters of biotech firms, VC firms, and PROs forged local networks that reached out as they developed, creating a national industry network from regional origins (Owen-Smith et al. 2002). Yet these regions remain important to understanding conditions in the industry. Evidence is mounting that the network effects that drive much of the action in biotechnology vary with the geographic location of partners (Owen-Smith and Powell 2004; Whittington, OwenSmith, and Powell 2006). Networks played an essential role in the development of stable regional clusters, but those clusters seeded the geographically dispersed structures that have come to characterize the field. We thus turn to analyses of network connections in the two largest and most successful US biotechnology regions in order to demonstrate that collaborative arrangements help to underpin successful clusters. Those regional communities vary in their character, evolutionary path, and approach to innovation. We draw upon a data-set of strategic alliance ties involving 482 DBFs and their more than 2,000 partner organizations over the period 1988–99 to

63


illuminate patterns in the structures connecting regionally colocated biotechnology firms. Data are drawn from Bioscan, an independent industry directory published quarterly. We focus on independently operated, profit-seeking entities involved in human therapeutic and diagnostic applications of biotechnology—but omit companies involved in agricultural and veterinary applications as those sectors draw on different scientific capabilities and operate in different regulatory environments. Our data-set, like the industry it represents, is dominated by US firms, although recent years have seen considerable expansion in Europe. The sample of firms includes both public and privately held firms, and the former include companies with minority or majority investments by other firms as long as their stock is independently traded. Large pharmaceutical companies, investors, government agencies, and PROs enter the data-set as partners that collaborate with biotech firms. We link these relational data to patent grant and citation information for the period 1976–99 drawn from the National Bureau of Economic Research patent citation database (Jaffe and Trajtenberg 2002). In total, there are 10,067 US utility patents issued to the 482 firms in our sample over this time period. Organizations are identified by type and location, which enables us to isolate ties among colocated organizations in two established biotechnology regions. The San Francisco Bay Area and Boston are well-studied examples of densely connected and intensely innovative regional economies. In our data-set, Boston is home to more than 14 percent of US firms in our sample. Bay Area biotechnology firms account for almost 21 percent of US firms. Together, these regions were issued 51.5 percent of the patents assigned to US biotechnology firms through 1999 and developed 32 percent of all biological therapeutics approved by the US Food and Drug (FDA) administration between 1988 and May 2004.2 Five of the ten best-selling biotechnology drugs in 2001 were developed by firms in these two regions. Boston and the Bay Area thus represent notable success cases for biotechnology regions.

The Bay Area and Boston Networks In order to examine the evolution of the Bay Area and Boston networks, we identify all organizations located in the two regions that have contracted with a local biotechnology firm. Our data-set includes four types of formal interorganizational connections and five types of organizations. In addition to biotech firms, we include VC firms, government agencies, large

64


multinational pharmaceutical corporations, and PROs in the partner sample. These diverse organizational forms are connected by four varieties of contractual ties. R&D connections represent agreements for shared research and development efforts. Finance ties reflect investments in one organization by another. Licensing ties are agreements that transfer the rights to intellectual property across organizations. Commercialization ties include downstream product development activities, ranging from clinical trials to manufacturing, sales, and marketing. During the period 1988–99, the Bay Area network is the larger of the two, involving 159 organizations (82 biotech firms, 12 PROs—most notably Stanford University and the Universities of California at Berkeley and San Francisco, one government laboratory—Lawrence Livermore Labs, and some 64 VC firms), connected by 243 local contractual ties. The Boston network is home to 113 organizations (57 biotechs, 19 PROS— including MIT, Harvard University, Massachusetts General Hospital, and the Dana Farber Cancer Center, and 37 VC firms), connected by 201 local contractual ties. Neither region was home to a multinational pharmaceutical corporation during this time period.3 While the regions differ in scale, in the demography of organizational types that occupy them, and in the availability of local VC funds (Powell et al. 2002), both are characterized by organizationally diverse and structurally cohesive networks. How, then, do the regions differ? Figures 4.1 and 4.2 track yearly changes in Boston and the Bay Area in terms of the distribution of dyads that comprise each region’s main network component. The main component of a network is its largest connected subset. In practical terms, the main component represents the largest group of organizations in a structure that can reach one another through network paths of finite length and thus captures the minimal level of connectivity necessary to enable broad information diffusion (Owen-Smith and Powell 2004). Put colloquially, imagine drawing linkages among nodes without ever lifting your pen. These figures paint a very different evolutionary trajectory for the two regions. The most basic unit of a network is the dyad. In this case, a dyad is a pair of organizations connected by a formal R&D, finance, licensing, or commercialization tie. Figures 4.1 and 4.2 characterize dyads in terms of the types of organizations that comprise them, without regard to the class of activity connecting a given pair. Three types of dyads are possible in these main components: Biotech firms can connect with each other (a DBF–DBF dyad), with PROs such as universities or hospitals (a DBF–PRO dyad), or

65

Accounting for Emergence and Novelty 90 DBF VC DBF PRO DBF DBF

80 70 60 50 40 30 20 10 0 1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

Figure 4.1. Boston main component ties by dyads and year, 1988–99

120 DBF VC DBF PRO DBF DBF

100

80

60

40

20

0

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

Figure 4.2. Bay Area main component ties by dyads and year, 1988–99

66

1999


with VC firms (a DBF–VC dyad).4 In addition to the distribution of dyads, regional networks grow at different paces and in different patterns. Consider Figure 4.1, which tracks the growth of the Boston network. Note first the pattern of growth in ties (each dyad represents a single tie) implied by the height of the histogram bars. Our data-set begins in 1988, when we find a relatively large number of ties in Boston. The number grows slightly into the early nineties and then levels off for several years before climbing again through the latter years of the data-set. The bars are shaded to represent the relative prevalence of different dyads in the network. Note the bar that represents 1988, which shows a remarkable reliance on PROs. Only a very small number of ties link biotech firms to each other or to local VC firms in 1988 or 1989. These ties grow as the network expands and these commercial connections dominate the network by the end of our time frame. The Boston network grew from origins in the public sector (Porter, Whittington, and Powell, 2005). Put differently, public science formed the foundation for commercial application (Nelson 1981, 1986). Industries where commercially viable technical advances emerge from the academic and public sector manifest more open technological trajectories than industries that rely more heavily on industrial R&D (Dosi 1982). The Boston biotechnology community is linked by shared connections to PROs early in its evolution. These connections remain an important part of the network, but increasing patterns of DBF to DBF and DBF to VC ties reflect the development of a commercial network that becomes structurally autonomous, while bearing the imprint of the public sector. Contrast this trajectory with the different pattern illustrated by Figure 4.2. There is no dominant network component in the Bay Area in 1988, though a cohesive network forms in 1989. Unlike Boston’s growth pattern, which saw a plateau in the early 1990s, the Bay Area grew markedly through 1996 before stabilizing in the late 1990s. These differences in volume and velocity are matched by very different dyad distributions. During the first two years when a main component existed, the Bay Area community was composed entirely of ties linking DBFs to local VC firms. Where the stability and technical diversity of Boston PROs anchored that network and fostered a more open technological trajectory (Owen-Smith and Powell 2004), the Bay Area relied heavily on the prospecting and matchmaking efforts of venture investors.5 Later years witnessed the increasing importance of VCs, a smattering of ties involving PROs, and— most importantly—dramatic growth in DBF–DBF connections. By 1999, direct links among Bay Area DBFs outweigh the other two types of dyads.

67


Both Boston and the San Francisco Bay Area evolved from dependence on a non-DBF organizational form to a state where significant portions of the network were made coherent by direct connections among sciencebased biotechnology firms. In other words, similar endpoints in the evolution of the networks were reached through different routes. While both relied on the inclusion of organizations different from biotechnology firms, Boston was anchored in the public sector, whereas the Bay Area was dominated by venture capitalists. The endpoints of these trajectories are similar as both regions came to depend heavily on collaborations among ostensible competitors, but their different starting points and the lasting involvement of different partners may have produced distinctive patterns of innovation. Distributions of dyads, however, cannot tell the full story of a network’s evolution; hence, we turn to an assessment of the overall topology of the networks. Figure 4.3 fleshes out differences across the regions with images of the networks in three distinct time periods. These snapshots were generated using Pajek,6 a freeware program designed for the visualization and analysis of large networks. The relative positions of nodes in these images are meaningful and result from two spring-embedded, graph-drawing algorithms. The first treats a network as a physical system where nodes repel each other and ties act as ‘springs’ that pull connected nodes closer together (Fruchterman and Reingold 1991). This algorithm moves unconnected nodes to the periphery of the image, and separates components (groups of two or more nodes) from one another. The second algorithm relocates connected nodes so that the Euclidean distances among them are proportional to their graph theoretic distance (Kamada and Kawai 1989).7 These images, then, are replicable representations where the relative position of organizations is a function of the connectivity of the system and the degrees of separation among nodes.8 The shapes of the nodes in Figure 4.3 represent different types of organizations: circular nodes are DBFs, diamonds are PROs, and triangles are VC firms. Tie patterns likewise represent different types of collaborations. Solid black links are R&D ties, dotted black connections represent financial investments, dotted gray are licenses, and solid gray linkages indicate commercialization deals. The width of a given tie reflects the number of connections linking a pair of partners. When multiple ties are present in a dyad, the pattern of the linkage reflects the most recent type of activity. To gain purchase on the differences between the two networks at a given point in time, read across columns in Figure 4.3. To get a sense of the evolutionary pattern within each region, pick a column and read down.

68


San Francisco Bay Area

Boston

1988

1994

1999

Figure 4.3. Boston and Bay Area networks: 1988, 1994, and 1999

69


Consider the first image of the Boston region, which bears out Figure 4.1’s emphasis on ties between DBFs and PROs. Note the central role played by diamond nodes and the relatively few connections linking circles to each other. The main component of this network stretches across the center of the image; most of the ties are single connections between different types of organizations. Tie types are rather evenly dispersed, but gray (commercialization) and dotted gray (licensing) ties are in the majority. In the region’s early years, DBF firms are stitched together into a coherent network by their shared connections to PROs. Harvard, the well-connected diamond in the lower right-hand corner of the image, and MIT, the center of the star in the upper left-hand quadrant, are the primary entry points to the network and their relative distance from each other suggests that the early years of the Boston biotechnology community may have been shaped by different kinds of academic involvement. In related work, we show that while both Harvard and MIT faculty have been active in founding Boston-based biotechnology firms, MIT scientists are much more active patentors, while Harvard faculty serve on more scientific advisory boards (Porter et al. 2005). The 1988 Bay Area image paints a very different picture as ties in that year did not aggregate to create a dominant component. Instead, the early years of the region appear to be characterized by small clusters of firms connected either to multiple venture capitalists or, less commonly, to Stanford University (the diamond at the center of the three node ‘chain’ near the bottom of the image) or University of California, San Francisco (UCSF) (the diamond in the dyad—with Genentech, whose founder was a UCSF scientist—near the center of the figure). The patterns suggested by the visualizations are echoed by careful archival research. In an analysis of the career histories of the founders of biotech firms in the Bay Area and Boston, Porter (2004) finds that Boston companies were often started by MIT and Harvard professors, many of whom maintained their university affiliations. In contrast, founders in the Bay Area were much more likely to come from VC or other biotech firms. Another key contrast was that Boston faculty founders were often senior professors with established reputations. When Bay Area faculty were involved in founding, they tended to be younger and much more likely to take a leave from their university positions. Almost all founders in Boston came from the region, while founders in the Bay Area came from diverse locales. Indeed, east coast faculty—from Yale, Columbia, and Duke—came to California to start companies. Turn your attention now to the second row of Figure 4.3, which represents the regional networks near the middle of our time series in 1994. The

70


Boston network has grown, but maintains a reliance on PROs and both Harvard (now at the bottom of the image) and MIT (the well-connected diamond node at the top of the image) remain important players in the network. This image also suggests the growing importance of local venture capitalists (notably in the ‘tree’ structures that descend from Harvard’s partners in the lower quadrant of the image) as well as the salience of DBF to DBF ties. Note the cohesive cluster (outlined by a dashed circle) formed by R&D connections among four Boston firms—Genzyme, Genzyme Transgenics (one of several spin-offs from Genzyme), Autoimmune (whose research tie to the Dana Farber Cancer keeps this nascent cluster attached to the larger network), and Creative Biomolecule. These firms, particularly Genzyme—a large and successful ‘first generation’ biotechnology company—seed the development of a dense DBF–DBF region in the Boston network. Compare this view to the image of the Bay Area, whose large main component reflects the dramatic pattern of growth captured in Figure 4.2. This image is dominated by ties linking DBFs (circles) to venture capitalists (triangles). PROs play a minimal role in the network. The single diamond node at the top left is Stanford, which is linked to a young biotechnology firm by a license.9 Note, however, the robust cluster of (often multiple) DBF to DBF ties outlined by a dashed circle at the top of the figure. This group, centered on Genentech—one of the first and most successful biotechnology firms—and Chiron—another large and established player in the industry—is characterized by diverse and repeated ties that directly link biotech firms to one another. Boston’s Genzyme triangle and the Bay Area’s Genentech cluster represent the beginnings of a network centered on collaborations among ostensible competitors. The size of these ties relative to those connecting younger DBFs to venture captital firms suggests a process by which newcomers are identified by investors and then linked into the DBF–DBF segment of the network by forging ties with incumbents or their partners. Where PROs are still the entry way and the gatekeepers of Boston in 1994, Bay Area VCs prospect for new talent, and established firms usher promising newcomers into an increasingly connected segment of the network. These patterns became more robust in 1999, the final year of our data. In both networks the pattern of DBF–DBF linkages expands and deepens (relevant regions are outlined by dashed ellipses) and both sections of the network remain centered on the region’s largest and most successful firms (Biogen and Genzyme in Boston and Genentech and Chiron in the Bay Area). Despite the clear emergence of purely commercial portions of

71


both networks, however, the regions still manifest significant differences. While VC firms are important players in the Bay Area network, they only rarely form multiple connections to the same partner and (as one would expect) their ties are overwhelmingly financial. The field of dotted gray lines in the lower left quadrant of the 1999 Bay Area image exemplifies this trend. VCs play an important connective and prospecting role in this network, but their one-dimensional network portfolios suggest that much of the innovative action may emerge from the dense and multiplex cluster of biotech to biotech ties. The linkages formed by PROs shrink in importance in the Boston network in 1999 (recall Figure 4.1). This distributional decline, however, masks the continued importance of these research-oriented public sector organizations. Universities, nonprofit research institutes, and hospitals forge repeated (and multiplex) ties to biotechnology firms (evidenced by the thicker connections linking diamonds to circles), and thus play a very different role in the region than do VC firms (the triangles that dominate a small portion of the Boston graph in the lower left segment of the image). Boston PROs remain important structural components of the network, however. Note the diamond nodes at the center of the image (just to the right of the dotted ellipse) that represent MIT and Massachusetts General Hospital. Both regions developed tightly interconnected DBF–DBF commercial networks, but they did so from different starting points and with very divergent types and levels of involvement from non-DBF partners. Though similar on many dimensions, we suggest that these disparate evolutionary trajectories have enduring effects on the nature of innovation in these regions.

The Form and Substance of Regional Innovation How do varied starting points and evolutionary trajectories leave lasting imprints on regional innovation patterns? We contend that the networks more clearly dominated by ‘open’ public sector organizations will result in innovations that rely less heavily on internal R&D and that draw more on research conducted in organizations other than biotechnology firms. In short, we expect patents assigned to Bay Area DBFs, a region whose network was always based more on commercial firms, to cite proportionally less non-DBF prior art and to rely more heavily on self-citations than do patents assigned to Boston firms.

72


We turn to data on citations made by patents assigned to Boston and Bay Area DBFs to examine how regional effects may shape the process of innovation. We begin by presenting information on the R&D outputs of regional firms in the aggregate from 1988–99. We then turn to consideration of shifting patterns in prior art citations by DBF patents. Next we consider the substance of regional innovation by assessing differences in rates of FDA approvals as well as variation on Orphan Drug Indications10 by region. Finally, we compare the patented innovations underpinning two comparable treatments for multiple sclerosis: Cambridge-based Biogen’s Avonex and Emeryville-based Chiron’s Betaseron. The differences in regional scale that we identify are matched by differences in the volume of innovation. Table 4.1 presents a comparison of R&D outputs by region for the period 1988–99. The 82 Bay Area DBFs in our sample generated some 3,800 US utility patents in this time period, which is an average of slightly more than 46 patents per firm. This contrasts dramatically with Boston DBFs’ average of slightly more than twenty-four patents per firm. In contrast, biotechnology firms located outside these two regions produced only slightly more than fourteen patents on average, suggesting the relative fecundity of both Bay Area and Boston DBFs. These output differences also mask a highly skewed distribution of patents within regions. Bay Area outputs are more stratified than those in Boston. The five most prolific Bay Area patentors account for 63 percent of regional patents, while the top five Boston patentors were issued 42 percent of the region’s patents. Despite these patterns, patents assigned to regional firms had very similar citation impact. Two-tailed t-tests discerned no significant difference between the impact of Bay Area and

Table 4.1. R&D outputs by region, 1988–99

Number of DBFs Number of patents Mean citations received (standardized) Variance in citations received Number of citations made Percentage non-DBF cites Percentage self-cites FDA approved therapeutics Orphan indications and products

Boston

Bay Area

Other locations

57 1,376 1.113 30.270 12,659 71% 12% 18 60

82 3,806 0.979 14.150 41,389 55% 35% 40 51

1,343 4,876 0.944 14.493 43,610 68% 11% 89 109

73


Boston DBF patents (t ¼ 0.774, p ¼ 0.439), but did suggest that firms in these regions develop higher impact intellectual property than those located elsewhere (t ¼ 3.837, p < 0.0001).11 The similar impact of Boston and Bay Area innovations masks broad differences in the distribution of highly cited patents within the regions. Patents assigned to Boston firms manifest a much higher variance in forward citations than do Bay Area patents, suggesting that Boston firms may more routinely engage in ‘exploratory’ innovative search, which typically yields a few very high impact patents at the expense of numerous innovations with lower than average future effects (Fleming and Sorenson 2001; Levitt and March 1988). On this view, the Bay Area’s lower citation variance is indicative of a more directed and incremental, exploitative strategy, which is what one might expect of firms that are supported by investor networks that are interested in demonstrable progress. Firms that pursue exploitative strategies generally develop numerous related improvements on established components of their research trajectories. Such incremental innovations are less valuable on average than the riskier outcomes of more broad-ranging innovation efforts, but convey important benefits in terms of overlapping ownership rights. Exploitative patents, then, will have lesser variance in their impacts than will patents that result from more exploratory efforts to develop blockbuster technologies. While impact variations suggest different patterns of search in innovation, prior art citations provide more direct insight into the precursors that firms rely on in developing new intellectual property. Such ‘backward’ citation data allow us to expand upon the relationship between regional networks and innovation in biotechnology. Consider two ideal-typical possibilities: First, firms embedded in networks composed largely of competitors and investors are primarily concerned with speed and with commercial development, hence they pursue a more focused innovation process that relies heavily on internal R&D and attention to the efforts of direct competitors (e.g. other DBFs). As they are situated in structures that lack a significant PRO involvement, such firms may be less likely to rely heavily on innovations developed externally. In contrast, firms that are located in networks anchored by PROs and that lack strong investor involvement may feel somewhat less overt pressure to pursue immediate commercial returns. To the extent that open, public sector research organizations alter the norms that govern information flow within a network, firms in such networks may reach more freely across organizational boundaries in efforts to develop new innovations and their patent citations may evince

74


less attention to the research efforts of competitors (Owen-Smith and Powell 2004). Again, we stress that these different patterns may reflect divergent time scales. Were we to have full data on Bay Area firms from the 1970s, we might well find patterns of relationships comparable to Boston in the 1980s and 1990s. We cannot rule out the possibility that these regional differences stem from the earlier start and success of Bay Area firms in bringing new medicines to market. Moreover, in their early years, several notable Boston-based firms opted to license their earliest lead products to large pharmaceutical companies in return for royalty payments (Robbins-Roth 2001). If these two conjectures have validity, we would expect innovations by firms in more overtly commercial networks—such as those in the Bay Area—to rely less heavily on prior art developed by organizations other than DBFs and to rely more strongly on citations to their own prior patents. In contrast, innovations made by firms situated in more open networks dominated by academic and public sector organizations—such as those in Boston—will rely on a broader cross-section of prior art sources and less extensively on internal R&D. Table 4.1 provides descriptive support for these claims. The 1,376 Boston inventions make 12,659 citations to prior US patents, while the 3,806 Bay Area patents acknowledge 41,389 links to prior art. (An average of 9.2 cites per patent in Boston and 10.9 cites per patent in the Bay Area. Firms outside these regions cite just under nine pieces of prior art per patent.) Similar levels of reliance on prior art, though, mask significant variation in terms of the sources from which precursors are drawn. Patents assigned to Boston firms rely more heavily on non-DBF prior art—a full 71 percent of citations—than do either Bay Area patents, which make 55 percent of their citations to non-DBF prior art, or nonregional patents. In contrast, slightly more than one-third (35 percent) of citations made by Bay Area patents are to their own prior art.12 Boston firms do cite their own prior art, but at a much lower (12 percent) rate that more closely accords with the overall trend in the industry. In sum, the R&D portfolios of Bay Area and Boston firms rely on quite different sources of knowledge, and these patterns appear to map onto the structure of the collaborative networks in each region. The form innovation takes, then, is related to the characteristics and trajectories of the networks that support it. While both regions have been quite successful in biotech, and are emulated across the globe, we have shown that their respective origins and paths of development are rather dissimilar. This pattern may continue at the level of market outcomes as

75


well as patents. To explore this possibility, we draw on FDA approval records to identify the fifty-eight new drugs developed by Boston and Bay Area DBFs. Fifty-three of those medicines were approved between 1988 and 2004. All five of the drugs that appeared on the market prior to that period were developed by two Bay Area Firms—Alza and Genentech. Again, these early approvals reflect the commercialization strategy pursued in a region with a strong VC community. Eighteen of these products are the work of Boston firms and forty stem from work by Bay Area DBFs. Another 89 therapeutics were developed by the 343 firms located outside these regions, but well-established firms such as Los Angeles’ Amgen, Philadelphia’s Centocor, and Seattle’s Immunex account for much of the action. In terms of market outcomes, the Bay Area appears to be both quicker and more prolific than Boston and both regions represent concentrations of success. This outcome is to be expected given a more commercially focused network and a development-oriented strategy that relies heavily on internal R&D. Indeed, seventeen of the first twenty of these drugs to come to market were produced by Bay Area firms. These differences in market outcomes, though, are much more suggestive of variations in strategy and focus than competency. Consider another source of information about the development of therapeutics, Orphan drug designations. The 1983 Orphan Drug Act was designed to enable the FDA to speed the development of therapies for rare diseases, and orphan designations offer tax breaks and regulatory assistance to organizations that develop such medicines. One hundred and eleven (111) orphan designations have been approved for Bay Area and Boston firms since 1985 (when the first such approval went to Boston’s Genzyme for the drug Ceredase for patients with Gaucher’s disease). Both Bay Area and Boston firms make use of orphan designations, but Boston firms, as one might expect for companies enmeshed in networks dominated by universities and hospitals, rely more heavily on indications for relatively rare diseases. The focus on orphan drugs reflects another difference as well. The Boston-based firms build their product portfolios with an initial focus on smaller markets and medicines that have the added security of orphan drug exclusivity. These medicines, while targeted at relatively small populations, are very much desired by their patient communities. In contrast, Bay Area firms favor medicines for larger markets in which the potential patient population runs in the tens of millions, and for which there is likely to be product competition from other DBFs and major pharmaceutical corporations. This high-risk, high-reward strategy

76


demands speed in product development, and shows the obvious imprint of the VC mindset. Descriptive patterns in prior art citations, forward citation impact, and market outcomes are complementary with observed variations in the evolution of the two regional networks. Despite sharing an industry and scientific base, Boston and Bay Area DBFs appear to differ systematically in the substantive focus of their R&D efforts. To further explore these differences in the focus of innovative activity, we turn to a natural experiment and compare the citation patterns for patents underlying two fairly similar biotechnology drugs. Betaseron and Avonex are competing therapies for remitting and relapsing multiple sclerosis and several clinical trials have directly compared their efficacy. Both drugs began life with orphan designations and both are variants of the biological compound interferon-beta, which differ only slightly in chemical makeup.13 The processes by which these compounds are produced are also very similar, and rely on Chinese hamster ovaries, though their differences are manifest enough that an infringement lawsuit between Avonex’s developer and Betaseron’s manufacturer (Biogen vs. Berlex Laboratories) resulted in a judgment of no infringement. Both drugs were approved during the 1990s (Betaseron in 1993 and Avonex in 1996). In short, these two drugs share notable scientific, clinical, and regulatory similarities, but they differ in the physical and organizational location of their development. Betaseron is based on research done by Cetus, an Emeryville, CA biotech firm that was acquired by Chiron, a Berkeley-based DBF. Chiron did the development work on Betaseron and shepherded the drug through the FDA approval process. Betaseron is manufactured and marketed under an arrangement with Berlex Laboratories, an American subsidiary of the pharmaceutical firm Schering-Plough. Avonex, in contrast is based on research done by Boston-based Biogen who also developed, manufactures and markets the drug. We use FDA-labeling information to identify the patents that underpin these drugs. We then turned to the NBER patent citation database to trace prior art citations by those patents and identify the sources of such prior art. In both instances we trace precursor inventions to three generations. Table 4.2 presents summary data for the innovations underlying these two drugs. The patterns suggested by Table 4.2 are in line with the overall results in prior art citations by region, and with our expectations based on the evolution of each cluster’s network. Betaseron relies on a set of four related patents initially assigned to Cetus (three were reassigned to Chiron

77

Accounting for Emergence and Novelty Table 4.2. Innovation data for Betaseron and Avonex, 1988–99 FDA approval date (initial indication)

1993-07-23

1996-05-17

Orphan status Developer Initial patent holder Distributor

Y Chiron Cetus Corporation Berlex Laboratories (Schering Plough) Remitting and relapsing Multiple sclerosis 4 4

Y Biogen Biogen Biogen

Initial indication Number core patents Number first generation citations Number second generation citations Number third generation citations Total patents (original þ 3 generations) Number of prior art patents owned by fiduciary firms Number of prior art patents from same region Number of non-US prior art patents Number of shared prior art patents

Remitting and relapsing Multiple sclerosis 1 14

31

32

16

108

55

155

6

0

2

4

26

61

39

39

Notes: Both intra-region cites for Betaseron are to other Silicon Valley DBFs (Genentech, ICN). The within-region cites for Avonex are to PROs (Mass Gen (1) MIT(2)) and a non-DBF firm (Ionics).

following the merger of the two firms; the fourth, a process patent for producing interferon, was reassigned to Berlex Labs). These four patents cite a small group of prior art patents (4). These four ‘first-generation’ precursors make another thirty-one second generation citations, which in turn cite a further sixteen pieces of prior art. All told, Betaseron rests on a history of some 55 interlocking patents. Avonex, which is based on a single compound patent, reaches more broadly into the prior art, relying on 155 separate pieces of intellectual property. None of the prior art on which Avonex depends is owned by Biogen. This last finding is particularly telling, as it suggests that Biogen developed its market leading therapeutic without the benefit of a thicket of intellectual property rights (IPRs), relying instead on a mix of partner’s intellectual property and public domain science. Differences in these two citation networks are instructive. Betaseron’s underlying IP network includes six patents developed by Cetus. Avonex, in contrast, relies on a single Biogen-owned patent that makes no citations to other intellectual property owned by that firm. While internal R&D was

78


surely not sufficient to the development of Betaseron, that drug relied much more heavily on a single DBF’s research effort than did its competitor. Both innovation networks reach well beyond the regions in which the two firms are situated. Betaseron cites only two patents held by other Bay Area organizations, but it is notable that both are biotechnology firms (Genentech and ICN). The Avonex citation network, in contrast, cites four patents held by Boston organizations, but none by DBFs. Three belong to PROs with whom Biogen has network ties (MIT holds two patents and the Massachusetts General Hospital a third). The fourth belongs to a non-DBF firm, a purification company called Ionics. While the citation networks are fairly small, comparing these two very similar drugs offers a natural experiment that holds constant important technical, clinical, and regulatory features of biotechnology innovations. Even when such factors are very similar, the patent citation networks underlying these two drugs differ in a fashion that reflects aggregate differences in regional innovation patterns and expectations based on the interorganizational networks that characterize each region. The Bay Area–based drug relies more heavily on internal R&D and on the research efforts of other firms. In contrast, the Boston-based therapy draws on a broader cross-section of prior IP owned by a wider range of organizational types.

Conclusions and Implications The Boston and Bay Area biotech communities became more similar over the twelve-year period under examination, shedding their respective reliance on PROS and VCs, and developing a strong firm-to-firm component. But these divergent roots have a notable impact on the innovation process. Boston-based companies that relied heavily on external sources of knowledge favored more exploratory efforts at discovery. This signature is captured by our measures of patent volume and impact, and by patterns of patent citations. Bay Area biotech firms were more self-reliant in terms of knowledge generation and more persistent in their efforts to further development of in-house intellectual property. Similarly, Bay Area firms were faster and more prolific in terms of new product development, as well as more likely to pursue novel medicines for larger markets where they might face stiff competition. In contrast, Boston firms were more deliberative in their commercial strategies and more likely to focus on medicines for identifiable and active patient populations

79


in need of relief from specific illnesses. Most remarkably, these differences persisted even when we held constant market, scientific, and regulatory factors by examining Chiron and Biogen’s approaches to the development of similar treatments for multiple sclerosis. Clearly, the continuing impact of VCs in the Bay Area and PROs in Boston is significant. We lack data on the early scientific roots of technical advance in the life sciences in Boston and the Bay Area. Perhaps the patterns we have observed are the outcroppings of diverse academic approaches to scientific research in the life sciences. Boston is home to the remarkable institutional combination of MIT, a powerful basic science institution that lacks a medical school, Harvard, another powerhouse institution in basic science whose medical school is located across the Charles River at a considerable remove from the main campus, and a number of research-oriented hospitals and institutes. The upshot of this institutional mix appears to us to be a corporate focus on expansive science and patients. In contrast, the biotech community in the Bay Area has its earliest origins in the ‘marriage’ of Herbert Boyer, a UCSF scientist, and Robert Swanson, a prominent venture capitalist, who joined together to create Genentech, one of the very first biotech companies. UCSF is an unusual institution, lacking disciplinary departments and a full panoply of research program and students. The organizational model at UCSF was an interdisciplinary, cross-functional approach to medicine, with an emphasis on translating basic science into clinical application (Varmus and Weinberg 1992). Genentech adopted and refined this interdisciplinary ‘team’ model, adding the impatience and restlessness of VC financiers and the attendant focus on ‘swinging for the fences’ by developing products for such major illnesses as heart disease, cancer, and diabetes. Here, an approach to translational R&D pioneered at an elite PRO is transferred by founding scientists to a region’s leading firm, eventually becoming a dominant arrangement for the region. One of the fundamental features distinguishing between the Boston and Bay Area networks is Boston’s early and continuing reliance on the region’s public sector research organizations. MIT, Harvard, Massachusetts General Hospital, the Whitehead Institute, Dana Farber Cancer Center, and other institutions anchored this network, catalyzed its development, and shaped firm innovation and product development. The imprints of evolutionary patterns, then, are a joint function of institutional roles and particular features that characterize such public sector organizations. In addition to providing stable anchors for networks, universities and hospitals contribute to more open information flows, more expansive

80


innovative trajectories, and, possibly, more patient-driven product development strategies. In short, PRO involvement is effective precisely because they operate in different environments and under different rules and constraints than their proprietary partners. In Boston, universities, research institutes, and hospitals—organizations institutionally committed to open information flow, science, and publichealth-based business strategies—altered the efforts of Boston firms by maintaining a formal and deliberate role in their region’s networks. In short, Boston area PROs altered their local networks using formal contractual arrangements that structure collaboration and the transfer of intellectual property rights. In contrast, Stanford and UCSF’s preference for informal, noncontractual ties in their regional network enabled financiers to shape innovative and organizational strategies. The implications are paradoxical: deliberate efforts by hospital and universities to control and shape information and resource flows in networks result in more open and expansive structures, while more informal, ‘hands-off’ approaches help create networks that are more tightly controlled and commercially directed. The role that universities play in regional development, then, appears more complicated than a simple model of technology transfer and technical training would suggest. This coevolutionary dynamic suggests important sources of regional variation and also highlights the potential consequences of organizational action in evolving networks. Firm strategies in Boston and the Bay Area bear the characteristic footprints of their most important partners. At the same time, partners are constrained by the activities of local firms. Consider a brief example. If success in the competitive arena of Bay Area biotechnology depends on early access to venture capital and that access smoothes entry into collaborations with the established firms that are key to innovative and commercial success, then savvy venture capitalists will seek to invest in newcomers whose strategies and arrangements match dominant patterns in the region: the very patterns that VCs’ early efforts helped generate and sustain. Not surprisingly, then, Stuart and Sorenson (2003) find that while VC support became abundant in the Bay Area in the 1990s, the odds for success declined. One possible reason is lock in around a dominant model. The contrast of Boston and the Bay Area, the most prolific biotechnology clusters in the world, should give pause to policymakers who look to successful clusters for models to emulate. Without awareness of underlying institutional variations and distinctive approaches to the development of new medicines, one could easily draw the incorrect inference that

81


combining PROs, VC, and small firms provides the ultimate recipe for successful economic development. We emphasize that similar approaches may be very deceiving and mask sharp differences in underlying causes of institutional and technical development.

Notes * We thank Kjersten Bunker Whittington, Maryann Feldman, Martin Kenney, Luigi Orsenigo, David Lane, Elaine Romanelli, and the participants in the 2003 and 2004 Genesis of Clusters workshops for their comments on earlier drafts. We are grateful for research support from the Hewlett Foundation and the Merck Foundation. 1. This section draws upon our earlier work on biotechnology (cf. Powell et al. 2005; Bunker-Whittington, Owen-Smith, and Powell 2005). 2. We base our calculations on approvals for initial indications by the US Food and Drug Administration’s Center for Biological Evaluation and Research. 3. More recently, Pfizer and Novartis have moved R&D activities to Kendall Square in Cambridge, MA. The largest biotech firm, Amgen, has acquired a smaller Bay Area firm, Tularik, and created a beachhead in that region. 4. Our data are structured as a two-mode network (Wasserman and Faust 1994) that tracks connections among DBFs and between DBFs and partner organizations. Linkages between non-DBF partners (e.g. PRO–VC collaborations) are exceptionally sparse and thus are not included. 5. This different trajectory may reflect left censoring in the data. If we had comparable data for the Bay Area for the late 1970s and early 1980s, we might well observe important DBF–PRO ties. In particular, we would expect more linkages connecting UCSF and Stanford to local DBFs. We do know that the Bay Area biotech community developed earlier than Boston (Robbins-Roth 2001), hence the direct comparison for the later 1980s and early 1990s may capture a slower take-off in Boston. 6. A freeware program developed by Vlado Batagelj and Andre Mrvar and available for download online at http://vlado.fmf.uni-lj.si/pub/networks/pajek/ 7. This distance is a function of the number of ‘steps’ it takes to traverse a network path connecting a given pair of nodes. Organizations that are connected by a tie are a distance one. 8. For more detail on network visualization using Pajek, see the appendix of Powell et al. (2005). 9. To be sure, this isolation does not imply that Stanford is not active in technology transfer, but at this point in time the bulk of its formal licensing activities are to firms outside the region. 10. An Orphan Indication is conveyed by the US Food and Drug Administration for products that treat rare diseases and thus have little potential to become

82

Accounting for Emergence and Novelty huge commercial successes. Orphan indications are valuable to firms in that they convey tax breaks and reduced regulatory fees as well as short-term market exclusivity. 11. Forward citations—citations from future patents to current innovations—are a commonly used measure of impact. We standardize citation counts by year and technical class to avoid heterogeneity across time and technical areas (Jaffe and Trajtenberg 2002; Trajtenberg 1990). 12. Despite a greater reliance on self-citations, 65 percent of Bay Area prior art citations are to patents developed outside the firm. The majority of innovative work goes on through networks. 13. Interferon-beta-1a (Avonex) is a recombinant compound whose amino acid sequence is identical to natural interferon-beta. Interferon-beta-1b (Betaseron) in contrast is a recombinant compound that differs from natural Interferonbeta by one amino acid.

83


Part II Considering the Developing Cluster Context


5 Anatomy of Cluster Development: Emergence and Convergence in the US Human Biotherapeutics, 1976–2003 Elaine Romanelli and Maryann Feldman

Industry clusters are a persistent and pervasive feature in the spatial arrangements of organizations. Little research, however, has explored either the organizational or geographic origins of entrepreneurs and firms who populate an industry cluster. Although some (e.g. Gompers, Lerner, and Scharfstein 2003; Romanelli 1989; Stinchcombe 1986) have considered the characteristics of organizations that may be more or less prolific generators of new organizations, little empirical evidence has been brought to bear on the question. A much larger literature has considered the influence of regional conditions—typically, metropolitan regions—on rates of entrepreneurial growth; however, there are few comparisons of patterns in cluster development among industries (Klepper 2002; Saxenian 1994; Sorenson and Audia 2000). Fundamental questions regarding the demographic process of cluster development thus remain unanswered. Do clusters emerge primarily through the investments of local entrepreneurs, local investors, and other local agents, with some clusters growing large while others languish due to failures of local resources or investment processes? Alternatively, do clusters develop through a geographic convergence of both local and nonlocal entrepreneurs, firms, and investments based on expectations or information about best places for particular kinds of business activity? Finally, assuming that at least some entrepreneurs and firms relocate and at least some venture capitalists invest outside their home regions, what factors elicit their attention and influence their decisions to relocate or invest in a distant region?

87

Anatomy of Cluster Development

In this chapter, we seek to shed light on these questions through an intensive examination of the organizational and geographic origins of entrepreneurs and firms in the US biotherapeutics industry over the period 1976 through 2003. Based on detailed information about the entrepreneurs of 688 biotherapeutics firms (both USA and foreign-owned), we explore how variations in the origins of local entrepreneurs and firms and patterns in cross-regional relocations influence cluster growth and convergence. We present detailed data on the organizational origins of entrepreneurs and firms to show, for regions which have developed relatively large clusters, how differences in the organizational origins over time relate to differences in the growth rates of the clusters and influence the relocations of distant entrepreneurs and firms.

Research Setting Commercial prospects for biotechnology have been widely heralded ever since 1953 when James Watson and Francis Crick ‘correctly theorized the structure and operation of the DNA molecule. This fundamental discovery led to increased research and understanding of the biochemical processes involved in the production and reproduction of life’ (Kenney 1986: 2). For about the next two decades, basic research in the now-called life sciences was restricted to university and government laboratories, along with those of a few large oil and chemical companies, as scientists sought to understand the role of DNA in cellular functioning. These scientists, and their science, formed a critical base of resources for commercial biotechnology in the early 1970s when the US federal government and the National Cancer Institute declared war on cancer, thus providing vast sums of money to support biomedical research. Almost a billion dollars—$989 million to be exact—was allocated to the National Institutes of Health, which was dominated by the National Cancer Institute, in 1981 alone (Robbins-Roth 2001). Perhaps even more important than the money or the science, though both were critical, was a focus on practical research goals—the diagnosis, treatment, and eventually the eradication of cancer. Dreams of extremely lucrative commercial applications—diagnostics and therapeutics—were not far behind. As discussed by Kenney (1986), collaboration between science and industry—what he called the university– industrial complex—was forged on a scale perhaps never before seen. Cetus Corporation, founded in Berkeley, California in 1971 by Ron Cape and Peter Farley, who brought scientific experience from both academic

88


and business laboratories is commonly attributed as the first commercial biotechnology firm in the USA. Like most early biotechnology firms, Cetus adopted an exceedingly broad approach to research and product development, aiming for commercial applications of biotechnology in areas as wide ranging as genetically engineered bacteria for alcohol production and oil-spill cleanups, as well as vaccines and therapeutic proteins for the prevention and treatment of human disease. This broad strategy developed in part from the interests of investors, which included Standard Oil, Shell Oil, and National Distillers (Robbins-Roth 2001). In part also, however, it emerged from early optimism about the generalizability of scientific techniques and processes for virtually all forms of life. Even long after the demise of Cetus, which was restructured as Agracetus in 1984 to focus on agricultural biotechnology through a partnership with the W. R. Grace Corporation, and eventually acquired and disbanded by Monsanto in 1996, business researchers distinguished what they called DBFs, encompassing diverse commercial applications, to set them apart from oil, chemical, agricultural, and pharmaceutical organizations who used a mixture of biotechnology and traditional product development techniques. Genentech, perhaps the first commercial biotechnology firm to focus on the development of pharmaceutical products using biotechnology techniques, was founded in 1976 by Bob Swanson, a venture capitalist with Kleiner and Perkins who had been an early investor in Cetus, and Herbert Boyer, a biochemist at the University of California, San Francisco, who had partnered with Stanley Cohen, a Stanford molecular biologist, in the development of a successful ‘gene-splicing’ technique in 1973. The Cohen–Boyer process patent, awarded in 1980, and licensed in 1982 to seventy-one companies across a broad range of industries, set off an explosion of commercial investment. Perhaps more important for the evolution of this industry, business investors and academic scientists, following the lead of Swanson and Boyer at Genentech, continued the practice of teaming in the formation of new biotechnology firms. Although this blurring of these traditionally distinct spheres of interest in basic versus applied research was fraught with controversy (Kenney 1986), academic scientists were drawn to entrepreneurial activity not only by the allure of significant monetary reward but also by the chance to participate in what promised to be one of the great leaps forward in the diagnosis and treatment of disease in history. Thus biotechnology, which might have remained mainly a new technique in the laboratories of big business, became the basis of a wholly new kind of industry.

89


From these beginnings, biotechnology has burgeoned into one of the largest and fastest growing new industries, involving more than 2,400 firms in more than 30 countries.1 Along the way, the firms have abandoned their early strategies for developing commercial applications of biotechnology in virtually every arena of life and chemical processes. The mysteries of life have proved more complex and confounding than was ever conceived by the founders of the early firms. Split first into broad categories of human diagnostics and therapeutics, agricultural biotechnology, and industrial and environmental biotechnology, firms today focus on very specific techniques for the production of new, bioengineered drugs, plants, and chemicals. Nonetheless, as traditional pharmaceutical firms decry the dearth of new blockbuster drug discoveries, and merge (and then merge again) to increase the size and scope of their product pipelines, the proliferation of new biotechnology firms, especially in the areas of human diagnostics and therapeutics, has continued. Histories of the biotechnology industry abound, including Kenney’s insightful examination (1986) of the early and often rocky collaborations between commercial investors and university scientists, and we will not attempt to review it further here. Suffice it to say that the business– academic collaboration has persisted. Venture capitalists continue to seek out promising science, and scientists, for the formation of new biotechnology firms especially in the area of human diagnostics and therapeutics. Scientists increasingly view entrepreneurship as a natural stage in their career paths, blurring the distinction even more between academic and commercial research. Just as important, large-established organizations, including now some of the early biotechnology firms (e.g. Amgen and Genentech) invest in the research of young biotechnology firms through licensing agreements and equity stakes that challenge the profitability, if not the science, of smaller firms. And they routinely spin out new firms, either as new independent or subsidiary organizations, to focus their research and investments. For our purposes here, it is important mainly to note two things. First, human biotherapeutics has emerged as a distinct segment of the overall biotechnology field. As we describe below, more than 650 organizations, including start-ups, spin-offs, direct entries by established firms, mergers, and joint ventures, have been located in the USA over the period 1976 through 2003. Second, the origins of entrepreneurs and firms in the industry are diverse, heralding from a variety of different organizations. Thus, the industry holds good promise for an investigation into the demographics of cluster development. Although we cannot offer strong claims

90


about the generalizability of our research, since it examines clustering for just one industry, we can suggest that, in its focus on comparing cluster development in more than seventy metropolitan statistical areas (MSAs) spread over thirty-eight states that became home to even one biotherapeutics firm, we greatly extend previous literature which has focused on case histories or formal mathematical modeling.

Research Method We investigated the organizational and geographic origins of entrepreneurs and firms in the US biotherapeutics industry over the period 1976 through 2003. Biotherapeutics refers to the segment of the biotechnology industry that is focused on the discovery and production of drugs for the prevention and treatment of human disease using biotechnological techniques (e.g. genetic engineering, monoclonal antibodies, and gene therapy). Biotherapeutics firms were identified based on evidence, from biotechnology industry research publications (in particular, BioScan, 1987 through 2004), of direct investments in the development of drugs, whether in research, clinical trial, or production and marketing phases. BioScan is a comprehensive industry directory that provides information about the founding dates, management teams, and geographic locations of biotechnology firms as well as their products and strategic alliances. These data were extensively augmented through searches of business and industry press publications available from Lexis–Nexus from the beginning of the industry, firm websites, and financial documents (e.g. prospectuses, annual reports, and 10-K filings) available from the US Securities and Exchange Commission. The result, we believe, is a comprehensive, though likely not exhaustive, longitudinal database on the organizational and geographic origins of firms in the US biotherapeutics industry. We identified 688 firms—including both USA and foreign-owned organizations—that were engaged in human biotherapeutics in the USA over the study period. Although we cannot claim that our sample is exhaustive of all human biotherapeutics firms operating in the USA over the study period, extensive supplemental inquiries into the histories of entrepreneurs and firms in the industry, as well as the evolution of the US biotechnology industry as a whole, revealed only a few, mainly very small and short-lived firms, that were not identified by the above methods. Thus, we believe that our coverage of the industry is comprehensive and accurate through about 2003. Of particular interest to questions explored in

91


this paper, we have no reason to believe our coverage is biased toward biotherapeutics activity in particular regions. Below, we describe the data we collected in more specific detail.

Life Demographics To track the organizational origins of entrepreneurs and firms in the industry, we collected data on the types of organizations from which biotherapeutics entrepreneurs emerged. Six categories of organizational origins were used: (a) universities or research institutes, (b) biotherapeutics firms themselves, (c) traditional pharmaceutical organizations, (d ) VC firms, (e) hybrids of the above categories in cases where two or more entrepreneurs emerged from different sources, and ( f ) other. Six categories were used to characterize the firms’ modes of entry into the industry: (a) start-ups, (b) spin-offs, (c) direct entries of established firms via investments in their own laboratories or the acquisition and absorption of an existing biotherapeutics firm, (d ) subsidiaries, whether newly created or acquired, (e) joint ventures, and ( f ) mergers. Four categories of exit mode were specified: (a) failure of the organization, (b) cessation of activities in human biotherapeutics, (c) acquisition, and (d) merger. In some cases, acquired firms were left intact as separate operating subsidiaries of the acquiring organizations, while in other cases their assets were absorbed into the activities of the acquiring organization. In the first scenario, even if the name of the organization did not change, we coded the exit of the original organization and the entry of a new organization as a subsidiary. Similarly, in the case of mergers, even when the merged organizations continued under the name of one of the merging organizations, we coded the exits of the merging organizations and the entry of a new organization. To avoid double counting, we coded the last year of existence for the acquired or merging organizations as the one in which the acquisition or merger occurred and the first year of existence for the new firm as the year following the acquisition or merger event.

Geographic Demographics The geographic origins and destinations of entrepreneurs and organizations were coded using the 2003 US Office of Management and Budget Bulletin No. 03–04 which lists MSAs and combined MSAs, among other groupings, based on information obtained in the 2000 Census. As described in the bulletin, MSAs were designated based on their having ‘at least

92


one urbanized area of 50,000 or more population, plus adjacent territory that has a high degree of social and economic integration with the core as measured by commuting ties.’ A total of 362 MSAs (not including Puerto Rico) were listed, encompassing 1,090 counties, approximately 35 percent, and about 83 percent of the US population. Combined MSAs (e.g. the Washington, DC and Baltimore MSAs) were defined as two or more geographically adjacent MSAs with employment interchanges (i.e. commuting ties) of at least 25 percent (or, in a few cases with employment interchanges of only 15 percent, but considered highly integrated based on the opinions of Congressional delegations). The bulletin lists 113 Combined MSAs in the USA, encompassing 310 MSAs. We classified the MSA locations of firms in the study, as well as the organizational origins of their entrepreneurs and antecedent firms, using zip codes, which allowed us to identify the counties and thus the relevant MSAs. The use of MSAs and combined MSAs, rather than city or state geographic boundaries, is attractive for identifying regions of activity in that MSAs are designated on the basis of regional economic integration independent of political boundaries. For example, the Washington, DC MSA— formally designated as the Washington–Arlington–Alexandria, DC–VA– MD–WV Metropolitan Statistical Area—encompasses three states plus the District of Columbia and twenty-one cities or counties which residents would recognize as part of the ‘greater’ Washington, DC area. Residents routinely traverse the political boundaries for living, work, shopping, and entertainment purposes. They read the same newspapers, the Washington Post, the Washington Times, and the Washington Business Journal, and debate important regional issues on a cross-regional basis. Major transportation facilities, including roads and rapid transit systems cross the political boundaries and decisions regarding the development of new transportation facilities involve officials from all political jurisdictions. By virtually any measure, greater Washington, DC, as represented by the MSA, is a single social and economic region. One potential problem arises from our classification procedures that must be discussed. Our study period is long, and the distribution of human populations across regions as well as the patterns of economic integration within regions changed over the study period. Our decision to classify organizations, as well as their origins and destinations, based on the 2003 classification system raises the likelihood that some organizations will be incorrectly located based on classifications at the time of the entry or exit event. For example, an organization founded in Worcester County, Massachusetts in any year up until 2003 would have been

93


classified as located within the Boston MSA; under the new classifications, Worcester has been designated as its own MSA, though the Boston and Worcester MSAs are designated as components of a larger, combined MSA. Thus, a firm that never changed its physical address could be classified as residing within different MSAs based on changes in the OMB classifications over the study period. By our procedures, such an organization is classified as being in the Worcester MSA over the entire period. To explore the extent of this difficulty, we tracked the classifications of counties (which are the core units of MSAs) from 1976 through 2003. The classification schemes changed most extensively every ten years following the decennial census; new MSAs were designated and old MSAs were reorganized in terms of their county components, sometimes combining counties that were previously classified as separate MSAs and sometimes reorganizing counties into two or more MSAs. Nonetheless, except for the addition of wholly new MSAs, such as Corvallis, Oregon in the 2003 classification, the MSA classifications over the study period were remarkably consistent. Approximately 85 percent of the MSAs listed in the 2003 bulletin were also listed in the 1976 bulletin; changes in their county compositions reflected mainly the addition of one or more adjacent counties. Indeed, except for the introduction of wholly new MSAs, the 2003 classification looks very similar to the 1976 classification, and they are more similar to one another than either is to the classifications of the intervening decades.2 Thus, assuming that the 2003 classifications represent trends in population growth and regional economic integration which were developing long before the official classifications, we feel comfortable using the single classification scheme to designate regions over the entire study period. To do otherwise would result in counts of regional biotherapeutics populations that varied simply on the basis of changes in classifications; moreover, there would be no way to assess relocation origins and destinations unambiguously.

Limitations These procedures give rise to two limitations to interpretation of the results of our study. First, we do not consider all regions potentially at risk of developing a biotherapeutics cluster. Although, as described below, the firms in our study were located, either originally or via relocation, in more than seventy US metropolitan regions, these represent only about 21 percent of all such regions in the USA. Full analysis of regional conditions that give rise to industry clusters would require consideration of all US

94


regions. All but eight of the top 50 largest metropolitan regions, however, are represented in our study, each having been home to at least one biotherapeutics firm. Twenty-five other regions are also represented; however, the majority was home to only one or two biotherapeutics firms. Thus, though not all regions are considered, we think we have sufficient range to draw conclusions about the demographic and geographic processes of cluster development. Second, it will not be possible to say whether our results are generalizable to other industries. Though the study is large, covering most of the firms engaged in human biotherapeutics over more than twenty-five years, it is still a study of only one industry. Our findings suggest that more formal longitudinal research and analysis of entire industries will be necessary to strengthen our understanding of cluster development processes.

The Anatomy of Cluster Development over Time and Space Table 5.1 shows the number of human biotherapeutics organizations established in the USA over the period 1976 through 2003, broken down by type of establishment, organizational origins or firms or entrepreneurs, and by whether at least one member of the founding team or organization (in the case of spin-offs, joint ventures, and mergers) relocated from a different MSA from the one in which the new firm was established. As shown, 688 biotherapeutics organizations, including both USA and foreign-owned organizations, were either established or entered into human biotherapeutics research in the USA over the study period. The majority, 455 (66 percent) were established via the start-up of a wholly new organization. Approximately 45 percent of these, 205 organizations, which accounted for 30 percent of all biotherapeutics organizations, were created by scientists out of universities or private research institutes. Subsidiaries, including both those that were newly created and those that were created via the acquisition of an existing biotherapeutics firm, accounted for 11 percent of the total. No other category accounted for more than 10 percent of the total.

Clustering and Geographic Convergence Table 5.1 also shows information about the relocations of entrepreneurs and firms from one MSA to another. Of the 688 firms, 220 (32 percent)

95

Anatomy of Cluster Development Table 5.1. US biotherapeutics firms by types of entry and organizational origins All firms Number Start-ups From university or research institute From an existing biotherapeutics organization From pharmaceutical organization Hybrid, from multiple source types From venture capital firm From other type of firm No information Spin-offs From university or research institute From an existing biotherapeutics organization From a pharmaceutical organization From other type of firm Direct entries by established firms Research in own laboratories Acquired a biotherapeutics organization Subsidiaries Newly created Acquired Mergers Joint ventures Total Migrant firms Grand total

Firms with nonlocal originsa

Percent

Number

0.30 0.10 0.03 0.07 0.09 0.03 0.03

41 20 10 25 22 5 26

0.00 0.05 0.02 0.01

0 7 3 1

455 205 70 24 50 64 21 21

0.66

55 1 34 11 9

7

75

0.03 0.11

0.04 0.07 0.04 0.02

20 39 16 4

688 60

0.00 0.03 0.01 0.01 0.09

0.08 0.01

27 48 30 11

0.19 0.09 0.05 0.11 0.10 0.02 0.11 0.08

62 52 10

Percent

0.09 0.18 0.07 0.02 220

60 748

280

a

These counts include organizations for which at least one of the founding firms (e.g. in the case of mergers) or entrepreneurs migrated from one MSA to another.

were created by entrepreneurs or firms that crossed a regional boundary at the combined MSA level to establish or relocate their organizations.3 Perhaps not surprizingly, since people and capital are more mobile than firms, the majority of these, 184 relocations (84 percent) were accounted for by start-ups and the creation of new subsidiaries. Spin-offs, mergers, joint ventures, and direct entries by established firms occurred predominantly in the same regions as their ‘parent’ organizations. In addition, sixty firms relocated wholesale, at least one year following the year of establishment, from one region to another at the combined MSA level. Table 5.2 presents information about the locations of US biotherapeutics firms. Firms were located in thirty-eight states plus the District of Columbia, encompassing 73 MSAs (19 percent, out of a total of 362 US MSAs) plus three counties that did not meet US Census Bureau criteria for classification as an MSA. Sixteen of the MSAs (22 percent of all

96


metropolitan regions with at least one biotherapeutics firm) were home to at least ten biotherapeutics organizations. When combined MSAs were considered, twelve regions (encompassing twenty-four MSAs) were home to at least ten biotherapeutics firms. These twelve regions account for 88 percent of all biotherapeutics firms that were either originally located in one of the regions or that relocated to one of the regions. All of these regions were among the top 50 in the US-based on population size, though three regions that ranked among the ten largest in population, including the Dallas, Detroit, and Atlanta combined MSAs, were not among the top 12 for biotherapeutics firms. All were home to at least one major research university, accounting for approximately 34 percent (51 out of 150) of all ‘research—intensive’ universities as ranked by the Carnegie Foundation (2000). Moreover, all were among the largest recipients of grants from the National Institutes of Health, though many other regions received similar numbers and total amounts of funding.

Table 5.2. Biotherapeutics firms in MSAs and combined MSAs MSA

Combined MSA

Number

Percent

Number

Percent

San Francisco–Oakland–Fremont Napa San Jose–Sunnyvale–Santa Clara Boston–Cambridge–Quincy Manchester–Nashua NH Worcester New York–Newark–Edison Trenton–Ewing New Haven–Milford Bridgeport–Stamford–Norwalk Poughkeepsie–Newburg–Middletown San Diego–Carlsbad–San Marcos

102 2 44 99 1 10 75 11 10 2 1 88

148

0.23

110

0.17

99

0.15

88

0.13

Los Angeles–Long Beach–Santa Ana Oxnard–Thousand Oaks–Ventura Philadelphia–Camden–Wilmington Washington–Arlington–Alexandria Baltimore–Towson Durham Raleigh–Cary Seattle–Tacoma–Bellevue Houston–Bay Town–Sugarland Boulder Denver–Aurora Chicago–Naperville–Joliet Total

36 1 34 28 5 29 3 26 20 10 5 14 656

0.16 0.00 0.07 0.15 0.00 0.02 0.11 0.02 0.02 0.00 0.00 0.13 0.00 0.05 0.00 0.05 0.04 0.01 0.04 0.00 0.04 0.03 0.02 0.01 0.02

37

0.06

34 33

0.05 0.05

32

0.05

26 20 15

0.04 0.03 0.03

14

0.02

97


Thus, regional size and resources appear to be important in the locations of biotherapeutics clusters, but not fully explanatory. As shown in Table 5.2, these regions were home overall to the establishment or relocation of 656 (88 percent) biotherapeutics firms over the study period. Of these, four regions (combined MSAs)—San Francisco (148 establishments), Boston (110 establishments), New York (99 establishments), and San Diego (88 establishments)—accounted for 59 percent (365 MSA entry events) of the locations and relocations of biotherapeutics firms across all regions. The next most clustered regions—Los Angeles, Philadelphia, Washington, DC, and Raleigh–Durham—were home to an average of thirty-four biotherapeutics firms each. Finally, the least clustered of the top 12 regions—Seattle, Houston, Boulder, and Chicago—were home to twenty-six or fewer biotherapeutics firms. Of the remaining fiftytwo regions, most were home to only one or two biotherapeutics firms. Overall, ninety-two firms (12 percent) were located in or relocated to regions outside the top 12. Thus, the US biotherapeutics industry appears substantially clustered. Although, even over almost thirty years of this industry’s existence no single region has emerged as solely dominant, the majority of firms were established or have relocated to a relatively few regions. Moreover, the majority of these, 486 firms (68 percent) were founded by entrepreneurs or created from the activities of organizations that already resided in the region where the new firm was established. Thus, our findings provide moderate support for the argument that clusters arise primarily from the investments of local entrepreneurs, firms, venture capitalists, and other interested parties. Nonetheless, 32 percent of firms (n ¼ 220) were created by entrepreneurs or out of other firms that were located elsewhere. This finding, combined with the finding of sixty firms that relocated wholesale to a new region at the combined MSA level, suggests that a considerable amount of movement of resources is also occurring. To get a better understanding of geographic patterns in entrepreneurial and firm relocations, we next examined whether the relocations of entrepreneurs and firms indicated a convergence of resources to the top 12 metropolitan regions.

Cross-Regional Relocations Tables 5.3 and 5.4 show directional patterns in the relocations of entrepreneurs and firms both between the top 12 regions and from regions, including foreign countries, outside the top 12 regions. For entrepreneurs, sixty-three relocations (61 percent) occurred between regions that

98


comprised the top 12. A striking majority of entrepreneurs who relocated from US regions outside the top 12 (89 percent) or from foreign countries (100 percent) relocated to one of the top 12 regions. Within the top 12, the majority of relocations (n ¼ 20) were done by entrepreneurs who moved from the New York region to other regions, followed by entrepreneurial relocations from the San Francisco region (n ¼ 10). The San Francisco and Boston regions were the principal recipients of entrepreneurs from all other regions, receiving, respectively, twenty and nineteen entrepreneurs from elsewhere. To get a clearer idea about regional gains and losses of entrepreneurs, we calculated the ratio of relocations away from the top 12 regions to relocations into the regions. The last column of Table 5.3 shows these ratios. A number greater than 1 indicates that a region was losing more entrepreneurs than it was gaining. By this measure, New York, Los Angeles, and Chicago all showed a net loss in their numbers of biotherapeutics entrepreneurs. Table 5.4, which shows the relocation patterns of firms that moved from one region to another, reveals a slightly different pattern. A somewhat greater number (45 percent) of firm relocations were from regions outside the top 12 regions. Again, the majority (70 percent) relocated to one of the top 12 regions. Within the top 12 regions, while New York was again the loser of the greatest number of firms (n ¼ 8), it was also the recipient of the largest number of relocating firms (n ¼ 12). Five of the relocations to New York were from regions outside the top 12. The Raleigh–Durham, Houston, and Denver–Boulder regions showed net losses based on the number of firms that relocated in and out of the regions. Thus, we find evidence of geographic convergence toward the top 12 regions. Most of the entrepreneurs and firms from outside the top 12 regions relocated to one of the top 12 regions. Moreover, the large majority of relocating entrepreneurs (90 percent) and firms (76 percent) moved between regions in the top 12. Overall, the San Francisco, Boston, and San Diego regions were the predominant recipients of relocating entrepreneurs and firms. New York, which was also an important recipient of relocating entrepreneurs and firms, was also a considerable source.

Intraregional Organizational Origins As indicated above, the majority of US biotherapeutics firms were established by entrepreneurs or firms in regions where they already resided. To learn more about the growth patterns of the top 12 regions, and perhaps to obtain insights regarding the substantial differences in their growth rates

99

100 Table 5.3. Entrepreneur relocations across combined MSA for top 12 regions Other— Foreign San Boston New San Philadelphia Raleigh– Washington, Los Seattle Houston Denver– Chicago US Francisco York Diego Durham DC Angeles Boulder Other—US San Francisco Boston New York San Diego Los Angeles Philadelphia Washington, DC Raleigh– Durham Seattle Houston Denver– Boulder Chicago Total

3 8 4 2 1 0 2 3

1 2 1 5 2

1

2

2 3 6

3 2

1 1

4 1 1

1 3 1 1

1 1

1 1

1 1 1

1

1 1 1

1 4 3

1

1

2

1

0 1 1

1 1

1 28

12

1

2 1 1

1

10

3

20

8

7

0

1 4

4

0

2

0

5

Table 5.4. Firm relocations across combined MSA for top 12 regions Other— Foreign San Boston New San Philadelphia Raleigh– Washington, Los Seattle Houston Denver– Chicago US Francisco York Diego Durham DC Angeles Boulder

Other—US San Francisco Boston New York San Diego Los Angeles Philadelphia Washington, DC Raleigh– Durham Seattle Houston Denver– Boulder Chicago Total

8 3 1 5 2 1 0 3

2 25

1

1

2

1

3 1 2

Number of relocating firms 1 1 1 1

1

1

1 1

1

2

1 1

2

2

2

1 1

1 1

0 0 0 0

1

1 1

1 1

2

4

4

8

2

1

3

2

3

1

1

101


over time, we examined the longitudinal organizational and geographic origins of entrepreneurs and firms in each of the regions. Table 5.5 shows patterns for San Diego and Table 5.6 for Los Angeles. They exhibit very different patterns in the organizational and geographic origins of their firms. As we will discuss, with only two exceptions (Raleigh-Durham and New York), these patterns were shown for all other regions and appear to distinguish regions, among the top 12, which became home to large biotherapeutics clusters and those which did not. San Diego, as one of the top four regions, was interesting for study because it was not previously home to other high-technology clusters in contrast to the San Francisco and Boston regions. There are several important things to notice about Table 5.5. First, the San Diego region was home to only six biotherapeutics firms during the first ten years of the industry, from 1976 through 1985. This contrasts with much faster growth rates in New York with thirty-three firms, San Francisco with twenty-five firms and Boston with sixteen firms during the same period. This is a quite striking difference when we consider that San Diego has greatly narrowed the number of firms in the region as compared to these other regions. Second, during the period 1986 through 1995, while the number of scientific start-ups continues and even increases a bit, a large number of new firms are established with entrepreneurial and firm origins in the original six firms. Thus, it appears that San Diego achieved its growth substantially through the founding of new firms by scientists and managers with experience in other local biotherapeutics firms. Although not shown, similar arrays of data for the San Francisco and Boston regions show the same pattern. The significance of this pattern will become clearer when we examine Table 5.6 for the Los Angeles region. Third, beginning during the period 1986–95 but increasing during the period 1996–2003, we see a significant inflow of relocating entrepreneurs and firms as well as capital via the investments of distant firms in San Diego firms.4 Finally, during this final period, we see a slight increase in the number of entrepreneurs and firms who relocated out of San Diego and a substantial increase in the number of San Diego firms that were acquired and their assets absorbed by firms outside the region. Before interpreting the patterns shown for San Diego, which, as noted above, were similar to those for the San Francisco and Boston regions, we should examine the patterns for the Los Angeles region which was home to a considerably smaller number of biotherapeutics firms, although still the fifth largest among the top 12 biotherapeutics regions. From 1986 through 1985, the first ten years of the industry, Los Angeles was home

102

Table 5.5. Distribution of biotherapeutics firms in San Diego

1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

Total entries

Science startups

1

1

2 1

2

2

1

2 9 1 8 4 3 6 6 2 8 2 8 5 3 7 2 1 3

Biotherapeutics start-ups

Hybrid start-ups

Biotherapeutics spin-offs

Venture capital start-ups

Pharmaceutical start-ups

Direct entries

Mergers, acquisitions, and joint ventures

In relocations of entrepreneurs

In relocations of firms

In relocations of capital

1

1

Local acquisitions and mergers

2 3 1

2

1 1

1

1

2

1 1

2

1 1 1 1

1

1 1

2 1

1 1 1

Out relocations of assets

1 1

1

1

2

Out relocations of firms

1

1 2

Out relocations of entrepreneurs

1

1

2

3 1 1 2

Failures

1

1

1 1

2 2 2

1 1 2

1

1 1 1

2

1

2

1 2 1

1

1 1

3 1 2

1

1

1

1 2 1 2

1 1

1 1 1

1 1

1 1

1 2 3 1 3 4 3 1 3 4

Table 5.6. Distribution of biotherapeutics firms in Los Angeles

1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

Total entries

Science startups

4 4 1

1 2





Direct entries

1 1

1

1


In reloca tions of entrepreneurs

In reloca tions of firms


Failures



1



1

1 1

1

4

Hybrid start-ups

1

1

1

1

1 2

1 2 2 1 1

1

1 1

1

1 1

1

1

2

1

1

1

1

1

2 1 1 1 4 1 3 3 2

1

1 1

1 1

1

1

1 1

1 1

1

1

1

1 1

2 1

1 1

2


to ten biotherapeutics, though only three founded by local scientists out of universities or research institutes. This is a slightly larger number of early firms than we observed for San Diego. The number of firms founded or relocated to the regions increases a bit during the period 1986 through 1995, but we see no indication of the growth in start-ups out of the earlier biotherapeutics firms that we observed for San Diego. This is perhaps surprising since Los Angeles was and is home to Amgen, one of the largest and most successful biotherapeutics firms in the USA. From 1995 through 2003, Los Angeles continued its slow pace of start-ups by entrepreneurs out of local universities and research institutes, but shows a slight increase in the number of entrepreneurs and firms that relocated to the region. Also, during the third period from 1995 through 2003, Los Angeles shows a noticeable increase in the number of firms it loses due to failure, the relocations of entrepreneurs and firms to other regions, and the acquisition and absorption of its firms by firms in other regions. Again striking in contrast to San Diego is the relatively smaller number of Los Angeles firms that were attractive as acquisitions for firms in other regions. Overall, the pattern shown for Los Angeles was similar to all of the other smaller regions, including Philadelphia and Washington, DC, which were home to about the same number of biotherapeutics firms as Los Angeles, and excluding the Raleigh–Durham region. Thus, out of the twelve top regions, ten exhibit developmental patterns that are highly similar to either San Diego or Los Angeles. San Diego, Boston, and San Francisco, while they continue a paces of biotherapeutics science founding throughout the study period, also exhibit a secondary pattern of biotherapeutics start-ups by entrepreneurs from other local biotherapeutics firms. By contrast, Los Angeles, in a pattern that is similar to all but one of the other smaller regions, exhibits no such secondary pattern. It appears that this secondary growth phenomenon, that is, the start-up of new firms by entrepreneurs from other biotechnology firms, is critical to the overall growth of the cluster. Note, we are not saying simply that higher rates of growth accounts for greater size; rather, we emphasize that it is the development of a particular type of growth that appears to distinguish these regions. Clusters grow when the knowledge and other resources created by the early firms in a cluster are ‘combined and recombined’ (Schumpeter 1934) by entrepreneurs from the early firms for the creation of new organizations. Two regions, Raleigh–Durham and New York, were anomalous with respect to these patterns. Since about 1997, Raleigh–Durham, though we do not show a table, has experienced a substantial increase in the number

105


of biotherapeutics firms founded out of only a few biotherapeutics firms in the region during the preceding twenty years. Although it is too early to draw strong conclusions about the increase, it appears that Raleigh– Durham may be following the pattern identified for San Diego, San Francisco, and Boston and that the region may thus be poised for substantial growth, though at a much later period in the history of the biotherapeutics industry. If the pattern continues, however, it suggests that cluster growth may not be strongly tied to historic periods. Rather, it may depend more on the characteristics of growth, in particular the tendency of scientists and managers from local, established firms to strike out on their own to found new organizations in the same region. New York exhibits a very different pattern altogether. As shown in Table 5.7, New York shows both a large number of science start-ups and a large number of direct entries during the first ten years of the industry. The large number of direct entries contrasts with other regions, which had few, but is not surprizing given the location in the New York region of a large proportion of US pharmaceutical firms. New York also shows a large number of entrepreneurs who relocate to the region during the early period, in contrast to other regions. Though not indicated in the table, many of these entrepreneurs were from foreign countries or US metropolitan regions that were not one of the top 12 regions. The pattern of relocation to New York is continued in later periods, primarily through the relocations of firms, again many from US regions outside the top 12. Perhaps the most striking feature of the table, however, is that New York, like Los Angeles, exhibits no secondary growth phase through the start-up of new biotherapeutics firms by entrepreneurs from other local biotherapeutics firms, despite the fact that the region was home to more start-ups during the early period than any other region. Moreover, after about 1992, the rate of founding from scientific or any other organizational origins drops to near zero. New York remains attractive, during the later period, to entrepreneurs and firms from other regions, but it has virtually ceased to be generative of any new firms locally.

The Development of Industrial Clusters Industry clusters develop within complex contexts of national and even international scrutiny and influence. Although explanations for the location and growth of industry clusters have emphasized local factors, including the presence of important resources and social processes that

106

Table 5.7. Distribution of biotherapeutics firms in New York Total entries

1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

2 1 6 6 5 4 6 4 5 4 7 4 3 3 6 2 2 3 3 1 1 3 4 4 0 3

Science startups


Hybrid start-ups

1 1 1 1 3 1 2 1 5 2 1 2




Direct entries

2 1 2 3 2

1

1


In reloca tions of entrepreneurs

1 1

2 1 1 1

1 1 2

2

In relocations of firms


Failures




2

1

2 1

1 1 1 1

1 1

1

1

1

1

1

1

1 1

2 1


1 1

1

1 1

2 1 1

2 1 1 3 1

2

2 3

2 1

1 1

2 1 1 1 1 1

1 1

1

2

1

1

2

2

3 1 1


promote the formation of new firms, few studies have examined the growth of industry clusters in the context of multiple regions that may be at risk for developing a particular kind of industry cluster. It often seems obvious, in retrospect that a cluster would have developed in a particular region. Historic analysis may point to seemingly unique resources that were available in the region; however, we cannot assume this inevitability with evidence, typically, on just one or two prominent clusters. In this chapter, we have explored three important questions relating to the location and growth of industry clusters on a comparative basis. First, we examined whether clusters grow predominantly through the investments of local entrepreneurs, local firms, and local venture capitalists in the formation of new organizations or the entry by established organizations in a new line of business. In support of prevailing theory, our findings reveal a critical importance of local investments in two ways. First, of over seventy-six metropolitan regions in the USA (all that were home to at least one biotherapeutics firm), the majority (68 percent) were founded by entrepreneurs and other investors in the regions in which they already resided. Second, for three of the regions with the largest clusters— San Diego, Boston, and San Francisco—the critical spur to growth appears to be a tendency of entrepreneurs to leave local, established biotherapeutics firms to found additional biotherapeutics firms. Most regions continued to generate new biotherapeutics by entrepreneurs out of local universities and research institutes at a relatively steady pace. Only those regions, however, that exhibited this secondary or second-generation growth from the biotherapeutics firms themselves grew to substantial sizes relative to other clusters. We can only speculate about the conditions and processes that led entrepreneurs and firms in some regions to produce the secondgeneration growth. Our findings are consistent with many case histories of cluster development that emphasize regional cultures and patterns of social interaction (Murtha Lenway, and Hart 2001; Saxenian 1994; Storper and Venables 2002). Second-generation growth, which involves entrepreneurs leaving established firms in a cluster to found competing new organizations, requires that the entrepreneurs believe in their abilities to attract capital and especially human resources to support their new organizations. It is difficult to conclude that such beliefs could develop unless the leadership of the earlier organizations was supportive of new entrepreneurial efforts. Saxenian (1994) explored this question in her examination of eventually profound differences in the growth rates of

108


semiconductor clusters in Boston and Silicon Valley. She argued that the leaders of semiconductor producers in Silicon Valley, most of whom had been the founders of their firms, were far more supportive of employees (managers and engineers) who sought to strike out on their own than the leaders of semiconductor producers, many of whom had not been the founders of their firms. In keeping with this argument, Gompers, Lerner, and Scharfstein (2003) showed that VC-funded organizations were more likely to generate other new entrepreneurial firms than organizations created or funded in other ways. These arguments suggest a potential explanation for our anomalous findings regarding the growth of a biotherapeutics cluster in the New York region. Although the New York biotherapeutics cluster grew faster at an earlier stage in the industry than the clusters in any other region, a large proportion of this growth came about from the entry of large, established pharmaceutical firms in the region. Although local universities produced entrepreneurial scientists in numbers commensurate with other regions, these firms did not generate the second-generation growth of other regions that eventually surpassed New York in the sizes of their clusters. It seems likely, based on Saxenian’s reasoning, that the established pharmaceutical firms, having little experience in memory of entrepreneurial activity—most of the large pharmaceutical firms are close to 100 years old—were not encouraging of any inclinations on the part of their talented employees to leave the firms to pursue an entrepreneurial enterprise. Perhaps these large pharmaceutical firms also exerted a stultifying influence on the pace of start-ups overall. We may speculate, again extending from Saxenian’s findings in semiconductors, that the interconnectedness of firms in the region, and the associated density of information sharing, was inferior, compared to that of other regions. Our findings, however, did not fully support a conclusion that cluster locations and growth are solely the consequence of local investments. More than 200 entrepreneurs (32 percent) relocated from one metropolitan region to another to found new biotherapeutics firms. Moreover, sixty firms relocated, following establishment in one region, to another region. This finding supports the idea that entrepreneurs are scanning regional environments, beyond their home regions, for evidence of more attractive locations and that they are willing to relocate. Our analysis of the directional patterns in entrepreneurial and firm relocations showed that, while the majority of relocations occurred between regions among the top 12, entrepreneurs and firms from regions outside the

109


top 12 were overwhelming likely to relocate to one of the top 12 regions. San Francisco, Boston, New York, and San Diego, which have become home to the largest biotherapeutics clusters in the USA, were the most frequent recipients of the relocating entrepreneurs and firms. These findings suggest that, while the most clustered regions jockey to attract entrepreneurial talent and the resources of established firms, the industry overall is converging toward a few, larger and larger clusters. Finally, we examined in greater detail the organizational and geographic origins of three of the top five clustered regions. In addition to the importance of second-generation growth, our findings suggest that the attraction of entrepreneurs and firms to a region is a tertiary influence on growth, occurring late in the history of the industry and the clusters. San Diego, which showed a strong increase in second-generation growth, and even Los Angeles, which failed to generate this form of growth, both began to attract entrepreneurs from other regions during the late stage of the industry. New York also attracted many entrepreneurs and firms during this period; however, for this region, the attraction appeared to be a continuation of trends developed very early for this cluster. Overall, most regions did not begin to attract entrepreneurs and firms until later in the development of their clusters, if at all. Although the US biotherapeutics industry may yet be too young to draw strong conclusions about convergent geographic patterns, these findings support theory about the role of expectations in cluster growth, if not necessarily their locations. It is not obvious that small early leads in the growth of industry clusters were influential in the future growth of the clusters. San Diego, which lagged far behind most other regions in its production of early biotherapeutics firms, eventually grew nonetheless to become home to one of the largest biotherapeutics clusters. And New York, which had more early biotherapeutics firms than any other region, including a large complement of science startups, has largely failed to generate growth in its cluster. With the exception of New York, which has been attractive to relocating entrepreneurs and firms throughout the history of its biotechnology clusters, the top regions did not begin to attract entrepreneurs and other firms until they had achieved considerable sizes from local growth and investments. Thus, it is hard to conclude that the locations of industry clusters are accidental. Industry clusters are not only a universal feature in the spatial arrangements of industries, but also an essential component in the economic evolution of industries. Clusters provide not only a near term source of

110


economic wealth, but also a long-term foundation for future economic growth. The knowledge, institutions, and infrastructure of one large cluster, in addition perhaps to a culture of entrepreneurial activity that may be generated by the presence of one large cluster, prepare the local environment for investment in future clusters, both those that may be closely related to the original cluster and those that may be broadly new. Although regional civic and business leaders appear to have recognized these facts, responding with all manner of incentives to attract entrepreneurs and firms to their regions, little research to date has compared demographic patterns in cluster development for evidence of most prolific development. This study presented here, which is the first that we know of to track cluster development processes among regions that may have been similarly at risk of developing a biotherapeutics cluster, suggests that only those regions which generate a community of entrepreneurial activity, represented by what Jane Jacobs (1969) called breakaway firms, that is, firms that are started by entrepreneurs with experience in other entrepreneurial firms in the same industry, may be capable of long-term cluster persistence.

Acknowledgments We are indebted to Martin Doyle, an MBA graduate of the McDonough School of Business at Georgetown University whose earlier master’s degree in microbiology and extensive industry experience aided in the classification of firms engaged in human biotherapeutics research and product development.

Notes 1. Source: BioScan listings, 1987 through 2003. 2. Though the OMB bulletins reporting MSA classifications describe numerical criteria, based on commuting ties, for designating the inclusion or exclusion of counties in particular MSAs, the judgments of the coders, and at times, of regional Congressional delegations also influenced the assignments. 3. We only coded relocations when entrepreneurs or firms crossed geographic boundaries at the combined MSA level. While some combined MSAs are very large, e.g. the San Jose–San Francisco–Oakland combined MSA is spread from

111

Anatomy of Cluster Development Napa to Santa Clara counties, most relocations crossed contiguous counties within the combined MSAs. We focused on this larger geographic level to avoid overcounting the extent of entrepreneurial and firm relocations. 4. These data represent only those investments that resulted in the majority acquisition of San Diego firms. We found much evidence of other, smaller equity investments in the San Diego firms, but were not able to collect complete data on such investments.

112

6 Policy-Induced Clusters: The Genesis of Biotechnology Clustering on the East Coast of China Martha Prevezer and Han Tang

China has recently witnessed the emergence of a number of potentially important biotechnology clusters on the east coast. Notwithstanding that several of these agglomerations are in a very early stage, a pattern can be discerned where the most important clusters are centered around Beijing, Shanghai, and Shenzen/Guangdong. The current, formative stage of the Chinese biotechnology industry implies that the coming decade(s) are likely to be characterized by continuing reorganization, reshuffling of factors of production and a shakeout of less competitive biotechnological sites. The purpose of this chapter is to examine the role of policies in propelling the emergence of biotechnology clusters. To simultaneously build new institutions and reform existing ones, alongside the introduction of policies that encourage greater entrepreneurship and risk-taking, is a highly complex task. The development of policies includes learning from previous institutional reforms (over the last twenty years), and providing a business friendly environment where ex-pats can resettle and find fertile ground for nurturing start-ups. China, as a transitional economy, has been moving from planning to greater market orientation. Still, firms remain predominantly state owned especially for most large- and medium-sized enterprises. Collective and privately owned firms are mostly smaller sized, and entry of new firms has been of these nongovernmental types of enterprise. The next section looks at science and technology policies in China and then focuses on

113

The Genesis of Biotechnology

those policies that are targeted specifically at developing biotechnology. Section 3 examines the structure of the biotechnology sector in China, emphasizing the three dominant biotechnology cluster regions on the east coast. Thereafter these three regions are compared in terms of size, knowledge base, and specialization. The final section concludes.

Toward a Market-Oriented Economy The classification of different policy stages outlined in Chapter 1 will serve as the structural framework for the analysis. Emphasis is on the first two phases. The first refers to the creation of background and preemergence conditions, while the subsequent phase features the coevolution of institutions alongside start-ups. These stages characterize the transformation from an—by and large—entirely state-owned and state-run economy to a reorientation toward the creation of start-ups and regional clusters in high-tech sectors. The preemergence phase (1985–95) included a strengthening of property rights, especially for science and technology enterprises, to allow for the creation of spin-offs from research institutes, universities, and large enterprises and to encourage autonomy in decision-making for those enterprises. These are not private property rights as recognized in the West, but they do represent greater autonomy for the enterprise in terms of decision-making. This followed from ‘The Decision on the Reform of the Science and Technology Regime’ in 1985, by which the central government allowed the foundation of Minying Keji Enterprises (MKEs) which are nontraditional, but government-owned technology enterprises. Most of these MKEs were spin-offs where the enterprise was set up by a group of scientists, primarily supported by research institutes and universities (Liu and White 2001; Lu and Lazonick 2001). In 1988 establishment of MKEs was further encouraged (‘Decision on the Further Reform of the Science and Technology Regime’), followed by a speech of Deng Xiaoping in 1992 which eliminated ideological obstacles to developing nonsocialist ownership types and unleashed a boom in the creation of MKEs (Table 6.1). In the subsequent phase these policies were consistently pursued and resulted in a considerable increase in start-ups from the mid-1990s onwards. The numbers of MKEs increased from 26,000 in 1992 to 109,400 in 2002 (Ministry of Science and Technology (MOST) 2003c). The incentives launched to promote increased start-ups included locational stimulus,

114

The Genesis of Biotechnology Table 6.1. Summary of science and technology and biotechnology policies

Science & technology policies

Phase I mid-1980s–95/6

Phase II 1995/6–2003

Creation of MKEs allowed and encouraged Improve infrastructure

R&D expenditures to enterprises increased Enhancing Technological Innovation And Developing High technology, Law of Commercialization of S&T Results 1996

Horizontal support to business sector

Enhance R&D Key Technologies R&D Programme

Attract returnee talent Lower entry barriers for start-ups Direct and indirect financial support for start-ups Establishing Science & Technology Industry Parks Founding Service Centres for Scientific and Technical Entrepreneurs Setting up Innovation Funds for Small Technology Firms Chun Hui Programme 1996 to attract overseas scholars to China

Biotechnology policies

National High-Tech R&D Programme (863 Programme) 1986

Torch Programme 1988 State Key Laboratory Programme

National Basic Research Programme (973 Programme) 1997 National Life Science and Technology Talents Forging Base Programme 2002 Encourage Overseas Scholars to Serve the Country in Different Ways Suggestion on supporting Senior Overseas Scholars Establishing biotechnology parks Incentives to biotechnology firms

lower entry barriers for entrepreneurs, policies to improve financial constraints, and allowing a Chinese VC industry to be established. Government expenditure on R&D also shifted from governmental institutions (from 43 percent in 1997 to 27 percent by 2002) toward R&D expenditure on enterprises receiving more than 60 percent in 2002. Locational incentives and design mainly relied on the imitation of other countries’ science and technology (S&T) industry parks (Lai and Shyu

115


2005). Since the first S&T industry park was established in 1991, fifty-three national level parks were founded across the country by the end of 2002. The number of enterprises in these sites increased from 2,587 in 1991 to 20,796 in 2000. This was complemented by the establishment of service centers that helped entrepreneurs setting up new businesses and the provision of incubator milieus. In 2002—according to the Ministry of Science and Technology—23,373 enterprises were in and 6,927 firms had ‘graduated’ from these service centers. Less success has been experienced as regards the establishment of an indigenous VC industry, which was high on the agenda already in the mid-1980s; it was not until around 1996 that VC companies began to emerge. Their numbers rose from 23 in 1996 with a capital of RMB3.5 billion to over 246 VC firms in 2001 with capital of RMB40 billion (China Venture Capital Yearbook 2002). The process started with some local governments that set up investment institutes, supplying funds to startups and incubators in high-tech parks.

Biotechnology Policies in China Biotechnology in China was targeted as a key industry in the National High-tech Research & Development Plan (the 863 plan) which was introduced in 1986. The biotech industry had received about a quarter of the funds from the program by 2000. The program is said to be responsible for generating 455 patents, 10,278 papers and 273 biotechnology projects, training over 3,000 postgraduate students and being involved in supporting around 30 biotech firms (MOST 2001). The Torch program, initiated in 1988, was set up to promote the commercialization of key high-technology projects through market mechanisms. It directed development in certain industries, principally the electronics sector, but increasingly also the biotechnology sector. Part of its focus was also the creation of high-technology industry development zones with the aim of assisting the application of R&D to production and commercialization. In the ninth five-year plan (1995–2000), the Chinese government further stressed the importance of biotechnology R&D. At the same time, indigenous companies realized the importance that government was giving to this sector and became interested in biotechnology. Local governments formulated similar policies supporting the development of their local biotechnology industry. A high priority was also put on encouraging the return of Chinese scientists working abroad through, for example providing a favorable environment for returnees to

116

The Genesis of Biotechnology Table 6.2. Returnees as percentage of top scientists in China, 2003 Scientist Academician of Chinese Academy of Engineering Academician of Chinese Academy of Sciences Academician of Chinese Academy of Social Sciences Senior Expert of Chinese Academy of Medical Sciences Academician of Chinese Academy of Agricultural Sciences Scientists in the ‘863’ Programme during the Ninth Five Year Plan (above Director of research project level) Electee of Millions of Talents Programme

Returnees (%) 81 54 4 37 14 72 18

Source: Service Centre for Returnees of Ministry of Personnel, China.

set up their own businesses in specially designed ‘Returnee Parks’. Since the Open Door policy was introduced in 1978, it is estimated that around 700,000 students went abroad to study, over 30 percent of whom were in the field of biotechnology or related areas (Qi C.Y. 2003) and roughly a quarter have returned to work in China. These returnees make a significant contribution to the Chinese Academies of Science (Table 6.2). A key ingredient in the development of biotech clusters has been to secure a skill base in biotechnology, which has been strengthened since the mid-1980s. By a country ranking for all fields of science (Table 6.3), sorted by papers, China was ranked nineteen by number of citations in 2003, and ninth with respect to the number of papers that have been cited. China’s biotech output increased sevenfold between 1991 and 2002, with the number of health-biotech papers published increasing from under 50 in 1991 to over 300 by 2002, roughly on a par with South Korea and well in advance of India and Brazil (The Economist, December 9, 2004). Another indication of scientific progress is the participation by Chinese scientists in the Human Genome Project, where they were responsible for 1 percent of the genomic sequencing. A significant part of this was done at the National Human Genome North Center in Beijing, at the Beijing Genomics Institute (the Beijing Huada Genomics Research Center) and at the National Human Genome South Center in Shanghai. Costs for skilled labor are also significantly lower in China than in the USA or Europe. The average annual salary for 2003 in pharmaceuticals and bioengineering was approximately $5,256.4 (Business Alert—China 2003). This compares with an average US salary in the biotechnology sector of $95,000 in 2003 (American Association for the Advancement of Science). About 21 million scientists, engineers, and other professionals were

117

The Genesis of Biotechnology Table 6.3. Top 20 country rankings of the most-cited 150 countries in all fields, 2003 Rank

Sorted by citations Country

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

USA England Germany Japan France Canada Italy Netherlands Switzerland Australia Sweden Spain Scotland Belgium Russia Israel Denmark Finland People’s R China Austria

Citations

Papers

Citations per paper

33,089,756 6,212,840 5,857,244 5,098,499 4,213,581 3,549,116 2,569,970 2,135,032 1,769,220 1,736,998 1,600,307 1,419,447 967,215 935,873 848,345 835,818 827,292 700,902 658,355 574,298

2,705,352 598,470 655,586 713,542 484,291 358,007 310,557 194,710 137,661 211,549 152,632 209,762 93,327 99,226 285,856 95,942 76,889 71,328 236,996 68,610

12.23 10.38 8.93 7.15 8.7 9.91 8.28 10.97 12.85 8.21 10.48 6.77 10.36 9.43 2.97 8.71 10.76 9.83 2.78 8.37

Source: ISI Essential Science Indicators Web-based product from the November 1, 2003 update covering a ten-year and eight-month period, January 1993–August 31, 2003. Notes: Country counts are based on the institutional affiliations given on published chapters. A chapter is attributed to a country if the chapter carries at least one address from that country. All addresses are considered, not only the address listed first. A country is only counted once when it appears more than one time on a chapter. All unique countries on a chapter are credited equally for the chapter. All citations received by a chapter are credited equally to all the countries on the cited chapter. No restrictions are made on the citing items in compiling the citation counts, other than that they are recorded from ISI-indexed journals only.

engaged in China in science and technology institutions and state-owned enterprises in 1999, up from 5 million in 1980. The number of science graduates in China above Masters level has risen by almost 80 percent since 1994 to nearly 10,000 graduates in 2002. In relative terms, (number of scientists and engineers in R&D per million population or R&D spending/GDP), China’s R&D ranks below that of Korea or Taiwan but is higher on average than that of Latin American countries (EIU 2000).

Three Biotechnology Clusters Compared Figures 6.1 and 6.2 show the geographic distribution of biotech research organizations and biotech firms respectively, in China, by province. As can be seen from the figures, three specifically densely populated biotechnology areas appear: Beijing, Shanghai, and Shenzhen (Guangdong).

118


Heilongjiang

Beijing

Jilin Liaoning

Xinjiang Gangsu

Inner Mongolia

ei

.

Heb xi Sh a

Qinghai

Tianjin

ng

Ningxia

Shangdong

Shanggxi Henan Anhui

Tibet

Jiangsu

Shanghai

Hubei Sichun

Zhejiang Guizhou

Yunnan

Hunan

Guangxi

Jiangxi Fujian

Guangdong Taiwan Hongkong Hainan

Figure 6.1. The geographic distribution of biotech research organizations in China Source: Database of Chinese biotechnology research organization and researchers released by China Biotechnology Development Centre of ministry of Science & Technology, Life Science & Biotechnology Bureau of Chinese Academy of Science and Association of China Biotechnology Engineering at www.biotech.org.cn (there are 513 biotech organizations in the database).

Characteristics of the Dominant Biotechnology Clusters The Beijing cluster contains the largest number of firms (177), followed by Shanghai (160), and Shenzhen somewhat smaller (126). Most of the firms are small. Of the 53 percent of the sample that answered these questions, almost 66 percent employ fewer than 100 employees, and around a quarter of them were medium-sized firms (more than 100 but less than 500 employees).Very few firms were large enterprises. Specialization is dominated by the health, diagnostics, and equipment sectors in all three clusters, whereas the activity is relatively low in other sectors (agriculture, chemicals, or services sectors). There is not a great difference between clusters in terms of specialization, even though the Shanghai cluster is the most specialized (health- and equipment sector) and also seems to be strongest when it comes to services (Table 6.4).

119


Beijing

Heilongjiang

Jilin Liaoning

Xinjiang Gangsu

Inner Mongolia Hebei

.

Sh

Qinghai

i

ggx Shan

Tibet

Tianjin

i

gx an

Ningxia

Shangdong

Henan Anhui

Jiangsu

Shanghai

Hubei Sichun

Zhejiang Guizhou

Yunnan

Hunan

Guangxi

Jiangxi Fujian

Guangdong

Taiwan

Hongkong Hainan

Figure 6.2. The geographic distribution of biotech firms in China Source: Company database in www.genetide.com (there are 1375 biotech firms in the database).

Table 6.4. Type of sector specialization of biotechnology firms: Beijing, Shanghai, and Shenzhen, 2003

Health/pharmaceuticals Equipment sector Agriculture Chemicals Diagnostics and reagents Services Other No answer Total

Beijing area

Shanghai area

Shenzhen area, Guangdong

44 27 10 2 32 7 9 46 177

48 23 2 7 14 11 6 49 160

43 8 10 5 30 4 10 16 126

The ownership patterns appear to show some differences between the three clusters. Beijing has a lower proportion of domestic privately owned enterprise than Shanghai and Shenzhen, and higher proportions of foreign-owned enterprises (where foreign owned includes both wholly 120

The Genesis of Biotechnology Table 6.5. Ownership type for firms, Beijing, Shanghai, and Shenzhen Beijing

Shanghai

Shenzhen

Number of companies

%

Number of companies

%

Number of companies

%

92 17

30.6 16.5

50 17

31.3 10.6

53 5

42.1 4.0

26

20.0

21

13.1

7

5.6

37

27.1

57

35.7

52

41.3

5 177

5.9 100.0

15 160

9.4 100.0

9 126

7.1 100.0

No answer Joint venture (foreign & domestic) Wholly foreign owned enterprise Mingying (private & collective) enterprise State-owned enterprise Total

owned and joint venture). This is due to the need for foreign firms in particular to be close to central government, in order to gain access to information on government plans and regulations, and also to take advantage of the greater R&D capacity in the Beijing area. Alongside this, the proportion of purely state-owned enterprises is lower in Beijing than the other two locations (Table 6.5). It is noteworthy that in two of the clusters (Beijing and Shenzhen) most large companies are domestic, whereas the opposite pattern prevails in Shanghai. Small- and medium-sized firms dominate, even in the foreignowned part of the industry.1 The biotechnology clusters are thus not dominated by large multinational companies (MNCs), but rather stem from entry into the clusters by indigenous firms.2

Founders and Location What these clusters all illustrate is a symbiotic relationship between the local government and firms, which has formed the backbone for the positive feedback that has existed between local government initiatives and entrepreneurial responses. As Qian (2003) points out, this mutually supportive relationship in Chinese provinces stands in contrast to the ‘grabbing hand’ of local government vis-a`-vis private enterprises in postreform Russia. The difference is not attributable to more established private property rights, nor to a well established rule of law that restrains government predation. Qian attributes it to the greater decentralization in China that allows local government to act with greater autonomy from the central state and, through its fiscal contracting system, to retain any

121


additional revenues it raises above the fixed share of revenue it remits to the central government. Jin, Qian, and Weingast (2001) established that following the reforms of the 1980s, provincial budgetary revenues and provincial expenditures became far more highly correlated than previously, with far lower extraction by the central government of any additional revenues raised at the provincial level. Provincial government in China therefore has fiscal incentives to support productive local businesses, benefiting directly from the revenues they raise. This decentralization and autonomy of provinces from the central government, while assisting in enhancing the relationship between the local government and private business, has also meant that provincial governments compete with each other to attract businesses. The huge rivalry between provinces, towns, and local districts leading to duplicate investments in infrastructure has also led to excessive red tape and trade barriers as people and businesses cross from one administrative region to another.3 The creation of new companies into the clusters took off from the mid1990s, partly as a response to the policies designed to encourage entrepreneurship and create incentives for fostering high-technology start-ups. One of the questions is whether the role of the provincial governments in generating clusters has differed between the three regions. There are basically two groups that have been chiefly responsible for the foundation of biotechnology companies in all three regions: the government and scientists returning from research abroad. The government took the lead in setting up its first biotechnology commercialization base in Shenzhen in 1993. The first start-up was the firm, Kexing, set up in Shenzhen. This grew out of China’s first interferon laboratory in 1987. In 1993, the National Science and Technology Committee founded Kexing in the Shenzhen Science and Industry Park to commercialize this new interferon. This was the first dedicated biotechnology firm in China, and was followed by an inflow of companies. Similar processes occurred in Beijing and Shanghai. Other new companies were founded by returning scientists from abroad. Returning scientists have made their location choices depending on their previous residency, or where they went to university. Since most of these returning scientists started their studies in the leading universities, they tended to return to either Shanghai or Beijing. In addition, guanxi, or the local network of connections, played a key role in the choice of location for these biotech firms. Locations with more flexible market mechanisms and government support for their venture also seem to have

122


been favored. The latter factors are particularly emphasized by the biotechnology firms located in Shenzhen, partly influenced by the presence of the Special Economic Zone (SEZ) in that area. That helped Shenzhen— as compared to other non-SEZ cities—to develop less bureaucratic local government and more flexible market mechanisms.4 The Shenzhen cluster has been seen especially as the location for commercialization of research which originated in Beijing, Shanghai, or overseas. For example, Proexact Biotech represents a typical case of a firm founded in Shenzhen by a returning scientist. It is a diagnostic equipment firm, founded in San Diego in 2000 with an offshoot founded in Shenzhen in 2001. The main founder was an overseas Chinese, Liu Zhong, who worked as chief scientist in several biotechnology firms in the USA. Attracted by the large market and low-labor cost in China, Liu Zhong and his team set up a branch of Proexact in Shenzhen in 2001. Proximity to universities and local research was not important in this case, as the technology for manufacturing their diagnostic equipment was relatively mature and the main R&D activity of Proexact was based in San Diego. They were attracted to Shenzhen by its relatively efficient labor force as well as a less bureaucratic local government and a more sophisticated bank sector. The local government actively supported start-ups by offering seed funding (Rmb100,000) for returning scientists starting their own business. In other words access to technology was not an issue, whereas ease of bureaucracy, access to capital, and the connections to enable the company to negotiate its way through government regulations were key factors. Beijing, in contrast to Shenzhen, has developed as the science and technology research center for China over the last twenty years. It has over 40 percent of national key laboratories in biotechnology, 25 percent of national engineering centers related to biotechnology and 15 percent of the country’s biotechnology research organizations. A third of biotech research projects supported by the National Nature Science Fund were undertaken in Beijing and a half of the biotech research projects supported by the ‘863 plan’ in 2002, well ahead of the other two locations as a research center (see Table 6.6). In addition Beijing, along with Shanghai, has a high concentration of top universities, compared with Shenzhen (see Table 6.7). There is some indication that the relative strength of biotech research in the Beijing and Shanghai areas is reflected in the types of firms that are created there. The Shanghai and Beijing firms have higher research intensities, defined as numbers of R&D staff, than firms in Guangdong (Table 6.8).5 There are reasons though to believe that R&D intensity is still relatively low among

123

The Genesis of Biotechnology Table 6.6. Biotechnology research projects supported by the National Natural Science Fund by location in 1999 Common projects

Beijing Shanghai Guangdong Others

Key projects

Number

Amount (RMB million)

Percentage of total number

Number

Amount (RMB million)

Percentage of total number

326 160 103 640

43.4 20.5 13.3 72.7

27.3 12.9 8.4 51.2

10 5 2 6

9 4.6 2.1 5.5

42.6 21.5 9.9 26.0

Source: 1999–2000 Beijing Bioengineering & New Pharmaceutical Industry Development Report, Beijing Bioengineering & New Pharmaceutical Industry Office and Beijing Science & Technology Committee, and Lei (2004).

Table 6.7. Comparison of universities and student numbers: Beijing, Shanghai, and Shenzhen Universities in the three clusters in 1999

Beijing Shanghai Shenzhen

Number

Student numbers

Graduate numbers

64 41 2

235,140 186,307 10,568

50,307 40,316 2,146

Source : Lv et al., High-tech Industry Policy and Its Practise, China Development.

Table 6.8. R&D staff in relation to total employees

0–10% 11–50% 51–75% Above 75% No answer Total number of firms

Beijing

Shanghai

Guangdong

9 21 1 5 141 177

47 31 10 4 68 160

— 67 9 — 50 126

firms in the cluster, compared with biotech firms in developed country clusters. One indication is that average R&D expenditures in the pharmaceutical sector are around 3 percent of sales revenue in China, compared with an average of 15 percent in most developed countries. However, Beijing has not been as accommodating a location for researchoriented firms to develop, despite its strong research base. The Beijing Huada Genomics Institute, which was responsible for the 1 percent of the Human Genome Project completed in China, was founded in Beijing in

124


1999, by a returning scientist, Yang Huanming. He had studied and worked in Denmark, France, and the USA and had the backing of the Chinese Academy of Sciences (CAS), but felt hampered by CAS regulations and lack of funding in Beijing and set up a branch of the BGI in Zhejiang province in 2001 to take advantage of its lighter regulation and supportive funding from the local government. The Beijing provincial government has a record of having a heavier hand in regulating firms and has attempted to direct firms to specialize within particular industry parks in the region and has not been as responsive to easing licensing or regulatory procedures as have the governments in Shanghai or Shenzhen. The strength in the Shanghai cluster resides in it having a broader and more diversified industrial base than either Beijing or Shenzhen. Shanghai houses leading universities, such as Fudan and Jiaotong, biotech research institutes, such as the Shanghai Life Science Institute of the Chinese Academy and the South Center of the Chinese National Human Genome, as well as being the traditional center for the Chinese pharmaceutical industry. The Shanghai cluster has benefited from having a better-developed infrastructure, a larger and more diversified business market and government preferential policies toward the biotech industry that have attracted firms into this cluster.6 Well-developed industries and strong biotech research performance in Shanghai have enabled biotech firms in Shanghai to obtain research inputs relatively easily.7 A typical start-up into this cluster was the foundation of United Gene in 1997 by Xie Yi, a professor at Fudan University and Mao Yumin, a former Dean of the school of Life Science at Fudan. They had the support of Shanghai’s Science and Technology Committee, which eased them through the regulatory requirements, and subsequently gained the backing of private real estate business, renaming the company Shengyuan Gene Development. In addition the local government has acted particularly forcefully to establish and promote the National Shanghai Biotechnology and Pharmaceutical Industry Base (NSBPIB) within the Zhanjiang High-tech Park in Shanghai (Pudong Park). It was set up in 1996 with backing from the Ministry of Science and Technology, the Ministry of Health, the Food and Drug Administration, the CAS and support from the Shanghai government—thus with support from both central and local government. It has influenced the movement of key research institutes to resettle in the Park, such as the Chinese Human Genome Center, the Institute of Materia Medica and the Shanghai Chinese Medical University and its affiliated hospital. By the end of 2003, eight national level research institutes or centers had been founded or migrated to the NSBPIB and seven leading

125


biotech research centers in Shanghai were relocated to the NSPBIB. There has also been the usual panoply of preferential policies directed to the biotech industry—lowering entry barriers, simplifying bureaucracy, exemption from taxes, encouraging the sharing of research resources, subsidizing investments, and attempting to attract VC into Pudong Park. On the face of it, it has had some success in attracting around 160 pharmaceutical and biotechnology companies into the Park by 2004. However, many of these firms are still state-owned or have substantial state involvement and it is questionable how far real market signals are being allowed to develop in these clusters and how much self-organizing of entrepreneurs is taking place. One criticism meted out against the NSPBIB in particular is that research has become overly ‘plan’ oriented, being dependent on phase-released funding by the government, and that such research is not as able to keep pace with market demand or market developments. The government has overly emphasized technological development at the expense of marketing or other support activities that would usually accompany the development of a cluster in a developed, more market-oriented country. There has been the development of some new technology platforms, on the back of returning scientists from the USA, but without the required people in marketing and law.8 An additional criticism of the NSBPIB, which can also be levied at all the biotech clusters, is the failure to allow firms to exit from the industry and for their resources to be absorbed or taken over by more efficient firms. This is particularly acute in the NSBPIB and leads to a lack of shaking out and restructuring which would enable the cluster to specialize and mature more efficiently. With firms not being allowed to fail, the risk-taking is borne by the State, which limits the development of market signals. If this degree of State involvement continues, this will stunt the growth and maturing of the cluster. To summarize, there are some differences in terms of the role of government between our three emerging biotech clusters and also in their various structures and the types of firms that are settling there. The Beijing cluster, having a stronger research base than the other two, appears to attract more foreign firms than the other two clusters, perhaps in order to be close to central government and to enable foreign firms to establish networks necessary to get round the various regulatory procedures. It does appear that government involvement and direction into the biotech industry parks, in particular in Beijing and Shanghai, is very substantial and inhibiting to the development of true market responses. The Shenzhen area benefits from having a longer legacy of greater market

126


orientation with its links to Hong Kong and the autonomy given to the SEZs, and there is more evidence of market forces operating in this cluster. However, the Shenzhen area has no history of substantial research in the biotechnology field and is dependent on its manufacturing labor pool, whose lower costs are eroding as the area develops. There does not appear to be a great difference in terms of specialization between the three clusters, which is perhaps due to the relative immaturity of these clusters, although there are signs that the Shanghai cluster is developing greater diversity in terms of provision of services. Overall the Shanghai cluster, while having a highly dirigiste government involvement for instance in the resettling of research institutes into Pudong Park and other such forceful tactics, nevertheless has a more diversified industrial base as well as research skills in the area on which to build its cluster, and has a lighter regulatory touch than does the Beijing region. It therefore stands the greatest chance of developing into a more sophisticated and marketoriented cluster over the next decade.

Conclusions The biotech clusters that we have described are the product of the interaction between various policy initiatives and a number of other factors. We have identified two main phases in the evolution of policy: from the mid-1980s to the mid-1990s when policies were mainly concerned with sowing the seeds of institutional reform, creating new forms of property rights, setting out strategic programes for the development of biotechnology and other high-technology sectors. This was followed from the mid- 1990s and onward by policies more focused on incentives to assist start-ups, attracting potential entrepreneurs back to China from abroad, and developing regional clusters around science parks. These policies have been attempting to compensate for some of the immaturities and weaknesses in the Chinese economy: the lack of support for high-risk ventures from the banking sector, the lack of VC and the relatively weak influence of foreign capital in this sector. So can we argue that the Chinese government—both central and provincial—policies have been successful in establishing these hightechnology clusters? How much state involvement is necessary in a transition economy, to overcome the obstacles of inadequate property rights, poor rule of law and insufficient protection for private property from an overpowerful state? There are some indications that in various respects

127


the Chinese State is managing their reform quite successfully, better than other transition states such as Russia for example, in allowing market mechanisms to infiltrate at the margin and to establish positive incentives for relationships between the state and private business (Lau et al. 2000). Thus considerable entry of new businesses does occur, which are for the most part private- or collectively owned enterprises. In parallel, there is a restructuring and phasing out, and greater autonomy or spinning-off of state-owned enterprises; and the price mechanism is being reinforced gradually from more marginal transactions to becoming the more prevalent allocative mechanism. It nevertheless remains a policy-driven country—driven and responding to central planning and technology policies. This has its positive aspects with a beneficial relationship between provincial governments and entrepreneurs, encouraging start-ups and using fiscal incentives, for both government and firms, to assist in that process. There is considerable competition between provinces in trying to attract and stimulate business in their region which has a positive effect in bidding down entry barriers and in creating a race to lower the regulatory burden on firms. The role of returnees in transferring their technical and managerial experience to all of these clusters in China has been crucial. Policy input has been significant as well in focusing on attracting entrepreneurial and scientific talent back to China from abroad, and a significant proportion of start-ups has been created by returnees. In terms of choice of location, they have been attracted by links to their previous universities (in Beijing or Shanghai) or by the greater ease of doing business, particularly in Shenzhen, but also in Shanghai. The proximity to central government has been a factor in locating near Beijing. The role of institutional reform and greater openness has been particularly important in Shenzhen. And the strong-arm of policy has been very significant in helping to construct the Shanghai cluster. The building up of the Zhangjiang High-tech Park is an example of powerful government being able to influence the relocation of major research institutes and other institutions which in turn have stimulated the creation of start-ups and the attraction of international R&D company laboratories there. However, there are also negative aspects to this degree of state involvement in the genesis and shaping of the cluster. One criticism is that the strong-arm of government has been overly powerful in setting up these seeds for clustering and too influential on the types of activity that become established at these locations. This represents a reversion to a

128


‘Plan mentality’ where everyone goes by the rules, which affects the way that business is set up and transacted, responding to the letter to government incentives rather than seeking out market-driven initiatives which are responsive to customer demand or international standards. At the other extreme, foreign firms that have entered these locations have done so to exploit the incentives, bringing sometimes poor ideas and ventures to the market in order to make money. The real issue will be how quickly this industry and these clusters can adopt international standards, allowing shake outs and exits of less productive firms to occur, enabling them to become integrated with the international market. Chinese professors are good clinicians but relatively weak on research management knowledge and skills. Venture capital is not yet knowledgeable enough to make good judgments along those lines, and the provincial governments inhibit closure of firms which would increase unemployment. There is insufficient specialization between the clusters—in part due to competition between the regions, the composition of the clusters as yet does not fully reflect the regional differences between the areas. The clusters are all sustaining businesses in similar areas and not yet being distinguished by their particular research strengths in Beijing or Shanghai or their manufacturing strengths in Shenzhen. The clusters are too technology oriented with insufficient business support and market orientation. This would apply particularly to the Beijing and Shanghai clusters with their stronger research strengths and greater government direction of technological development. There are some threats to cluster development—at present they benefit from great cost advantages in comparison with the USA or European high-technology clusters. However, labor shortages have already developed in the Shenzhen area and wages have been rising accordingly. They need their own hinterland and diversification, that is development of local supply skills as well as local markets to avoid being isolated coastal hubs, inferior in standards and not integrated yet into the international market, but not sufficiently developed or diversified to enable growth to occur in the more local adjacent regions. Only the Shanghai cluster at present shows signs of having this diversified hinterland and has the best chance of developing into a thicker, market-driven cluster. There are signs that the Shenzhen region, in developing the Greater Pearl River Delta Area, is conscious of the need to extend further beyond its coastal hub into a deeper and more developed hinterland, through its ‘9þ2’ policy to link the nine provinces in southern China to the two

129


special administrative regions of Hong Kong and Macau. Without this, the cluster relies on being a cheaper manufacturing and export station, which advantages will be eroded by rising wage costs and shortages of labor skills.

APPENDIX Data Collection The data below are compiled from various sources and websites and based on telephone and personal interviews with companies. Since there is no official directory of biotech firms in China, we collected the company data from a variety of sources. First, we looked at five biotech firm databases released on the Internet: database of the Ministry of Science and Technology (MOST) (2003a), and the Chinese Biotech Research Organizations and Researchers (2003) released by China Biotech Development Center of Ministry of Science & Technology, Life Science & Biotechnology Bureau of Chinese Academy of Science and Association of China Biotech Engineering in www.biotech.org.cn; Bio Yellow Page in www.bio-engine. com; company database in www.genetide.com; company database in www.cnm21.com; and company directory in www.yahoo.com.cn. Based on the data extracted from these websites, we constructed our own biotech company database of three locations: Beijing, Shanghai, and Guangdong. There are a total of 177 biotech companies in Beijing, 160 in Shanghai, and 127 in Guangdong. Data include date of foundation, numbers employed, numbers of R&D staff, ownership type, which we classify as state-owned enterprise, mingying enterprise, joint venture and wholly foreign-owned enterprise, specialization including biopharmaceutical, diagnostics, equipment, agriculture, chemical, technology service such as sequencing and others that cannot fall into these categories. We employed four categories of ownership in our database. The joint venture refers only to a company that involves both foreign and domestic capital. This definition is in line with the accepted understanding of joint venture. The mingying enterprise appeared after the Open Door Policy was introduced. It includes (a) family or private enterprise, (b) foreign enterprise including joint venture, (c) mingying keji qiye, (d) village enterprise, (e) joint stock enterprise, and ( f ) state-owned mingying enterprise, of

130


which property rights are still state owned but individuals or private organizations are responsible for operating the enterprise by form of lease, outsourcing, etc. (Xiao, Liang 2002). The Development of mingying Economy, the China Economic Times, November 2, 2002. This definition of mingying enterprise is somewhat broad. We excluded foreign enterprises from the mingying category and followed the remaining parts of the definition. Companies may be active in more than one sector, but we assigned each company to one sector only based on their main field of activity. We looked at companies’ websites to weed out nonbiotech companies. We then did telephone interviews to explore the data further and screen out nonbiotech companies. Furthermore, we did in-depth telephone/face-to-face interviews with some biotech companies and government agencies for information on the formation of biotech industrial clusters.

Table Appendix 6.1. Details of interviews used in the analysis Date of interview

Company/ Government agency name

Location

Interviewee

Feb 24, 2004

Pudong Biotech & Pharmaceutical Association/NSBPIB Development Co., Ltd. Shanghai Biotech & Pharmaceutical Association Shisheng Cell Technology Pacific Biotech Wanxing Biotech SHDNA Biotech Newgenco Bio Science Proexact Biotech LvHan Biotech Sea Forrest Biotech Jiena Bio-products Chuangshiji Modified Genetic Technology Huashengyuan Genetic Engineering Bolaote Biotech Blue Star Biotech Cy-tech Bio Science AGTC Genetic Technology Bioson Bioengineering Absea Biotech PaiteBoen Biotech Development Green Gingko Biotech Biosea Biotech

Shanghai

Ms Lin Hui

Shanghai

Mr Cao

Shanghai Shanghai Shanghai Shanghai Shanghai Shenzhen Shenzhen Shenzhen Shenzhen Shenzhen

— Ms Xu — Mr Pan — Dr Liu Zhong — — — —

Shenzhen

—

Shenzhen Guangzhou Shenzhen Beijing Beijing Beijing Beijing Beijing Beijing

— — — Mr Dong Xiaoyan — — — — —

Feb 24, 2004

Feb 26, 2004 Feb 26, 2004 March 1, 2004 March 9, 2004 March 11, 2004 Feb 27, 2004 March 4, 2004 March 4, 2004 March 4, 2004 March 9, 2004 March 9, 2004 March 16, 2004 March 16, 2004 March 17, 2004 April 5, 2004 March 22, 2004 March 24, 2004 March 25, 2004 March 25, 2004 March 26, 2004

131


Notes 1. Due to a sizeable share of missing data, this information should be cautiously interpreted. 2. This contrasts with the situation in Ireland (see Barry, Chapter 8) where foreign MNCs are large and dominate the cluster. 3. The Economist, November 20, 2004. 4. This point was stressed by Dr Liu Zhong, the founder of Proexact, a diagnostics firm based in Shenzhen. 5. Even though the numbers of R&D staff as a proportion of total employees in Beijing and Shanghai appear to be quite significant, these figures should be viewed as indicative, since no information is reported for more than 50 percent of the firms. 6. Interviews with Pacific Biotech and Wanxing Biotech. 7. Interview with DNA Technologies, a technology service provider in Shanghai. 8. Interview with William Keller, former President of Roche China and now consultant to NSPBIB.

132

7 The Emergence of a European Biotechnology Cluster: The Case of Medicon Valley Pontus Braunerhjelm and Christian Helgesson

Europe’s biotechnology industry entered a dynamic phase during the last decade, characterized by entry of new firms, increased research funding, and enhanced access to VC (Allansdottir et al. 2001). In fact, the latter part of the 1990s saw a doubling of European biotechnology firms. The steps toward a more vibrant European biotechnology industry were particularly prominent in Germany, Great Britain, France, and some of the Nordic countries with Sweden in the lead. Another manifestation of the revitalized interest in biotechnology is the emergence of potentially strong and internationally competitive biotechnology clusters. Some of these, for instance in Cambridge and Oxford in Great Britain, and in Stockholm (Sweden), have existed for a relatively long period, while others—most ¨ nich, Rhine/Neckar, and Rhineland, clearly in the German regions Mu and in Île-de-France outside Paris—are quite new. Thus, Germany, Great Britain, France, and Sweden host several promising sites of biotechnology agglomerations. Still, the industry is in a formative phase and a restructuring and shake-out is likely to change the future spatial distribution of the European biotechnology industry. In this chapter we will scrutinize why and how the Medicon Valley became a leading European biotechnology cluster.1 The antecedents to the emergence of Medicon Valley is argued to basically be representative for most European biotechnology clusters, even though each site has its own particular trajectory to the present position. Medicon Valley displays one characteristic feature which separates it from most other biotechnology

133

Emergence of a European Biotechnology Cluster

clusters; its geographic location stretches over the southern part of Sweden ¨ resund. The distance and the northern part of Denmark, known as O ¨ ) and Copenhagen in between the capital of southern Sweden (Malmo Denmark is less than 20 km. Thus, despite geographic proximity, depending on where agents are located in the Medicon Valley they may encounter differences in the institutional framework as well as in cultures and traditions. Presumably such differences may influence the functioning of the cluster. This makes Medicon Valley an especially interesting case since there is a possibility to compare the performance and dynamics of the cluster in a region where institutional differences do appear, simultaneously as the two countries share a number of other characteristics. Considering that the regions belong to separate national states with differences in language and culture, a shared history of approximately 134 years of war, and the fact that the regions have been separated by a small body of water—which changed in 2000 when Sweden and Denmark were connected by a bridge—the emergence of a common biotechnology cluster is an exceptional achievement (see e.g. Frank 2002; Nilsson, Svensson-Henning, and Wilkenson 2002; Palludan and Persson 2003). The region has been branded as Medicon Valley to highlight its stronghold in the life sciences—that is its prevalence of universities, hospitals, research parks, firms, service providers, and VC belonging to that sector. We will examine the process in which the Medicon Valley biotechnology cluster came into existence—the igniting spark—and the preemergence conditions. An additional aim is to picture the evolution of the cluster in the two nations and whether the dynamics of the cluster differs in the respective country. Finally we address the role of policy in these processes. Analytically we adopt the three stage framework outlined in Chapter 1. The analysis is based on officially available statistics and semistructured interviews representing universities, biotechnology companies, research parks, VC companies, local organizations, and authorities (see the Appendix). The next section describes the composition of the Medicon Valley and the history behind today’s cluster. Thereafter we examine the dynamics of the cluster with special attention toward differences within the two regions. Finally we discuss the role of policy in the evolution of Medicon Valley.

The Medicon Valley Biotechnology Cluster—Present and Past On the Danish side, Medicon Valley is geographically located to Hovedstadsregionen, West Sja¨lland, Lolland, and Bornholm in Denmark, while

134


Regions - Öresund Hovedstadsregionen (DK) W Själland, Lolland, Bornholm (DK) Skåne North East (SE) Skåne North West (SE) Skåne South (SE) Malmö Region (SE)

© SCB 2000

¨ resund region Figure 7.1. Map defining the O ¨ rsundsstatistik. Source: Sweden statistics and O

Ska˚ne—separated into three subregions—represents Sweden (Figure 7.1). The core of the cluster consists of the Danish region denoted Hovedstadsre¨ . Research in the natural sciences goes gionen and the region around Malmo ¨ a long way back in the Oresund region. The Lund University was already founded in 1666 and research in natural sciences has been carried out since at least 1680. The University of Copenhagen has an even longer history: education and research in law, philosophy, and medicine can be traced to 1479. The region also has a strong tradition in the agricultural, brewing, and pharmaceutical industries. The pharmaceutical firm Astra had been engaged in research in the region since 1913 while its Danish counterpart Lundbeck started shortly after World War I. It is also noteworthy that the first Swedish pure biotechnological firms appeared in this region. Both Bioinvent and Biora, founded in the 1980s, originated from different research projects at the Lund University. According to Andersson, Andersson, and Wichmann ¨ resund region was the origin of about 60 percent of the (1993) the O pharmaceutical firms in Scandinavia in the beginning of the 1990s.2

135


Obviously the setting for a biotechnology cluster was strong and there was no need for a substantial intervention in the first stage—that is to build a strong knowledge base in the region.3 Researchers in the region are also highly ranked with reference to the number of articles published in medical journals (Andersson, Andersson, and Wichman 1993). Although that research and knowledge accumulation in natural sciences had been extensive over a long period, the genesis of the Medicon Valley cluster is a relatively recent phenomenon. If one should try to date the igniting spark or triggering event that led to the Medicon Valley cluster, the initiative in 1983 by Professor ¨ rjel and Lund univerSture Forseń, together with county governor Nils Ho sity director Nils Stjernquist, to start the Ideon Science and Technology Park just adjacent to Lund University, seems a strong candidate.4 In that period the southern part of Sweden suffered from deindustrialization, stagnation, and declining economic performance. Traditionally the Ska˚neregion—despite the presence of an old and prestigious university—was dominated by a low- and medium-technology industry. The idea to establish a science and technology park was imported from the USA where similar parks hosting technology intensive start-ups already existed. The objective was to embed start-ups in a favorable environment, including relatively cheap office and laboratory space, a basic infrastructure, closeness to competitors as well as firms with complementary competences, for example services related to patenting, VC, legalities, accounting, and public relations. The creation of such a dynamic milieu, conducive to learning and interaction between agents, was expected to facilitate and enhance commercialization of academic research. The initiation of Ideon however had a much wider ranging influence than was expected: it sparked a second wave of establishment of research parks in the region. Most well-known examples are Symbion Science Park in Copenhagen, CAT Science Park in Roskilde, The Hoersholm Research Center, Krinova Science Park in ¨ Science Park. In the original Ideon Kristianstad, and Medeon Malmo science park, ICT-, biotechnology-, and functional foods firms are dominating. Most of the firms have close ties to the research departments at Lund University. Even though the 1983 initiative can be seen as the igniting spark, the report referred to above constitutes another important milestone in the evolution of the Medicon Valley biotechnology cluster. This study became instrumental in identifying the number and variety of firms and other organizations in the region, their quality in comparison

136


with other cluster environments, as well as the interdependence and interrelations between them. Thus, the report served to imprint the awareness by different agents of a cluster environment that stood up well in comparison with other regions. But the report also emphasized future challenges due to the acceleration in globalization and regionalization.5

The Cluster Initiative Building on the increasing awareness of a strong foundation in biotechnology, the first steps were taken to actually form a cluster in the region. Who were the prime movers propelling this evolution? And what role did policy play? In the beginning universities and organizations were the prime movers. As late as 1995, five universities in the region began discussing how cooperation between the universities in the two countries could strengthen the scientific knowledge base and also make education and research more effective and competitive. A couple of years later (1998) this resulted in a joint effort by nine universities to create what is now ¨ resunds University (O ¨ U). Basically O ¨ U formalized a collaborcalled the O ation that was already going on between these universities. Still, a platform was established which deepened, widened, and structured the ¨ U has grown to fourteen collaboration between universities. Presently, O universities that partly coordinate research and higher education and share data bases and laboratories. Another decisive step to strengthen and deepen the integration of the region was the agreement in the mid¨ resund region by 1990s to geographically connect the two sides of the O building a bridge. The discussions primarily between universities also spread to other groups. Representatives from the university sphere presented their ideas to industry and their organizations. The discussions focused on areas where cooperation between academia and industry could help to empower the competitiveness of the region. The conclusions were—which ¨ U and those reported parallels those that led to the establishment of the O by Andersson, Andersson, and Wichmann (1993)—that life science/ biotechnology was an area where the region had the strongest future potential with an internationally competitive edge. More specifically, the areas of highest commercial opportunities were identified as diabetes, immunology and inflammation, neuroscience, and cancer. The result of these discussions showed up as a loosely formalized network in 1997 under the name Medicon Valley Academy (MVA). It started

137


off as a three-year project with a budget of e2 million, of which Copenhagen university and EU accounted for a third each, while the remaining third was split between other organizations. The board of MVA consisted of (and still does) five representatives from the local universities, three from research hospitals, three industry representatives, and two board members affiliated with the regional development organizations. The headquarters of MVA is located in Copenhagen. The main objective was to constitute a platform for continuous meetings and discussions regard¨ resund region. As time went by the MVA ing the development of the O came to engage an increasing number of individuals linked to universities, biotechnology companies, VC companies, hospitals, authorities, and research parks. Since 2000 MVA has been financed through member fees. ¨ resund In a parallel move to the more formalized structure of MVA, the O University decided to expand its collaboration and improve the links to industry. To achieve this end, a network was formed and several ¨ resund IT Academy, O ¨ resund Food Network, O ¨ resund platforms—e.g. O ¨ resund Logistics—were created. Representatives Environment, and O with different backgrounds were encouraged to meet with the purpose of solving common problems, creating opportunities and finding synergies across firms and sectors. These platforms were also linked to the MVA.

Second-Stage Deepening of Medicon Valley As shown above, the universities played a crucial role in the formation of the Medicon Valley cluster. The universities took the first steps toward closer cooperation and denser links between the different academic institutions on the Danish and Swedish sides, followed by a plethora of measures to improve and deepen the interaction between universities and industry. In the second stage, participation of local politicians and representatives from local authorities became more evident. Policymakers and local governmental bodies started to realize the potential of an underutilized knowledge base. A key issue for the regional organizations was therefore to market the region in order to attract both national and international investment. The marketing strategy involved not only MVA, but also Copenhagen Capacity and the recently formed regional organization Position Ska˚ne. The number of service providers and VC firms started to increase in the region. By 1995 there were nine VC firms in the region which increased to

138


thirty-three in 2002. Moreover, VC firms specialized in seed capital investments also increased: from three in 1995 to twenty-four by 2001. Hence, a support structure began to develop that complemented the already strong presence of research organizations and, to some extent, firms. Most of the VC firms are located on Danish territory in the Medicon Valley cluster. The expanding local venture capitalist industry also gradually increased their competencies. They also became more visible in the region and engaged in the region’s development outside their roles as venture capitalists, for instance by participating in entrepreneurship courses at the universities. This development nurtured Medicon Valley’s reputation as an internationally leading biotechnology cluster. The main reason to start a new venture or relocate into the Medicon Valley region is claimed to be the self-reinforcing centripetal forces of successful high-technology clusters: the critical mass of universities, large pharmaceutical firms and hospitals in the region that has created a knowledge base of a highly skilled labor force which serve to attract both large and small firms, as well as labor with skills in the relevant areas. The connectivity between companies, academia, and hospitals seems fluid. Firms are often involved in university research programs through, for example, employees splitting their time between academia and industry. Frequently, they have special employment schemes for Ph.D. students and/or post docs. This glue is claimed to create a flow of information across different agents in the cluster, thereby Networks/partnering Most partnerships are intra-MV Number of corporate partnerships (Per company) 3.5 3

Corporate partnerships

2.5

Strong basic research collaboration within MV Number of corporate partnerships (Per company) 3.5 Unspecified 3

Academic partnerships

2

2 1.5

1.5

0

U

A

th

or N

ni

te

R es

to

M

er m

A th or

M V m e R r i U ni est ca te d of E K U i O th ngd er om O th Sw ed er e D en n m ar k A si a O th er

0 ic a f d E O Kin U th gd e O r S om th w er e D den en m ar k A si a O th er

1 0.5 V

1 0.5

N

Drug licensing Clinical research Basic research

2.5

Figure 7.2. Interlinks between firms and academia in Medicon Valley Note: Data for 29 survey respondents. Excluding large pharma. Source: Boston Consulting Group (2002) survey.

139


fostering a mutual elevation of competence, but also reducing the economic risks in different ventures. Not least, the latter aspect is of importance for the dense interactions in the Medicon Valley cluster—to transfer uncertainty into quantifiable economic risks.6 The risk of running a business in the biotechnology industry is considerable. According to the respondents, it costs somewhere between e55 million and e120 billion for a drug to reach the market, including costs for research, development, clinical testing, and marketing. Still. The probability that a drug will reach the market is very small. The extensive cooperation across agents implies risk-sharing and admits access to a larger knowledge base (Figure 7.2). Small biotech companies frequently cooperate with larger companies, or do contract drug-discovery research.

Policies to Support and Strengthen Medicon Valley We have shown how Medicon Valley evolved out of a strong knowledge base and local initiatives to enhance and diversify the region’s competitiveness and diversity. The issue we raise in this section concerns the current state of the Medicon Valley biotechnology cluster, the integration of the binational cluster, and the extent to which policies have responded to the needs of the region. At the aggregate level, both Denmark and Sweden have strengthened their revealed technological advantage in biotechnology during the 1990s, particularly Denmark. Moreover, despite less R&D resources in absolute numbers, Denmark reveals a stronger patent performance than Sweden between 1990 and 2000, irrespective of whether measures of granted or applied patent applications are used. Thus, data indicate that Denmark has gained a competitive edge over Sweden at the aggregate level.7 A specific feature of Medicon Valley is its binational structure. Altogether Medicon Valley has approximately three million inhabitants of which 35,000 work in the sectors of biotechnology, pharmaceutical, and medical technology. The different sectors overlap and a distinct separation between these areas is not possible. Approximately there are about 116 biotech firms located in the cluster, where an estimated 70 percent (82 firms) are located on the Danish side. These are complemented with 70 firms in the pharmaceutical industry, 130 in medical technology and about 15 clinical research organizations, while the number of service providers and investors are in the range of 250.8 As already mentioned, the academic system consists of 14 universities, where 140,000 students are enrolled and 12,000 scientists are employed.

140

Emergence of a European Biotechnology Cluster Table 7.1. Comparison of volume of biotechnology-related articles from three regions Region

Number of publications

Per thousand capita

15,000 15,000 38,000

7 5.5 6.5

Stockholm–Uppsala Medicon Valley Boston

Source: DTU; Medicine; Boston Consulting Group analysis and estimate.

Table 7.2. Number of citations in biotechnology-related articles Region Boston Bay Area (San Francisco) Medicon Valley Stockholm–Uppsala Paris East England (Cambridge–Oxford) Mu¨nchen

Number of citations

Per thousand capita

180,000 135,000 50,000 45,000 90,000 50,000 25,000

16 15.5 11 10.5 10 12 8

Source: DTU; Medicine; Boston Consulting Group (2002) analysis and estimate.

Every year approximately 1,500 Ph.D.s are conducted, dominated (75 percent) by the faculties of natural science.9 The strong academic position is also supported by the impressive number of published articles and ¨ recitations in biotechnology-related research that originates from the O sund region. Medicon Valley is just behind the Stockholm–Uppsala and Boston regions with respect to the number of published articles per capita (Tables 7.1 and 7.2). In addition, the region hosts eleven university hospitals out of a total of twenty-six hospitals.10 From the beginning, there was also a considerable presence of large players located in the region—AstraZeneca (Sweden), Novo Nordisk (Denmark), Leo (Denmark), Lundbeck (Denmark), Ferrosan (Denmark), and Ferring (Denmark).11 The four most R&D-intensive firms—AstraZeneca, Novo Nordisk, Lundbeck, and Leo—employ the major amount of researchers in the private sector. Out of approximately 5,000 researchers, these four firms account for 4,000, of which 3,000 can be found on Danish ¨ rkbacka 2002). Another sign of the region’s accumulated strength soil (Bjo is that some international flagship firms have located part of their activities in the region such as Maxygen and Biogen (both from USA and located in Denmark). There has also been a strong regional entry in the biotechnology

141

Emergence of a European Biotechnology Cluster 18 16

Number of start-ups

14 12 10 8 6 4 2 0 1995

1996

1997

1998

1999

2000

2001

2002

Year

Figure 7.3. Number of biotech start-ups in Medicon Valley, 1995–2002 Source: Medicon Valley Academy.

Table 7.3. Number of firms and employees in Medicon Valley, 2002 Medicon Valley Number of firms Sweden Biotech Meditech Pharma Total

34 39 13 86

Denmark 81 90 58 229

Employees

Employees/Firms

Sweden

Denmark

Sweden Denmark

618 1,580 1,975 4,173

2,359 7,363 17,316 27,038

18.2 40.5 151.9 48.5

29.1 81.8 298.6 118.1

Firms established 1996–2002, employment effects and employees per new establishment Firms established 1996–2002 Percentage of total

34

74

39.5%

32.3%

625 15.0%

1,531 5.7%

18.4

20.7

37.9%

17.5%

Source: Bjo¨rkbacka (2002).

sector up to the year 2000 (Figure 7.3). After 2000, the number of entrants started to dwindle but has recently begun to increase. Regarding the dynamics of the Medicon Valley, the Danish part seems to fare better then the Swedish, judging from the size, composition, and geographic distribution of biotechnology/life science firms (Table 7.3). Taking all sectors into account, employment is almost seven times larger in the Danish part of Medicon Valley, whereas the corresponding figure in

142


the biotechnology sector is about four times larger. A huge difference can also be noted when it comes to the number of firms and start-ups in the respective country’s part of the region. There are signs of a catching up process on the Swedish side, as suggested by a larger share of firms established in the period 1996–2002 and a higher percentage share of employees in those recently started firms, as compared to Denmark. Still, employment growth was somewhat faster in the newly established Danish firms (21 versus 18 employees). There are no self-evident explanations to the seemingly different pattern of performance in the respective part of Medicon Valley, and it is far too early to draw any definitive conclusions. For instance, both nations are traditional welfare states and are ranked as the two countries with the highest overall tax pressure in the world (public expenditure as percentages of GDP exceed 50 percent). Yet, some distinctive differences between the national parts of the cluster can be detected. First, the Danish part of the cluster comprises Copenhagen, Denmark’s capital, with a natural concentration of resources— political, financial, and human—that serves to reinforce a structural change toward more knowledge intensive production segments. The larg¨ —is considerably smaller, being est city on the Swedish side—Malmo Sweden’s third city in terms of population. In addition, communication is facilitated due to the large international Danish airport (Kastrup), offering direct flights all over the world. The advantage of smooth traveling possibilities is important, but the Swedish disadvantage should have almost vanished after the completion of the bridge. As shown in Table 7.4, a larger share of the labor force has a tertiary education in the Danish part of Medicon Valley. This is mirrored by a much larger share of employment in the high-technology sector which exceeds 20 percent, but is a mere 4 percent in Ska˚ne. Correspondingly, Ska˚ne’s traditional specialization in manufacturing production is demonstrated by the large share still working in that industry: about 20 percent of the labor force as compared to 10 percent in the Danish part of Medicon

Table 7.4. Medicon Valley labor force with tertiary education, 1999 Labor characteristics of labor force (25–64) Tertiary education Employment in high-technology sectors Employment in manufacturing sectors

Denmark (%)

Sweden (%)

24 20 10

21 4 20

Source: Bjo¨rkbacka (2002).

143


Valley. In general, a stronger specialization in high-technology sectors serves to facilitate absorption of new knowledge and a restructuring toward new, science based, industries. It also implies that there is a pool of labor more likely to have the competencies demanded in high-technology sectors. At the firm level, labor costs are about the same in the two countries. However, in Denmark 90 percent of total labor cost is wages while the share in Sweden is only 58 percent. The difference is due to high indirect labor costs in Sweden, of which a large part is a pure tax not related to pensions or other benefits tied to the employee.12 Hence, from an individual’s point of view, disposable income is higher in the Danish part of the Medicon Valley cluster which makes it a more attractive site for labor then the Swedish part. In addition, Denmark’s labor market regulation is regarded to be quite flexible. As ranked by the World Economic Forum, the labor market flexibility in terms of hiring and firing is about twice as high in Denmark as in Sweden, which should be attractive for employers. It is also noteworthy that Denmark was the first country to develop and adopt a specific regulatory regime for the biotechnology sector. The propensity by Danish policymakers to react to the needs of the biotechnology sector gave them a certain advantage as compared to many other countries, particularly neighboring Sweden, and helped to diminish sectorspecific problems. This advantage will, to some extent, be reduced as a harmonization process is under way in the EU area after 2004. One outcome of the greater attention paid by Danish politicians to the biotechnology sector is claimed to be highly competitive conditions for clinical tests of drugs, which normally is an administrative burdensome and costly ¨ rkbacka 2002). procedure (Bjo Part of the explanation for the differences in performance on the Danish and Swedish sides of the cluster may also stem from variances in industry structure and different attitudes and culture with regard to entrepreneurs and small firms. Sweden has been dominated by large firms for almost a century, whereas Denmark has a tradition of hosting small- and mediumsized firms. On the other hand, Sweden has had a strong pharmaceutical sector with successful and large firms like Astra, Kabi-Vitrum, and Pharmacia being important players on the world market for many years. A substantial part of these firms had production and research units in the Ska˚ne region and even though the firms have either been acquired or merged with other international firms, much of the biotechnology clusters in Ska˚ne originates in those firms. Previous experience and links to those firms should also facilitate interaction and contacts with other firms and internationalization.

144


Conclusions The process toward the Medicon Valley biotechnology cluster stretches over a long period of time. The knowledge base was already in place, but the diffusion of knowledge outside the universities was relatively weak and the integration between the Danish and Swedish part of the cluster by and large nonexistent. Even today the degree of integration between ¨ resund bridge has considthe two countries is unsatisfactory, albeit the O erably enhanced mobility and communications within the Medicon Valley. Despite the improvement in recent years, almost 50 percent of the biotechnology firms report that they do not have any cross-border ¨ rkbacka 2002). links (Bjo The universities have played a decisive role in creating a common ¨ resund region, which subsequently evolved platform and network in the O into the Medicon Valley cluster. Over time firms and their organizations were also involved in developing industry–university collaboration and interlinks. The process gained momentum as participants from different sectors became aware of the mutual interest and the market opportunities in doing so. The last group to join the process of regional awakening were politicians and public sector organizations (excluding the universities). An important part in the development of the region was the conceptualization and marketing activity of Medicon Valley. That made the region’s potential in biotechnology, based on the size, multitude and variety of factors, more visible to outsiders (and insiders). In terms of policy, a few challenges are obvious for high-tax nations such as Denmark and Sweden. A particular feature of the Medicon Valley cluster is the binationality which prompts that tax and social legislation becomes harmonized. Until recently employees of Danish firms living in Sweden had their income taxed according to the Danish tax code (lower than in Sweden) but could enjoy the Swedish welfare schemes (more generous than in Denmark). These irregularities are likely to hamper the future integration of the cluster. More important is biotechnology’s global position. In Sweden some of the superstars in biotechnology have been offered tempting research environments and salaries in the USA. A relocation of such individuals can be expected to have severely detrimental effects on the cluster. There are also policy differences with respect to education, labor market regulations, and sector specific needs that may impact the development of the Medicon Valley biotechnology cluster. In short, Danish policymakers seem to have responded earlier and more forcefully to support industry in

145


general, and the biotechnology sector in particular. The similar sensitiveness cannot be observed on the Swedish side, or seems to have developed much later. Such supportive measures do not imply subsidies, but rather to secure the supply of skills, reduce the red tape burden, and promote wellfunctioning markets. Therefore, a world-class knowledge base must be complemented by an institutional framework and a business climate that is strong enough to match other potential and emerging biotechnology clusters. In the Swedish part, firms predict problems in recruiting international experts and attracting international investment with reference to tax issues. International experts can choose regions where they can keep a larger portion of their salary and international investment may prefer regions where investment in R&D will signify a tax deduction for the investing company, as in England and France. Hence, not only in Europe, but also worldwide, is the biotechnology industry in a very formative stage where a shakeout will take place sooner or later. Consequently, some of the emerging clusters—such as Medicon Valley—may not have the same strong position in a decade or so. The competition for talent, firms and finance is intense, and regions that manage to foster a competitive environment have the greatest possibilities of developing viable, future biotechnology clusters.

Notes 1. When we refer to biotechnology the following areas are included: pharmaceuticals and drugs, agronomic biotechnology, environmental biotechnology, biotechnological equipment and services, food/health-products, and bioproduction (molecules or microorganisms). See Vinnova (2001). 2. About 15 percent of the Swedish biotechnology firms have located in the Swedish part of Medicon Valley (ISA 2004). 3. This differs markedly from the other cases analyzed in this volume: the Chinese biotechnology industry (Chapter 7), the Irish ICT industry (Chapter 8), or the Israeli VC industry (Chapter 9). ¨ rjel was also one of the founding fathers of Malmo ¨ hus Invest, Sweden’s 4. Nils Ho first VC firm. 5. The respondents also claim that the report was the first step in the process ¨ resund-bridge connecting Ska˚ne and that finally led to the building of the O Denmark. 6. In the entrepreneurial literature this has been defined as a key ability of entrepreneurs (Knight 1921).

146

Emergence of a European Biotechnology Cluster 7. 8. 9. 10. 11. 12.

See Allansdottir et al. (2002) for additional details. Medicon Valley Academy (www.mva.org). ¨ resundsuniversitetet. O Medicon Valley Academy. As reported by Frank (2002). About 20 percent of indirect labor costs are estimated to be a pure tax.

APPENDIX ˚ berg, The following actors have been interviewed (in order of participation): Ulf A ¨ derstro ¨ m, Lund UniPosition Ska˚ne, Claes Wenthzel, Perbio Science AB, Bengt So ¨ resundsuniversitetet, ¨ rgen Wieslander, Wieslab AB, Bengt Streijffert, O versity, Jo Kurt Nilsson, Glycorex Transplatation AB, Per Belfrage, Biomedicinskt Centrum, ¨ ller, Ideon Center, Birgitte Lund University and Medicon Valley Academy, Hans Mo ¨ ld, Medeon Thygesen, Biogen Manufacturing ApS, Charlotte Munck af Rosenscho ¨ Science Park, Gudmundur Kristjansson, Region Ska˚ne, Ha˚kan Nelson, MalMalmo ¨ hus Invest AB, Bjo ¨ rn Quistorff, Panum Institute, University of Copenhagen, mo Martin Bonde, Combio A/S, Peter Buhl Jensen, TopoTarget A/S, Elliot Goldstein, ¨ resund Healthcare Management A/S, Henrik Lawaetz, Maxygen, Niels Mengel, O Scandinavian Life Science Venture, Thomas Varming, Neurosearch A/S.

147

8 The Emergence of Ireland’s ICT Clusters: The Role of Foreign Direct Investment Frank Barry

Ireland is the most Foreign Direct Investment (FDI)-intensive economy in the EU, with around half of its manufacturing workforce and a higher than average share of services’ workers employed in foreign-owned firms. Its FDI inflows have become increasingly high-tech in nature. Only 12 percent of employment in foreign-manufacturing firms in Ireland in 1974 was in sectors classified as high-tech by the OECD. The current figure comes to almost 60 percent. The sectors accounting for the bulk of foreign-firm employment in Ireland are information and communications technology, pharmaceuticals, medical devices and internationally traded services such as shared services and call centers. Our focus in this chapter is on ICT, and specifically on computer hardware and software. There are around sixty foreign firms and seventy Irish firms in the hardware sector (which consists of office and data processing equipment (ODP) and electronic components), with foreign firms on average around ten times larger in employment terms. Market leaders such as IBM, Apple, Hewlett-Packard (HP), Dell, and Intel all have a substantial presence in the country. Ireland, furthermore, is reported to be the world’s largest exporter of software. Over 140 overseas software companies have operations there, including market leaders such as Microsoft, EMC, Oracle, Novell, Siebel, Accenture, Sun, and SAP. There are a further 600 Irish-owned software companies. Employment levels in the computer hardware and software sectors are depicted in Figure 8.1. Employment growth in both sectors was dramatic over the course of the 1990s, which became known as the ‘Celtic Tiger era’. We see also from Figure 8.1 that while employment in domestically owned

148

Emergence of Ireland’s ICT Clusters 25,000 20,000 Foreign hardware 15,000

Domestic hardware Foreign software

10,000

Domestic software 5,000

2000

1997

1994

1991

1988

1985

1982

1979

1976

1973

0

Figure 8.1. Employment levels in computer hardware and software in Ireland

computer hardware firms is negligible, employment in indigenous software firms has kept pace with that in the foreign-owned segment.1 Ireland exhibits a strong production and export specialization in these two sectors. Discussions of clusters, however, typically focus on much smaller areas than the Republic of Ireland, which, with a land mass of around 27,000 square miles, is slightly larger than the state of West Virginia. How clustered are the hardware and software sectors within Ireland? The first point to note is that economic activity in Ireland is much more clustered around the capital city, Dublin, than is the case in most of the rest of Europe. Hardware is even more tightly clustered around Dublin and software substantially more so, than is the case for the rest of industry and services. The Greater Dublin region, for example, accounts for around 40 percent of all industrial employment and industrial establishments in the state, but for about 50 percent of hardware employment and hardware firms, and these latter proportions are rising over time. Furthermore, the Dublin region accounts for over 80 percent of employment in software, compared to only 40 percent of aggregate services employment. While there are some other clusters—notably around the second tier of cities such as Galway in the west and Limerick/Shannon in the southwest—to a large extent our analysis may be taken to be of clusters in the Greater Dublin region.2 In successful late industrializers, according to Amsden and Chu (2003), the public authorities often play an important role in network development and, less controversially, in ensuring that the appropriate

149

Emergence of Ireland’s ICT Clusters

conditions are in place for the Marshallian agglomeration processes to take hold. Consistent with this, the notion of institutional learning—the ability of the public administration system to extract, accumulate, and use effectively the insights that become available to it—will emerge as a crucial element in the present story.3 This will be discussed primarily with reference to the actions and strategies of the Industrial Development Agency, IDA-Ireland. The chapter is structured as follows. The next section provides some background on Ireland’s FDI-oriented development strategy, and considers the constellation of factors that led to the emergence of the Celtic Tiger economy of the 1990s. Sections 3 and 4 focus in on computer hardware and software, which were among the sectors exhibiting the most rapid growth in this period, while Section 5 explores the set of factors, including the actions and strategies of the country’s development agencies, that allowed Ireland attain such a strong position in these sectors. In the concluding comments we discuss the type of clustering that can be thought to prevail in ICT in Ireland today.

The Emergence of the Celtic Tiger Economy of the 1990s Ireland missed out on the general postwar European boom, abandoning protectionism only in the late 1950s, some time after most of the rest of Western Europe. Among the key features of its belated move toward outward orientation was the adoption of a zero tax rating on profits derived from manufactured exports. This led to a rapid inflow of foreign enterprises using Ireland as an export platform primarily for sales into Continental Europe. While economic conditions improved substantially with the move to free trade, cemented by the Anglo-Irish Free Trade Agreement of 1965 and accession to the EU in 1973, there was little or no convergence on average EU living standards for almost three decades, as seen in Figure 8.2. Three factors are generally agreed to have held Ireland back over this period: (a) insider domination of the labor market, which yielded pay increases well ahead of productivity, alongside high unemployment and substantial emigration, (b) a failure to expand the educational attainment of the labor force in line with the rest of Western Europe, and (c) a procyclical fiscal expansion in the late 1970s which led to a debt and deficit crisis that took a further decade to bring under control.

150

Emergence of Ireland’s ICT Clusters 120 100 80 60 40 20

2002

1995

1988

1981

1974

1967

1960

0

Figure 8.2. Irish GNP per head as percentage of EU-15 average, 1960–2002 Notes: National income is measured at purchasing power standards rather than market exchange rates. Source: European Commission AMECO database for GDP per head at PPS, adjusted for the difference between GDP and GNP as given in the quarterly bulletins of the Central Bank of Ireland.

The Emergence of the Celtic Tiger A fortuitous combination of changes in policy and the external environment occurred in the late 1980s. The effects were dramatic, giving rise to the ‘Celtic Tiger’ label. Real national income per head rose from less than 65 percent of the EU average to achieve rough parity by the end of the 1990s. Unemployment tumbled from a high of 17 percent in 1987 to less than 4 percent in the early years of the new millennium, while numbers at work expanded by more than 50 percent. The beneficial shocks included a change in fiscal strategy in 1987 which created space for future tax reductions. These, in combination with the country’s newly developed, social partnership model’ of wage determination, bolstered cost competitiveness. The doubling of the EU Structural Funds in 1989 allowed a rapid resumption in the badly needed infrastructural projects which had been put on hold as part of the change in fiscal strategy. And, crucially for the purposes of the present chapter, the lead-up to the Single Market saw a huge increase in FDI flows both into and within Europe, of which Ireland captured a sharply increased share. We briefly discuss each of these factors, highlighting where appropriate the contribution of institutional learning.

151


Initial attempts to tackle the looming government debt crisis had relied on tax increases, with Ireland exhibiting the fastest growing tax-to-GNP ratio in the OECD. The tax burden raised wage demands however, exacerbating unemployment and raising social welfare spending. A combination of factors in 1986–7 paved the way for a new and ultimately successful stabilization attempt, which relied on cuts in government spending instead. Supportive developments included a currency devaluation, which improved cost competitiveness against the UK, and a liftoff in the world economy and especially the UK in 1987, which meant that the Irish expenditure cuts were (benignly) countercyclical. The year 1987 also saw the introduction of social partnership, whereby government, unions and employers came together to agree on a general path for wages over the following three-year period. Successive governments have used the process to purchase wage moderation via the promise of future tax reductions, with tax cuts accounting for about one-third of the rise in real take-home pay since the partnership process began. This winning combination of expenditure restraint underpinned by the promise of tax reductions had in fact secured the agreement of the major social groups—employers, unions and farmers—in advance of its implementation. That such a consensus emerged—through the forum of the National Economic and Social Council which brought these groups together—illustrates the depths of the crisis facing the country at that time, and serves as an example of the concept of ‘institutional learning’. The process of negotiating the partnership agreements, furthermore, has been argued to promote a shared understanding of how the economy functions and of the appropriate response to different economic shocks. In line with these views, Baccaro and Simoni (2002) find that social partnership changed the wage leadership process. Wage increases pre1987 had been driven by the rapid productivity growth of the foreignowned sector, while increases over the partnership period have been driven instead by the much slower productivity growth of the indigenous sector, leading to substantial reductions in overall unit costs. The timing of the expansion in EU Structural and Cohesion Funds from 1989 was also fortuitous. In addition to raising the level of FDI inflows that the economy’s infrastructure could handle, the aid would also have impacted on the type of FDI Ireland was able to attract, with the humanresource program of the Structural Funds contributing to the skill levels of the Irish workforce. The Structural Funds also contributed to institutional learning. As FitzGerald (1998) notes:

152

Emergence of Ireland’s ICT Clusters the need to satisfy the donor countries, through the EU Commission, that their money is well spent has resulted in the introduction of a set of evaluation procedures which has helped change the way the administration approaches public expenditure. In the past the only question, once money had been voted by parliament, was whether it had been spent in accordance with regulations. Now there is increasing interest in assessing how effective the expenditure has been.

The final beneficial shock to which the economy was subject was the development of the Single European Market. This led to a doubling (in real terms) in the amount of investment undertaken by US firms in the EU between the early and the late 1980s, and a quadrupling of Ireland’s share. In part this may have been due to the playing out of Marshallian agglomeration and bandwagon or demonstration effects in the newly benevolent environment. Perhaps even more important, however, was the liberalization of public procurement policies that the Single Market entailed. This prevented larger EU countries from using the threat of blacklisting publicly funded purchases of a firm’s products as a lever to influence their location decisions, a practice which had operated to Ireland’s considerable disadvantage, as argued by MacSharry and White (2000). Of course, the increasingly high-tech nature of the FDI attracted to Ireland over this period would not have occurred had the educational attainment levels of the workforce not been increasing rapidly. Ferreira and Vanhoudt (2002) conclude that increased educational throughput— especially given the vocational/technical slant of the skills provided at third level—and the sectoral (high-tech) composition of the increased FDI inflows were self-reinforcing factors that proved decisive for the boom. The number of jobs in foreign-owned industry in Ireland grew by 40 percent between 1987 and the year 2000, solidifying Ireland’s position as the most FDI-intensive EU economy. Before coming to discuss the specifics of the ICT sector in Ireland, we will briefly review the factors that make Ireland such an attractive location for FDI.

Determinants of Ireland’s Success in Attracting FDI One of the most important factors in accounting for Ireland’s success in attracting export-platform FDI is the continuing low rate of corporation tax.4 The standard corporation tax rate is currently 12.5 percent, which is less than half the EU 15 average. Desai, Foley, and Hines (2002) report a measure of the average effective corporation tax rates on US overseas investments for 1997, in which the Irish effective rate again comes in at less than half the EU average.

153


In terms of geographic and political factors, EU membership is clearly crucial. Proximity between FDI home and host locations remains a statistically significant determinant of FDI inflows, perhaps because of the impediments distance places on the speed and ease of communication. Thus, the UK and Ireland—as English-speaking environments with strong cultural connections with the USA—are particularly favored locations for US investors in Europe. Other aspects of the general business environment that are also likely to be of importance include the functioning of the labor market, the quality of public infrastructure and the efficiency of the public administration system. Surveys of MNC executives consistently rank the availability of appropriate skills as one of Ireland’s important advantages. Ireland has been successful in implementing a science-based education strategy which enhances its attractiveness to multinational corporations. Ireland has converged on the OECD average in terms of attainment of at least a university degree or equivalent, and has surpassed the OECD in terms of the proportion attaining third-level diplomas or their equivalent, while the extra Irish throughput in tertiary education is concentrated in electronics and the natural sciences. Another decisive factor in attracting FDI to Ireland has been the expertise of the country’s Industrial Development Agency. This will be discussed further below, in the specific context of the computer hardware and software sectors, to which we now turn.

The Computer Hardware Sector in Ireland Table 8.1 shows the shares of Ireland and various other countries in world exports of computers and peripherals and of electronic components. Within Europe the increase in Ireland’s share is seen to have come at the expense of the larger and traditionally more prosperous EU states such as France, Germany, and the UK. The declining shares of Japan and the USA suggest that similar processes are at work in other regions of the world economy also. Table 8.2 reports the share of computers and electronic components in manufacturing employment in several EU countries, relative to its overall importance in the EU15. Employment in both hardware segments is seen to be particularly important in two economies on the EU’s western periphery—Ireland and Scotland—which together are thought to have accounted for almost 70 percent of the personal computers sold in Europe in the late 1990s. In both countries, the sector is largely under

154

Emergence of Ireland’s ICT Clusters Table 8.1. Country shares in world computer hardware exports Shares of world exports SITC 752 Computers and peripherals

Europe France Germany Ireland Italy Netherlands UK Asia Japan Taiwan Hong Kong Korea Rep. China Singapore Thailand Malaysia Philippines Americas USA Canada Mexico Costa Rica

SITC 75997 Components

2000

1992

2000

1992

0.04 0.05 0.05 0.01 0.08 0.08

0.05 0.07 0.02 0.03 0.04 0.09

0.02 0.04 0.06 0.01 0.05 0.04

0.04 0.05 0.05 0.03 0.05 0.07

0.08 0.09 0.02 0.05 0.06 0.11 0.01 0.04 0.03

0.21 0.07 0.02 0.03 0.00 0.13 0.01 0.00 0.00

0.09 0.09 0.07 0.07 0.04 0.08 0.05 0.00 0.02

0.16 0.04 0.06 0.02 0.01 0.06 0.03 0.04 0.00

0.17 0.01 0.04 0.00

0.23 0.02 0.01 0.00

0.18 0.02 0.02 0.01

0.23 0.04 0.01 0.00

Source: UN trade statistics.

Table 8.2. The relative importance of computer sector employment in EU countries Computer equipment Nace 3002 Belgium Denmark Germany Spain France Ireland Italy Austria Portugal Finland Sweden UK Of which: . . . . . . . Scotland Netherlands

0.21 0.55 0.82 0.48 1.48 10.42 0.48 0.15 0.06 0.31 0.46 1.79 7.90 1.54

Electronic components Nace 321 0.79 0.65 0.90 0.44 1.80 3.77 0.69 1.75 0.71 1.07 0.79 1.10 3.05 0.54

Source: Eurostat New Cronos. Note: Data not available for Luxembourg and Greece.

155


foreign—and predominantly USA—ownership, and is relatively clustered: in Ireland—as we have seen—in the Greater Dublin region, and in Scotland in what has come to be known as Silicon Glen.

Computers and Peripherals Why were Ireland and Scotland so successful in the computer assembly segment? Dedrick and Kraemer (2002) argue that PC assembly firms have no need to cluster, and are indeed, in the USA, quite dispersed across the country. They argue that their concentration in countries like Ireland and Scotland has more to do with infrastructure and government incentives than Marshallian-type agglomerations. Both were relatively low labor cost locations within the EU15, with more abundant supplies of skilled labor than lower-cost economies such as Greece, Spain, and Portugal. Ireland furthermore was a low corporation tax environment while Scotland benefited from UK regional grants and other financial and fiscal incentives. Both countries were established locations for computer assembly even before the era of the personal computer.5 Ireland played host to Digital and a number of other minicomputer companies as well as a mainframe assembler in the 1970s, and a campaign began in the early 1980s to develop the country as a major European location for electronics and computer software. To the extent that the IDA accumulated valuable sector-specific knowledge through its interactions with early foreign entrants in the sector and that the employment of this knowledge facilitated the replacement of the earlier firms by PC assemblers over the course of the 1980s, an element of clustering—through the successful intervention of public-sector agencies—might be deemed to be involved. This theme will be developed further below. The Irish and Scottish computer assembly sectors appear to have reached their employment peaks around 1998, after which a shake out occurred. In Ireland, of the five microcomputer makers and one contract manufacturer located there in 1998, only Dell and Apple continued to assemble in 2002, and the latter’s assembly operations had been dramatically downsized. Computer assembly has been shifting to lower-cost locations within each of the triad markets over this period. In the North American case, the shift has been to Mexico, while in Asia, assembly was shifting from Singapore toward lower-cost locations such as Thailand, Malaysia, and China. Extrapolation of the global and European trends will see computer assembly for the EMEA market shifting further toward Central and Eastern

156


Europe. For Ireland, the major question is as to whether Dell will move. Fields (2004) argues that Dell has always shown a preference for locations near to its most critical material supplier, Intel, reflecting its preoccupation with avoiding supply disruptions. Within a few years of Dell choosing to locate in Ireland however, Intel shifted the activity of mounting its microchips into PCB boards to Asia, from where the boards are now distributed directly to computer manufacturers including Dell. Dell itself tends to create a cluster of sorts, since suppliers are required to maintain inventory, either through supply hubs or through production facilities, close to its plants to support the company’s build-to-order production method. These operational clusters are likely to be much less stable than technological ones (McKendrick 1998). Even if Dell and its cluster move from Ireland to Central and Eastern Europe, as Kraemer and Dedrick (2002) forecast, this will not necessarily mean that the Irish cluster is disentangling. The country appears to be moving up the value chain, into electronic components, while many of the hardware firms have shifted into export services activities in Ireland (Barry and Van Egeraat 2005).

Electronic Components Most of the innovation in computer hardware is now carried out by components’ suppliers, who tend to have quite high human capital requirements. Although there are many subcategories within the two segments, it would appear that a shift from NACE 30 to NACE 3210 generally represents a movement up the value chain.6 This movement up the value chain can be illustrated by the history of Intel’s operations in Ireland. When the company first came to the country in 1989, it assembled PCs and motherboards as well as producing microprocessor wafers. When it consolidated cartridge assembly in its Philippine and Puerto Rican plants, it refitted its Irish plant for much higher-level wafer production. The performance of the Irish plant has been impressive. As Durkan (1998) notes: The IFO plant in Ireland contributes 40 percent of the worldwide Virtual Factory White Papers in the .25 micron technology, and expects to reach the same level in the new .18 micron technology shortly. Furthermore IFO ranks in terms of IMEC paper submissions in the top 2 of Virtual Factory sites, contributing about 10 percent compared with an average of 2 to 3 percent. There have also been other

157

Emergence of Ireland’s ICT Clusters positive developments. The Irish plant developed a safety culture and system that has been adopted across the whole Intel operation worldwide and Ireland realized ISO9002 certification first. NSAI are the compliance auditors for Intel sites worldwide.

Intel recently opened a new IT innovation center in Ireland and resumed work on its e2.5 billion FAB 24 fabrication facility. In May 2004, it announced a further e1.6 billion investment to build a FAB 24.2 plant, which will reduce the size of microprocessors by a further 40 percent. Ireland had been in competition with the company’s Israeli and US plants for this investment. Highly skilled semiconductor design engineers will fill most of the new positions. Following construction in 2006, Intel will have invested e6 billion in Ireland since it first set up operations there in 1989.

Hardware R&D More than 80 percent of OECD R&D spending in the ODP sector takes place in the USA and Japan, which account for only a little over 60 percent of OECD employment in the sector. This shows, unsurprizingly, that R&D is far more concentrated geographically than are other measures of activity in the sector. Ireland, which accounts for almost 3 percent of OECD employment in ODP equipment, accounts for a mere 0.2 percent of R&D expenditures in the sector. Surveys of multinational executives indicate that the low corporation tax environment may play a role in the low R&D spend per employee in Ireland, since R&D counts as a cost within a manufacturing unit and R&D costs can be written off at higher tax rates in the MNC’s home base. The Irish government has recently taken a small step toward addressing this, by the introduction in 2004 of a 20 percent tax credit for incremental R&D.

The Computer Software Sector in Ireland The boundaries between hardware and software are blurred, as illustrated by IBM’s shift of focus in the 1990s from selling PCs to selling IT services and solutions. Foreign companies also, when they shift manufacturing processes out of Ireland, frequently replace them with customer support or software development centers. Table 8.3 reports the importance of computer software employment in EU countries, again measured relative to

158

Emergence of Ireland’s ICT Clusters Table 8.3. The relative importance of computer software employment in EU countries Belgium Denmark Germany Spain France Ireland Italy Netherlands Austria Portugal Finland Sweden UK

0.89 1.25 0.61 0.62 1.05 1.32 1.04 1.25 0.78 0.27 1.25 1.95 1.47

the EU15 average, in this case as a share of employment in manufacturing and market services. Software employment records its highest share of private-sector employment in Sweden, the UK and Ireland. Scotland, though a substantial player in hardware production, plays no such role in software. Instead, the computer services industry in the UK is concentrated in the Greater South East, the wealthiest UK region. Within software there is an important distinction between mass market packaged products, which tend to be produced by large MNCs, and other software activities—including custom and niche software and business solutions—in which domestic firms tend to dominate. The EU market is roughly evenly divided between the two sets of activities.

Mass Market Packaged Software Most of the localization of software for the broader EMEA market takes place in Ireland, which is said to account for around 50 percent of all massmarket packaged software sold in Europe (OECD 2002). Transfer pricing raises problems in evaluating output levels in this sector. Even in employment terms, however, the packaged software sector is more important in Ireland than in other EU economies.7 Eurostat data register employment in this sector in only eight EU countries, with employment numbers as shown in Table 8.4. The mass-market packaged software sector in Ireland is engaged in the manufacturing, localization, and distribution (MLD) of software packages. This is not a particularly high-skill segment of the sector. Around 50

159

Emergence of Ireland’s ICT Clusters Table 8.4. EU employment in mass market packaged software Nace 2233 Reproduction of computer media

Spain France Ireland Italy Netherlands Finland Sweden UK

2000 663 875 5,591 342 168 16 194 3,576

Source: Eurostat New Cronos.

percent of employees in these operations are typically engaged in the relatively low-skill manufacturing stage, while around 30 percent are involved in localization. In the case of Microsoft’s Irish operations, some 90 percent of staff involved in localization had third-level qualifications in information technology or linguistics, while 35 percent were nationals of mainland European countries. Even though MLD is not very high-tech in nature, the sector has nevertheless moved up the value chain over time. The key players in the MLD sector (including Microsoft, Lotus, Oracle, Symantec, Informix, and Corel) first established software manufacturing facilities in Ireland around the mid-1980s, duplicating and shrink-wrapping disk copies of software programs developed by the parent company and arranging for the printing and assembly of manuals. The second phase, again beginning with Lotus and Microsoft, saw these companies adding localization— involving translation into other languages and cultural and technical formats appropriate to the destination markets—to the process. The third phase of the sector’s development saw the transfer of the responsibility for distribution, which had previously been handled by local distributors, to the Irish operations themselves, thus making Ireland an operations hub. MLD activities account for about half the jobs in the foreign-owned software sector in Ireland. The other half are accounted for by the software development sector. This is substantially more highly skilled, and will be discussed further below. What are the factors that are likely to have drawn the mass-market segment to Ireland? First, the need for localization and the requirement

160


for a supply of workers with the requisite linguistic knowledge mean that production has tended to respect the triad boundaries between the Americas, EMEA, and Asia. Ireland has remained a favored location in Europe because of its tax advantages and because, as an English-language environment, it has been able to attract sufficient numbers of the young Continental Europeans upon whom the localization segment relies.

Niche Software and Computer Services The remainder of the software sector comprises NACE 72 (computer services and related activities). This segment includes custom software (which is provided for individual companies), niche software (written for specific business sectors) and other software services, which are provided both for organizations and for consumers. In this segment, countries with high computer penetration rates such as Sweden, Denmark, Finland, the Netherlands, and the UK have higher weights than the rest of Europe. Their relatively strong showing in this segment reflects the fact that many computer services are essentially nontradeable. Though Ireland achieved phenomenal growth in the software sector throughout the 1990s, this was just sufficient to allow the country attain a middle ranking relative to other EU countries and to the regions of the UK economy in terms of NACE 72. While UK software and computer services companies, however, are found to obtain only around one-third of their revenues from exports, and French and German companies from 25 to 30 percent, exports accounted for 85 percent of the revenues of Irish indigenous firms by 2002. This strong export orientation is explained by the fact that about half of Irish indigenous software firms are engaged in the development and sale of niche products in sectors such as Banking and Finance, Telecommunications, and computer/Internet-based training. The emergence of this product orientation on the part of Irish firms is in part ascribable to the substantial presence of MNCs across all manufacturing and services sectors in Ireland. O’Gorman, O’Malley, and Mooney (1997) report that, for a large minority of indigenous software firms, sales to Irish-based MNCs, particularly in pharmaceuticals but also in financial services, telecoms, and food and drink, are quite important, while O’Riain (1997) describes how some indigenous firms, which began by providing custom services, expanded over time into producing consultancy kits which eventually became exportable products.

161


Software R&D Software R&D expenditures are less concentrated geographically than is the case for ODP, and Ireland captures more than twice as high a share of OECD software R&D than it does of hardware R&D. The R&D spend within the indigenous software sector is also substantial, with Forfa´s (2003) reporting that levels per head are about the same as in the foreignowned segment. Within the foreign sector, software development in Ireland accounts for somewhat less than half of total employment but is more high-skill than the rest. One part consists of branches of major computing services or IT consulting companies (including EDS, IBM, ICL, and Accenture). The other is an adjunct to nonsoftware electronics corporations such as Motorola and Ericsson, with operations focused on the production of embedded software and applications for products such as mobile phones. Hochtberger, White, and Grimes (2004) present case studies of three foreign firms engaged in software development in Ireland—HP, Electronic Data Systems (EDS), and IBM. What follows is a brief synopsis of these case studies, where we focus on material relevant to the issue of clustering, to which we turn our attention in the next section. Hewlett-Packard employs around 4,000 people across three locations in Ireland, with its European Software Centre located in the west coast city of Galway. The Galway affiliate’s MD told the authors that ‘the corporate relationships that most influence the center’s development are not with the other Irish HP divisions but with other HP affiliates in the USA that are within the same line of business.’ Furthermore, the various Irish divisions ‘do not report into each other. They do not really have anything to do with each other, although . . . we coordinate our efforts from the public relations point of view’. Linkages are strong however with local third-level educational establishments, particularly in terms of research carried out at the Digital Enterprise Research Unit of the National University of Ireland, Galway, and through a graduate recruitment program and involvement in curriculum development at local Institutes of Technology. US computer-services firm EDS first established an Irish presence in 1990. Its Dublin affiliate was the first designated ‘solution center’ designed to export services to international clients from a location outside the USA. Among the reasons why Ireland was chosen was its track record as a host location for foreign MNCs. As the Irish affiliate performed well in its dealings with a number of EDS’s most significant clients, the company came to appreciate more and more the skills that the Irish workforce

162


offered and the Irish affiliate was allowed to extend its scope toward greater process development. Interestingly, while EDS will often work together with competitors on particular projects, these relationships emerge at the global corporate level rather than arising from the clustering of IT MNCs in Ireland. IBM had been manufacturing in Ireland since 1960 but most of its Irish workforce over the course of the 1990s came to be employed in services. Around one-third of their staff works in its sales and support center for the EMEA region, while most of the remainder is employed on its technology campus outside Dublin. With the boom of the 1990s, the Irish subsidiary has become more involved in services’ provision in Ireland. The Irish IT agglomeration is found to benefit IBM in an unusual way. According to the research interview of Hochtberger, White, and Grimes (2004) the agglomeration means that Ireland is in some sense a microcosm of the global arena. The interviewee suggested that competing against other global firms in the Irish marketplace provides IBM with a close-up of other firms’ global strategies, from which the entire corporation can learn.

Spatial CoLocation, Clustering, and the Organic Development of ICT in Ireland Thus far we have identified a high degree of spatial colocation among firms in the hardware and software sectors. What accounts for Ireland’s strong showing in these sectors? A number of elements—including the corporation-tax regime, the skills and experience of the IDA and the orientation of education policy toward meeting the needs of MNCs— have already been alluded to as important in Ireland’s overall success in attracting FDI, while the Single Market and the restoration of macroeconomic stability in the late 1980s were clearly important in facilitating increased FDI inflows across all sectors. As will be argued below, the corporation tax regime in combination with the improving skill levels of the labor force would have been particularly conducive to high-tech sectors, while agglomeration and demonstration effects and the impact that the large vibrant MNC sector has have on the sectoral structure of indigenous industry would also have contributed to clustering. Elements of cumulative causation would then set in. The vibrant indigenous software sector contributed to the birth of a VC industry, increasing the locational advantages that Ireland has to offer. This was aided by the institutional learning of the public service, which was able to affect the

163


types of infrastructural development that further growth in these sectors required.

The Role of the Corporation-Tax Regime The importance of Ireland’s low corporation tax regime in stimulating FDI inflows has already been alluded to. Given other requirements such as adequate physical infrastructure, a plentiful supply of appropriately skilled labor and stable government, a low-corporation-tax regime may be particularly attractive to sectors with high R&D- and advertising expenditures, as it can be difficult for international tax authorities to locate the exact source of value added in such sectors. This makes it easier to manipulate transfer prices in order to shift profits to low-tax locations such as Ireland. According to Davies and Lyons’ categorization (1996), such advertising and R&D-intensive sectors accounted for over 65 percent of foreign employment in Irish manufacturing in the year 2000.

Agglomeration and Demonstration Effects and Impacts on Indigenous Industry The importance of Marshallian agglomerations and demonstration effects in influencing firms’ decisions to locate in Ireland is studied empirically by ¨ rg, and Strobl (2003).8 In a regression on firm entry in sector j at Barry, Go time t they find all three agglomeration effects, as well as demonstration effects, to be of the right sign; they confirm that flagship projects are of special importance and that new firms are influenced most strongly by the location decisions of incumbents of the same nationality. With respect to the impact that foreign-firm presence might have on indigenous firms, in principle this can on balance be positive or negative. The impact on indigenous firms can be negative, for example, when the latter are crowded out of either product or factor markets. There is little product-market competition between indigenous and foreign firms in the Irish case, however. Not only are foreign firms highly export-oriented, they export to different markets than indigenous firms, and they are in any case in different product categories. Labor-market crowding out might also be deemed to be minimal because of the extent of integration between the Irish and UK labor markets, which reduces the impact of local developments on Irish wage levels. Positive interactions can arise when indigenous firms act as subsuppliers to foreign-owned firms, or when productivity spillovers occur. There is

164


econometric evidence for Irish manufacturing that indigenous-firm entry is positively affected by foreign-firm presence in the same sector and in industries downstream of that sector and foreign presence has a lifeenhancing effect on domestic firms in high-tech sectors, which is suggestive of technological spillovers. While indigenous-firm presence in the hardware sector remains very small however, this is not true of software. Based on survey evidence, O’Gorman, O’Malley, and Mooney (1997) find that ‘demand from overseas MNCs in Ireland as customers did have a significant influence in strengthening the capabilities of many indigenous software firms, probably a large minority of them’. Also, ‘for a minority of about one-quarter of companies, selling to overseas MNCs in Ireland has been of significant direct assistance in gaining access to export customers, for example, through referrals by MNC customers to foreign affiliate companies or to other customers abroad’. Another group of software firms, accounting for perhaps one-tenth of indigenous software employment, arose as firms in other industries, such as telecommunications or computer hardware, spun off their software divisions (O’Riain 1997). The impact of foreign presence on the entry rate of Irish indigenous manufacturing firms may also be related to their role as ‘incubators’ for new entrepreneurs. A study of the electronics industry in the early 1980s suggested that this was the case. With respect to software, O’Gorman, O’Malley, and Mooney (1997) find that one-third of Irish indigenous entrepreneurs in the sector had worked in foreign firms immediately before the start-up of the new firm, while two-thirds had worked in foreign firms at some stage in their careers.

Endogenous Responses to the Emergence of Clusters These findings—both econometric and from survey evidence—suggest that there is some element of conventional clustering in the Irish ICT data. The presence of such clusters is likely in turn to impact on the economic or infrastructural environment in ways that will make it more conducive to the development of these particular sectors. Industry associations, for example will typically arise to lobby government to effect such changes. The 2003 document—Creating a World Class Environment for ICT Entrepreneurs—produced by a subgroup of the Irish Business and Employers Confederation, represents an example of such a lobby group in action.

165


Other environmental changes beneficial to the industry will also take place, perhaps with state prodding, as in the case of the development of a VC industry in Ireland. The Business Expansion Scheme, which offered tax incentives for long-term equity-capital investments in certain sectors of the economy, was introduced in 1984 and the Irish Venture Capital Association was established as the industry’s representative body in 1985. Members have invested close to 1 billion euro over the last decade, with the vast bulk invested in Ireland and a high proportion going to technology companies.

Targeting FDI: Role of Policy in the Emergence of Ireland’s Specialization in ICT Ireland was one of the first countries to adopt an FDI-based development strategy, and the IDA is widely recognized as one of the most effective investment promotion agencies in the world.9 Its modus operandi is described by former Irish Finance Minister and EU Commissioner Ray MacSharry and long-serving managing director of the IDA Padraic White, in MacSharry and White (2000). The agency first identifies, partly interactively, the high-growth sectors and subsectors that are thought to provide a good fit for Ireland’s resources and development aims. Having attracted several computer and components firms in the 1970s, for example and being favorably impressed by their performance in situ, the IDA launched a campaign in the early 1980s to develop Ireland as a major European location for electronics and computer software. The agency’s next step involves approaching the strongest companies in these niche areas with a view to persuading them to locate in Ireland. Intel, for example was pursued by the IDA for over a decade before deciding in 1989 to open a European plant, with Ireland ultimately emerging as the favored location.10 After maintaining contacts for more than two decades with IBM—a company which had traditionally shied away from export-platform activity—the IDA, partly on the basis of the success of the Software Development Centre that the company had set up in Ireland to meet its in-house development needs, eventually persuaded them that such a move could be beneficial, leading to the opening of an export plant in Ireland. The sectors successfully targeted by the IDA all had relatively high-skill intensities, medium as opposed to high plant-level economies of scale and relatively low transport costs, making them suitable for relocation to

166


high-skill peripheral regions (Midelfart et al. 2000). Targeting by the IDA helped capture these sectors for Ireland rather than having them go elsewhere, and the agency played a crucial role in advertising Ireland’s advantages, in convincing potential investors that apparent difficulties could be overcome, and in capturing the important ‘flagship projects’ that appear to have been of importance in cluster development. Crucially, however, experience and track record have given the IDA a degree of bureaucratic clout unusual for an investment promotion agency, allowing it to extend its influence into areas not traditionally recognized as lying within the industrial policy remit. By bringing the concerns of industrialists forcefully to government, for example, it played a major role in forcing through the modernization of the country’s telecommunications infrastructure in the late 1970s and early 1980s, which allowed Ireland develop a head start in attracting the then newly off-shoring ITenabled services sectors. When it noticed in the late 1970s a looming disparity between electronics graduate outflows and its own demand projections, it was able to secure rapid government action to institute one-year conversion courses to furnish science graduates with electronics qualifications. A huge expansion in the capacity of electrical engineering courses in the state followed, positioning the country well to profit from the subsequent explosion in the global software sector. Enterprise Ireland, a sister agency of the IDA tasked with the development of indigenous industry, also provides an example of international best practice in terms of the national linkage programs it established to further integrate foreign enterprises into the Irish economy (Battat, Frank, and Shen 1996; UNCTAD 2001).11 Enterprise Ireland also has a relatively strong involvement in venture capital. It is thought to account for 11 percent of the funds under management in Ireland compared to an average public-sector involvement of 7 percent across the rest of Europe. And, more recently, the development agencies have been to the fore in pushing for and overseeing the implementation of a new public emphasis on science, technology and innovation, once convergence on average Western European living standards had been achieved and the threat of increased corporation-tax competition from Central and Eastern Europe emerged.12 Recognition of the importance of these issues was heralded by the release in 1996 of the first-ever Irish Government White Paper on Science, Technology and Innovation. It is underlined by the fivefold increase in investment in these areas under the current National Development Plan, by the funding

167


by Science Foundation Ireland (SFI) of five joint partnerships between thirdlevel research institutions and industry, and by the introduction of a 20 percent tax credit for incremental R&D in the Finance Act of 2004. Within ICT alone, the last two years have registered a number of significant developments under this new strategy. Bell Labs has announced its intention to set up a major R&D center at Lucent Technologies’ Dublin facility, linked with the establishment of a collaborative academic center at one of the city’s universities. Similarly, HP announced the establishment of a world-class Technology Development Centre at its manufacturing facility outside Dublin, while its European Software Centre entered into collaboration with the National University of Ireland-Galway in establishing the Digital Enterprise Research Institute. Intel has established an innovation center at its main site outside Dublin while increasing its investment in its research center near Limerick. It has also partnered three Irish universities in an academic Centre for Research on Adaptive Nanostructures and Nanodevices. IBM, over this same period, announced further significant investments in its Irish R&D software facility in Dublin. The decision was influenced, according to one of the directors of the company, by the availability of the necessary skills, the strong support of the IDA and the growing emphasis on scientific research by SFI.

Conclusions Ireland displays a strong employment and export specialization in both computer hardware and software, with both sectors concentrated in the Greater Dublin region and in a small number of other urban centers. Thus the preconditions for cluster development are established. The reasons for the high degree of spatial colocation of both sectors in Ireland, and indeed for Ireland’s overall success in attracting FDI, include the low corporation-tax regime, adequate supplies of skilled labor, a conducive macroeconomic and business environment and the skills and experience of the country’s industrial development agencies. The first two of these at least are likely to have been of particular importance to the high-tech sectors that the agencies have targeted. We have cited both survey and econometric evidence, furthermore, of the importance of previous firm entry—particularly of ‘flagship projects’— in drawing other firms in the same and related sectors to locate in Ireland. At least one industry analyst has suggested that Dell’s decision to locate in Ireland was influenced by Intel’s prior location there. While it was pointed

168


out above that the Dell and Intel plants in Ireland no longer interact directly with each other, path dependence may already have set in, so that this need not destabilize the cluster. Dell, in turn, through its requirement that suppliers maintain either factories or supply hubs within a short distance of its assembly sites, tends to generate its own cluster of surrounding firms. These kinds of developments, in combination with the state-sponsored National Linkage Program, have resulted in a substantial increase in electronics-sector linkages. These input–output links are one of the three Marshallian factors generating agglomerations. The others are technology spillovers and ¨ rg and Strobl thick factor markets, especially for specialized skills. Go (2003) provide empirical evidence for technology spillovers from foreign to indigenous high-tech manufacturing firms in Ireland. Given the very small size of the indigenous hardware sector, however, it is difficult to imagine that these spillovers could have been particularly dramatic, a ¨ rg (2002) in their study of the conclusion reached also by Kearns and Go electronics sector.13 There can be little doubt on the other hand that the science-based thirdlevel educational system in Ireland has contributed enormously to the development of both ICT sectors. Ireland produces more science and engineering graduates per thousand of the population cohort aged 20–34 than the 11 other EU countries included in the European Commission (2003) data, and surveys of multinational firms invariably rank this as important in their choice of location. There is little evidence, however, of the emergence of ‘deep clusters’, where a region contains complete or nearly complete supply chains. This may not be particularly surprising, though, given that the present era is characterized by a very high degree of global production sharing, particularly in trade category SITC 7, which includes electronics (Yeats 2001). We have also found little evidence in Ireland of clusters that exhibit ‘dense patterns of interactions among local firms that differ quantitatively and qualitatively from the interactions that the firms have with those not located in the cluster’. Drawing on Kolko (2001), Quah (2001) suggests that ICT need not generally involve these kinds of complex interrelationships; geographic clustering in this sector, it is argued, arises primarily because of the high-skill requirements of the industry. In the most general sense, clusters are based not just on critical mass but on some particular source of advantage than promotes organic development. Amsden and Chu (2003) argue that, in the case of Taiwan and other successful late industrializers, this source of advantage was embodied in

169


the insights and actions of the state’s development agencies. A strong argument can be made that the same is true for Ireland.14 The development of both the hardware and software sectors in Ireland, as we have seen, owes much to the expertise and policy interventions of the IDA. The organization was instrumental in effecting dramatic improvements in the country’s telecommunications infrastructure, which allowed a whole new range of IT-enabled services industries to come on stream. It was instrumental in ensuring an adequate supply of electronics engineers, which not only improved the country’s attractiveness to foreign electronics firms but also proved highly beneficial when software later emerged as the dynamo of ICT.15 More recently, Ireland’s development agencies have been to the fore in pushing for and overseeing the implementation of a new state emphasis on science, technology, and innovation—efforts which appear already to be bearing fruit. How, though, can the IDA be thought of as a localized source of advantage? Can its actions not easily be copied by its competitors? There are two reasons why this may not be the case. The first is that its internal structures appear to make it amenable to institutional learning. The second is that its powerful position within the state bureaucracy may not be easily replicable elsewhere.16

Notes 1. The numbers for hardware come from the Forfas employment survey, which distinguishes between employment in foreign and domestic firms going back to 1973. About one-third of recorded employment in hardware is in electronic components, with the remainder in office and data processing equipment. 2. For an analysis of the emergence of an ICT cluster in Galway, see Green et al. (2001). 3. In the managerial literature, ‘organizational learning’ refers to learning within a given organization while ‘institutional learning’ refers to improvements in the quality of interactions between organizations that relate to each other in a given context. Here the latter term is assumed to embrace the former. 4. Empirical evidence on the importance of corporation taxes in determining FDI flows is presented by Gropp and Kostial (2000) and Altshuler, Grubert, and Newlon (2001). 5. See Van Egeraat and Jacobson (2004) for a detailed history of the Irish and Scottish computer hardware industries. 6. Within the pre-enlargement EU there was a net gain of 100,000 jobs in NACE 3210 and a loss of 34,000 jobs in NACE 30 between 1993 and 2000, with 10 of

170


7.

8.

9. 10.

11.

12.

13. 14.

15.

16.

the 13 EU countries for which data are available recording an increase in employment in the former relative to the latter. These companies also pay very substantial taxes to the Irish exchequer. In 2001, for example, Microsoft, though employing only around 2,000 people, paid almost 5 percent of that year’s total Irish corporation tax take. The importance of demonstration effects is further evidenced by Barry and Bradley (1997) who write that ‘surveys of executives of newly arriving companies in the computer, instrument engineering, pharmaceutical and chemical sectors indicate that their location decision is now strongly influenced by the fact that other key market players are already located in Ireland.’ See e.g. Loewendahl (2001). The story is told of how, at a late stage, the company became paralyzed by fears that engineers with the requisite experience could not be found in Ireland. The IDA commissioned interviews with over 300 Irish engineers, working mainly in the USA, who had the appropriate experience, and was able to report to Intel that over 80 percent of them expressed a willingness to return to Ireland if offered a good career opportunity with a quality company. In electronics, for example, the percentage of raw materials and components procured locally by foreign firms rose from 8 percent in 1984 to 24 percent in ¨ rg and Ruane 1998). 1994, and, for domestic firms, from 19 to 32 percent (Go The development agencies comprise the IDA, Enterprise Ireland (the support body for indigenous industry) and Forfa´s, the national policy and advisory board for enterprise, trade, science, technology, and innovation. The other high-tech sectors in Ireland have a higher ratio of indigenous to foreign employment than computer hardware does. Some might argue that a process that relies on the actions of a state agency cannot be regarded as organic. The presence and modus operandi of the IDA might be regarded alternatively, however, as a unique feature of Ireland’s infrastructure. This complies with Porter’s key thesis that competitive advantage is created and sustained through a highly localized process. And, as an example of the type of externalities associated with Jane Jacobs, the software demands of many of the country’s manufacturing and financialsector foreign multinationals contributed to the emergence of an exportoriented indigenous software sector. MacSharry and White (2000) offer several reasons for why this might be the case: (a) institutional resistance on the part of Foreign Ministries to allowing other agencies establish such a strong foreign presence, (b) difficulties in securing the right caliber of proactive people to run such agencies, (c) the fact that governments rarely provide investment agencies with a clear development mandate, or the funds to carry out this mandate. ‘Very few countries’, they conclude, ‘have been able to create the combination of circumstances and people to forge an effective national investment promotion agency’.

171

9 The Emergence of Israel’s Venture Capital Industry: How Policy Can Influence High-Tech Cluster Dynamics Gil Avnimelech and Morris Teubal

While in principle any company in any country could issue securities on the US NASDAQ, building a new high-tech cluster may well depend on the emergence of an indigenous VC industry. Israel is perhaps the most prominent example of a policy-induced VC industry that coevolved with the transformation of the old military-dominated electronics industry into a Silicon Valley-type technology intensive cluster. That makes the Israeli experience unique, being one of the few countries that have managed to successfully emulate the Silicon Valley model of VC. Still, the emergence of a domestic VC industry has been preceded by more than thirty years of development of favorable background conditions and preemergence events. This chapter documents the Israeli experience and identifies the cornerstones of the policies that help to transform Israel from an agrarian economy into one of the leading global ICT economies. In doing so, we consider VC as an industry in its own right. It is important to recognize that supply of VC does not suffice: attention must also be paid to the creation of demand for VC services as well as taking the context and timing of policy seriously into account.1 Furthermore, VC emergence in Israel was a policy-led process which thus points to necessity of political capabilities. Central to the Israeli development was the onset of a catalytic, cumulative process with positive feedback and continuous learning that involved the entire high-tech cluster. Thus, Israel provides an example of the possibility of a rapid latching into the global ICT revolution, suggesting that there are important policy lessons from which not only other

172

The Emergence of Israel’s Venture Capital Industry

advanced industrialized economies in Europe and Asia (e.g. Taiwan and Singapore) can benefit from, but also top-tier developing economies like India and China. The following section lays out the chronology of the development of the Israeli VC industry, while the next section show how the VC industry was a component in a broader strategy aimed at enhancing Israel’s innovative and technological capabilities. Thereafter the dynamics and the transmission between stages in the process are described. We conclude by considering the reasons why government policies could fail with possible lessons for other countries.

The Israeli VC Chronology While the USA was the innovator and also the global leader in the VC industry, Israel constitutes a profile of a successful follower. The VC industry took about half the time to emerge in Israel as compared to the USA. A striking difference between the USA and Israel concerns the role of policy in VC emergence although in both countries policy played important and often indirect roles. VC emergence in Israel was a policy-led process since it was the result of a deliberate, targeted policy, which, by virtue of its scope, was clearly the dominant factor in the creation of a cumulative process of growth. In contrast, the USA’s Small Business Investment Companies (SBIC) program was not directed to VC per se, but was rather the Federal Government priority to support small- and medium-sized firms. The direct effect on the emergence of the US VC industry was presumably not the dominant one.2 From an evolutionary perspective, identifying the beginning of the VC industry, and whenever relevant, the market for VC services, is obviously a major aspect of the analysis. The pre-emergent phase appeared in 1970– 84, where both the technological infrastructure and the financial infrastructure for the subsequent emergence of a VC industry was established through a number of critical events that were however not directly related to VC.3 Beyond indigenous R&D capabilities they include the beginning of global product and capital market links and creation of a favorable environment for foreign investment. In addition, financial institutions increased their investment in high-tech industry, and technological entrepreneurship began to appear. From 1985 to 1992 a VC industry with a clear identity did not yet exist although some—mainly informal—VC activity and experimentation

173


occurred. Start-up activity increased, largely in response to layoffs from the defense-related industries coupled with enhanced commercial opportunities in the software and communications areas. The presumption is that the VC industry took off sometime during the 1993–2000 period. During this period there was an acceleration of growth of VC activity as witnessed by entry of large numbers of players both on the supply side, in terms of numbers of VC firms and the number of VC funds raised, and on the demand side by start-up companies (see Avnimelech and Teubal 2004a, 2004b; Avnimelech, Kenney, and Teubal 2005). After the initial phase in 1993–5 there was an accelerated growth phase in 1996–8, characterized by experimentation and collective learning, both with respect to VC strategies and their organization. Many strategies, routines, and organizational forms did not survive; some did and were increasingly adopted by a varying numbers of VCs. Fierce competition was paired with cooperation and a selection process through the market.4 The VC industry also began experimenting with institutions and with collective organizations (Israel Venture Association). Table 9.1 summarizes the main characteristics of Israel’s Silicon Valley model of high-tech cluster which developed during the 1990s, compared with the situation prevailing at the end of the 1980s and 1970s. Critical elements were the use of limited partnerships organizational form, and a clear focus on early phase investment strategy by VC firms at that time. Limited partnerships reduced potential downside risk while early stage investing had a high potential payoff. An environment was also established in which firms strived toward a born global profile whose objective was to exit through global capital markets. During the subsequent rapid growth phase (1995–8) we observe an accelerated entry of new VC companies and of VC activity fed by a cumulative process with positive feedback effects.5 It is then that the industry attained a size which enabled it to sustain a large number of supporting services. Now the sector converged to a relatively stable distribution of strategies in terms of investment stages, routines and of organization forms (limited partnerships). It is paralleled by the creation and growth of large numbers of new start-up firms. After the global ICT meltdown in 2000, Israel VC experienced a crisis and restructuring. In many ways, the ability to restructure suggests the successful establishment of the model. Since start-up firms’ output represented an important share of aggregate economic growth toward the end of the decade, a long and painful process of restructuring and adaptation

174

The Emergence of Israel’s Venture Capital Industry Table 9.1. Israel’s high-tech cluster: selected structural elements, 1970s–1990s Accumulated during the decade Number of start-up firms creation Funds raised by VCs, million US $ (M$) Capital invested in Israeli start-up firms by VCs (including foreign), M$ Accumulated no. of IPOs (high tech) Accumulated VC-backed IPOs: Accumulated significant M&As by MNE: Capital raised in NASDAQ in the decade, M$ Mergers and acquisitions (M&A), B$ A. Figure for the end of the decade Number of international investment banks in Israel Number of VC companies Share of ICT exports in manufacturing exports ICT manufacturing exports M$ Software exports MS Civilian R&D as percentage of GDP ICT employees (thousands) ICT-skilled employees (thousands) Patents issued

1990s 2,500 8,500 6,650

1980s 300 50 50

1970s 150 0 0

126 72 75 10,750 18,200 1990s — 100 54% 12,950 2,600 4.8% 152 57 969

9 3 0 50 0 1980s 1 2 28% 2,450 75 2.8% 80 37 325

1 0 0 10 0 1970s 0 0 14% 350 0 1.8% 60 26 140

Source: CBS, OCS, IVA, IAEI, and USPTO. Note: Frequently the figures in the box are approximations due to gaps in the availability of data, the existence of various sources of information, including fragmentary information from nonofficial sources.

followed. New patterns of interaction and links between the high-tech sector and the rest of the economy came to existence. A difference with regard to the USA was that while the VC industry in the USA had already been consolidated and achieved sustainability, Israel’s was still in the process of building up and consolidating its VC industry. In the postcrisis restructuring of the USA VC industry in the 1970s, the new, flexible regulation of pension funds played a critical role. Subsequently, the pension funds became the main source of capital of the US VC industry. In addition, subsidies were granted through the USA’s Small Business Innovation Research (SBIR) program which supported early phase R&D in smalland medium-sized firms and start-up, which supposedly was important for US VC industry consolidation.6 As in the US case, policy likely played an important role in the successful restructuring of VC industries and in the subsequent consolidation. By then, the core of the industry consisted of those VCs that survived the crisis. This last phase was also characterized by a relatively stable set of VC strategic groups (defined by capabilities, strategy, and performance) and

175


by VC industry sustainability (Avnimelech 2004). The information at our disposal strongly suggests that Israel’s VC industry began its consolidation process in 2004 (Avnimelech and Teubal 2005a).

Israel’s Innovation and Technology Policy Israel’s policy toward innovation and technology can be traced to overlap with the chronological description of the genesis of the domestic VC industry. Three phases can be distinguished that follow the model described in Chapter 1. In the pre-emergent first stage from 1969 to 1984, building capabilities to diffuse R&D and innovate was at focus. This phase was characterized by horizontal grants to business sector R&D aimed at the increasing private sector research activities and the creation of high-tech companies and the first start-up firms. The following phase 2 (1985–1992) was directed toward strengthening of R&D in the business sector, increasing the number of start-ups, and VC experiments. From this highly experimental period a new model of start-up firms with links to global capital/product markets was extracted. There was also a sharp increase in business sector R&D grants and the initiation of the incubator and ‘magnet program’ (described in detail below), which supported cooperative and generic R&D. During this phase the first VC support program, INBAL, was introduced. A general expansion in private sector R&D could be seen, as well as an increased rate of start-up formation and an increasing demand for VC services. Most important was however the learning from failures—INBAL’s failure and from business experiments—that led to the identification of the need for new government policies and the selection of limited partnership form of VC organization. These experiences were used in the third phase (1993– 2000) which targeted VC and also aimed at accelerating the growth of the R&D-intensive high-technology sector. It is now that the VC program named Yozma Program is initiated, while the other parts of the innovation and technology policy continued. It is also now that the VC industry achieves prominence and an accelerated growth of start-ups can be witnessed, spurred by an increase in profitable initial public offerings, and a restructuring driven by mergers and acquisitions. We will now somewhat in detail describe the most critical components of Israel’s innovation and technology policy.

176


Creation of the Office of the Chief Scientist (OCS): Grants to Private Sector R&D The Horizontal Grants to Business Sector R&D program began in 1969 with the creation at the Ministry of Industry and Trade of a specialized agency, the OCS. This program was and continues to be the backbone of the country’s R&D and Innovation strategy. Until the early 1990s, more than 90 percent of OCS disbursements to Civilian R&D came from this program, which supports the R&D activity of individual companies oriented to new/improved products and processes directed to the export market.7 In contrast to a targeted program which is applicable to a specific industry or technology, a horizontal program is open in principle to all firms. In Israel it extended a 50 percent subsidy to every R&D project accepted by the OCS, regardless of the firms’ industry, product class and technology (Teubal 1993). The major objectives of the horizontal R&D grants program during early implementation were to promote learning about R&D/Innovation8 in general and to generate knowledge about potential areas where the country concerned might have or could develop a sustainable competitive advantage. Much of the former is collective learning; R&D performing firms mutually learn from each other; and a lot of this learning relates not directly to technology or R&D proper but to organizational and managerial factors.

Strengthening of Business Sector The 1984 R&D law further consolidated Israel’s support of business sector R&D. The objective was to support knowledge-intensive industries, through expansion of the science and technology infrastructure and exploitation of existing human resources; creation of employment including absorption of immigrant scientists and engineers. The outcome was a significant increase in R&D awards to industry; and recognition of software as an industry—a very significant event indeed. Table 9.2 demonstrates OCD R&D support from 1985 to 2003. From 1985 to 1992 several notable new program innovations were initiated. These include Inbal created in 1991, a government-owned insurance company, which gave partial (70 percent) guarantees to traded VC funds. Four VC companies were established under Inbal regulations. This early VC support program failed to create a VC industry but was a valuable experiment. The Magnet Program (1992)—a US$ 60 million a year Horizontal

177

The Emergence of Israel’s Venture Capital Industry Table 9.2. R&D support from OSC, million US$ Year

Total grants (and annual change)

Business sector R&D grants

MAGNET budget

Technology incubators

1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

106 (2.5%) 110 (2.8%) 113 (2.7%) 120 (6.2%) 125 (4.2%) 136 (8.8%) 179 (31.6%) 199 (11.2%) 231 (16.1%) 317 (32.2%) 346 (9.1%) 351 (1.4%) 397 (13.1%) 400 (0.8%) 428 (7.0%) 440 (2.8%) 431 (2.0%) 383 (11%) 369 (3.4%)

106 109 112 118 122 133 171 177 199 172 294 279 309 305 331 337 328 291 283

0 0 0 0 0 0 0 1 40 10 16 36 53 61 59 67 64 58 53

0 0 0 0 0 0 4 16 24 27 31 30 30 30 30 32 32 27 26

Royalties (growth)

BIRD-F awards

6 (33.3%) 7 (16.7%) 8 (14.3%) 9 (12.5%) 10 (11.1%) 14 (40.0%) 20 (42.9%) 25 (25.0%) 33 (32.0%) 42 (27.3%) 56 (33.3%) 79 (41.1%) 103 (30.4%) 117 (13.6%) 139 (18.8%) 135 (10.8%) 145 (5.2%) 153 (1.4%) 133 (5.4%)

NA NA NA NA NA NA 12 10 12 10 12 13 12 14 9 8 11 10 11

Source : Avnimelech (2004), the OCS, and BIRD-F.

Program supporting cooperative, generic R&D, involving two or more firms and at least one university created joint projects. The Technological Incubators program, initiated in 1992, was a program supporting entrepreneurs during the Seed Phase, for a period of three years. The incubators are privately owned and managed and both the incubator and the projects are supported financially by the government.

The Yozma Program The design of Yozma was an outcome of a long and intensive preparation which included visits of OCS officers to Silicon Valley, interviews with USA, entrepreneurs, venture capitalists and investment banks, and visits to Small Business Administration officers. It was based on implementation of US proven VC characteristics related to organizational form and operating routines, after taking care of necessary adaptations to the Israeli environment such as using the NASDAQ as an exit path rather than the local Stock Exchange. The explicit objective was to create a solid base for a competitive VC industry with critical mass where decisive elements were learning from foreign limited partnerships and to acquire and develop international networks.

178


The Yozma program was based on a US$100 million government-owned VC fund oriented to two functions: first, fund of funds- investment in 10 private VC funds, comprising altogether US$ 80 million, and, second, to direct investments toward high-tech companies to which US$ 20 million was devoted. The latter investments were to be undertaken by government wholly owned Yozma Venture Fund (which was privatized in 1997). However, the basic thrust was to promote the establishment of a domestic, private limited partnership VC industry that invested in young Israeli high-tech start-up firms (early phase investments) with the support of government and with the involvement of reputable foreign financial institutions (generally a private equity company). The funds involved were to be managed by new independent Israeli VC management companies.9 Each such Yozma Fund would have to engage one foreign institution together with a well-established Israeli financial institution. This emphasizes the point that the Yozma program favored entry of professional managers or of individuals with VC-related abilities into the infant VC industry. Moreover, the insistence on creation of a formal organizations as a precondition for becoming a Yozma fund suggests that its initiators understood the significant role of institutions in the process of learning, generating, and accumulating capabilities and reputation.10 In an approved fund that fulfilled these conditions, the government would invest 40 percent (up to US$ 8 million) of the funds raised. Thus US$ 80 million of government venture contribution seeded 10 Yozma funds (Table 9.3). The Yozma program was also instrumental in setting up most of the VC management firms that were responsible for the funds.11 Five VC firms were founded in 1993: Gemini, Star, Pitango, Walden; Inventech; three in 1994: Concord, Eurofund, and JVP; one in 1995: Medica; and one in 1997: Vertex. Over 200 start-up companies received VC funding, the total capital raised and invested is shown in Table 9.4. Yozma did not simply provide capital and risk-sharing incentives to investors as was common in other government VC support programs.12 Rather, the main incentive was on the upside, each Yozma fund had a call option on government shares at cost (plus 5 percent interest) for a period of five years. Demand side support was assured not by Yozma itself but by the other parts of the Israeli innovation and technology policy, such as grants to business R&D program and by the relatively recent technological incubators programs. Another major point was the pursuing of an

179

The Emergence of Israel’s Venture Capital Industry Table 9.3. Yozma funds and associated VC companies VC name

Foundation date

Number of funds 1993–2003

Capital under management 1993–2003 (in M$)

Capital to be raised in 2004

Star

1989

11 (1989–2000)

NA

Gemini Inventec Pitango Walden Yozma Concord JVP Eurofund Medica Vertex Total

1993 1993 1993 1993 1993 1994 1994 1994 1995 1997

3 (93, 97, 00) 1 (93, 97) 4 (93, 96, 00) 2 (93, 98, 00) 3 (93, 98, 01) 3 (94, 97, 00) 4 (94, 97, 99, 01) 2 (94, 99) 2 (95, 00) 4 (97, 97, 00, 02) 39

975 (90–7: 275, 98–9: 300, 00: 400) 346 (36, 110, 200) 33 (20þ13) 665 (20, 145, 500) 184 (33, 61, 90) 150 (20, 80, 50) 280 (20, 80, 180) 675 (20, 75, 175, 405) 72 (20, 52) 65 (15, 50) 545 (39, 46, 160, 300) $3,990M

$200M NA $350M NA $80M $150M $250M NA $70M $150M $1,250M

Table 9.4. Venture capital raised and invested Year

VC raised

VC under management*

VC invested (% of foreign)

VC investment as % of GDP

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

58 160 372 374 156 397 729 706 1,851 3,701 1,100 63 300

80 240 612 986 1,142 1,539 2,268 2,974 4,825 8,504 9,546 9,609 9,600

— — — — — — 440 589 (36%) 1,011 (43%) 3,092 (59%) 1,985 (59%) 1,140 (58%) 1,000 (61%)

— — — — — — 0.41% 0.54% 0.9% 2.6% 1.65% 0.96% 0.84%

Source: Avnimelech (2004). Note: Management companies that invested at the current year and with at least $3M available for investments at the end of the year.

aggressive early start-up firms phase investment policy, spearheaded by the governmentally wholly owned Yozma Venture Fund. A condition of receiving Yozma Fund status was participation of highly reputable and capable foreign limited partners. Table 9.5 indicates the foreign (and Israeli) Limited Partners that participated in the Yozma program. This was important since it became a mechanism for learning and expanding international networks. Personal networks were essential since international knowledge related to VC is basically tacit. Moreover, these

180

The Emergence of Israel’s Venture Capital Industry Table 9.5. Foreign partners of Yozma Fund and privatization VC Name

Foundation date

Privatization date*

Reputable foreign investors

Reputable Israeli investors

Gemini Inventec Pitango Star Walden Concord JVP Eurofund

1993 1993 1993 1993 1993 1994 1994 1994

1998 1998 1998 1998 1998 1998 1998 NA

Advent Van Lear Group, Docor HarbourVest, Chase Capital TVM, Siemens Walden International AVX AXA, Jafco, Bank of Taiwan Daimler-Benz

Medica

1995

NA

Vertex

1997

2001

Yozma

1993

1998

Buxter, Soros PE, Bank of Japan Singapore Technologies, Vertex international; NA No private investors

DIC Mercator Dovrat shrem PCM KLA Israel Kardan Technologies HVP Development Federmann Enterprises Teva

Note: * For Yozma Funds privatization means implementation of call option for government shares.

partners, through their international links and reputation, would contribute to the development of the portfolio companies of Yozma Funds. VC learning was also assured by participation of the Yozma Venture Fund manager (e.g. Yigal Erlich and other OCS officers who represented the government) at the board meetings of all Yozma funds (they acted as a central node in a vast information network). Also the stimulation of co-investment among Yozma Funds served that purpose. Culturally speaking the stage was set for a lot of informal advising and interaction among fund managers. The eventual privatization of Yozma Funds was part of Yozma’s design. The limited partnerships had a call option to purchase the government of Israel’s share approximately at cost anytime during a period of five years. Since the exit market became increasingly favorable till year 2000, most management firms implemented this option. One exception was Medica, a Biotech-oriented Yozma Fund which, due to this fact, generated lower returns (at the end of its five first years of operation) compared to most of the other ICT-oriented VCs. While Medica did not make use of the call option it eventually achieved an annual rate of return of more than 50 percent. The other case was Eurofund, which achieved a relatively low rate of return after its first five years of operation. Note that the option to purchase government’s share in Yozma funds is in effect an incentive to the upside since it will be materialized only when the fund makes more profits than the alternative profits that could be earned with the resources dedicated to this purpose. Note also

181

The Emergence of Israel’s Venture Capital Industry Table 9.6. Capital raised by private equity organizations in Israel, 1991–2003 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Yozma VCs Private VCs Publicly Traded VC PE Funds Investment Companies Total PE

0

0

149

40

35

0

52

0

0

0

0

0

0

49

27

33

72

100

309

568

594 1,552 3,682 1,304

76

118

0

54

22

0

0

0

29

8

44

185

6

86

0

0 9

45 34

128 40

242 20

6 25

24 80

56 134

67 141

108 149

89 601

0 83

110 0

435 5

58

160

372

374

166

413

839

810 1,853 4,557 1,393 272

558

Source: Avnimelech (2004).

that this design feature directly flows from the fact that Israel’s targeted VC program included a government VC contribution, mainly designed as a fund of fund function. Table 9.6 shows the changing structure of Israel’s private equity market (including Investment Banks) and VC industry since 1991. The main categories of the VC industry are Yozma Funds which have been privatized VC companies;13 together with other private VCs firms (Other), and public VC firms traded in the Tel Aviv Stock Exchange. Yozma Funds started in 1993 where a total of US$ 149 million was raised, with the amount gradually (but not monotonically) declining to zero in 1998 and afterwards. The data are consistent with the fact that Yozma did not crowd out prior private VC (which already existed before 1993) but positively stimulated this category of funds which, through accelerated growth, became the mainstay of Israel’s VC industry and high-technology industry. Thus, Yozma triggered a cumulative process of growth of Israel’s VC industry. Note also the relative insignificance of the publicly traded component of Israel’s VC industry: while such companies existed prior to Yozma (1992, in the aftermath of the Inbal program) their share was never significant beyond 1993–4 and continuously declined thereafter (except perhaps during 1999–2000).

Emergence of a Venture Capital Industry: Policies and Pragmatism The Israeli experience in developing a VC industry is closely linked to the three stage model, where smooth transmission across the three stages has been a distinct feature of the process. The third stage is worth

182


emphasizing, i.e. the importance of adopting a clear strategic view on the promotion of high-tech and innovative clusters and the leadership role by policymaker in implementing targeted policies. Targeted policies seem to be inherently more complex than horizontal policies and therefore require a longer and more complex preparatory period. There are clearly a priori reasons for this, particularly in contexts with very little experience with innovation and the appropriate policy. Consequently, business sector capabilities on the one hand coevolved with enhanced policy capabilities and the introduction of new policy institutions. This process is a specific instance of the coevolutionary processes between technology, industry structure, and institutions (Nelson 1994). Table 9.7 summarizes. Government policy played an influential role in the formation of the industry. In the initial phase grants to business R&D became the backbone of Israel’s innovation policy beginning in 1969 and provided the basis for the development of a new civilian-oriented high-tech cluster during the 1990s. During the intermediate phase, a new set of policies were implemented to support high-tech start-up in a process of experimentation and learning by both the public and private sectors. In the late 1980s and early 1990s officials in the treasury and the OCS realized that despite massive government support for R&D there were clear market and system failures, which blocked the successful creation and development of start-up companies. While an important problem was insufficient finance for post-R&D activities (especially for start-up firms whose access to bank finance was Table 9.7. Start-ups, VC funding, OCS grants, exits, and closures, 1991–2003 Year

Foundation*

VC backed start-ups

OCS Grant to start-ups

IPO** (VC backed)

M&A*** (VC backed)

Closures

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

NA NA NA NA NA NA 250 300 450 550 150 98 134

NA NA NA NA NA NA 119 152 208 372 164 108 106

179 (109) 241 (165) 245 (179) 291 (218) 236 (146) 257 (200) 219 (158) 217 (156) 212 (169) 172 (126) 260 (193) 284 (199) NA

3 (0) 9 (1) 10 (3) 9 (2) 9 (4) 27 (14) 22 (9) 17 (8) 24 (18) 26 (25) 2 (2) 0 (0) 0 (0)

0 (0) 0 (0) 0 (0) 3 (2) 5 (5) 10 (10) 6 (6) 14 (14) 12 (12) 23 (22) 7 (7) 6 (6) 8 (8)

NA NA NA NA NA NA NA NA 94 161 386 412 118

Sources: IVA, Money Tree, OCS, and Globes Newspaper. * Estimates; ** established after 1990; *** not including fire sales or buybacks.

183


limited) this was only part of the problem. No less important were weaknesses in management abilities, business know-how, production/marketing capabilities (and orientation) and links/networks with global capital and technological markets. In response to this a gradual shift in the OCS’s policy objectives gradually took place—from promotion of R&D to explicit enhancement of start-up firm formation, survival and growth.14 The head of the OCS, Yigal Erlich, pondered how to make his office’s support more effective. He could, according to interviews done in 1998 and 1990, not find even one real success ‘similar to those we see today’. The basic problem was lack of capability to grow after the product development phase. By identifying a joint finance and marketing/management skills’ gap the system failure was defined and characterized in terms of absence of a particular type of financial institution—venture capital. The Inbal program was the first attempt at targeting the VC industry. It was launched by the treasury in 1992 one year before the implementation of Yozma. Its central idea was to stimulate publicly traded VC funds by guaranteeing the downside of their investments. The mechanism used was a government insurance company (Inbal) that guaranteed VC funds traded in the Israeli stock market (TASE) up to 70 percent of initial capital assets. The program imposed certain restrictions on the investments of the VC companies covered by the program (Inbal funds). Four Inbal funds were established. They and the Inbal program as a whole were not a great success. Inbal funds’ valuations in the stock market were low, similar to holding companies’ valuations and the funds encountered bureaucratic problems. More significant was the fact that the program did not attract any adding value agents or capabilities. Moreover, the funds did not succeed financially and did not raise additional capital. Eventually all four Inbal funds quit the program.15 The Inbal program not only failed to overcome the market failures related to the pool of capital aspect of the VC industry but it also did not target any of the system failure causes related to VC industry emergence. There was no mechanism for drawing professional VC agents into the program, it did not generate VC companies with adding value capabilities, and it failed to promote collective learning. In addition, there were links created to later stage VC pools or to a significant IPO market. Also, its model of VC company organization was not imitated, and the social impact of the Inbal program was very low. Still, it is important to mention that policymakers and businessmen alike learned from Inbal’s weak impact particularly about the disadvantages of public VC organizations. These included company taxation (which a limited partnership could

184


avoid), the difficulty of having investors contribute to the operation of the fund as well as difficulties in rapidly exploiting the reputation earned from early exits in order to raise new capital. There were severe limits on managements’ decision-making flexibility and on management compensation and no upside incentives. Awareness of these favorably influenced the design and implementation of Israel’s successful Targeted Policy—the Yozma program.16 The start-up of firms experienced a quantum jump in the early 1990s and the stock of start-up firms must have approximated 300 by 1993 (Avnimelech and Teubal 2004a, 2005a). This mass of start-up firms companies and their quality implied a very favorable environment for the fledging VC industry as far as the demand side or deal flow is concerned. Indirect evidence of the existence of high-quality start-up firms is the fact that VCs in the pre- and early emergence period (Star, Giza and Mofet, and to a lower extent Athena) were either making profits or soon going to do so.17 What is implied is that very rich pickings existed among the growing stock of firms; and that the small numbers of leading investors and VCs active at the time identified and invested in them.18

System Failure Leading to the Yozma Program19 Despite the favorable background and preemergence conditions created during the 1969–92 period a number of specific causes of system failure stood in the way of a purely market-led successful emergence of a VC industry around 1993. These factors, duly identified by policymakers at the time and reflected in the design of Yozma, stretched from difficulties in accessing intelligent and reputable foreign partners, lack of critical mass of capabilities and resources, an overall coordination of policies, to strict VC strategies and a credible signaling from highest political level that the development of VC and high-technology clusters was a key sector in the Israeli economy. Hence, successful VC emergence depended on overcoming the above obstacles. Yozma succeeded in triggering a successful cumulative process of VC growth and impact, resulting in a new system of innovation and a new Silicon Valley type of high-tech cluster. Also ICT high-tech exports quadrupled during the 1990s—from US$ 2,100 million in 1990 to US$ 8,800 million in 2002; and as part of the enhanced importance of the start-up firms segment of high-tech, the share of start-ups related output to total high-tech output increased considerably during the decade, as shown in Table 9.8.

185

186 Table 9.8. ICT and software manufacturing: sales, exports, and employees Year

ICT sales

ICT exports

ICT employees

Sale per employee

Software sales

Software export

Software employees

Sale per employee

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

3,300 3,600 4,000 4,600 5,200 5,900 6,500 7,200 8,000 8,600 12,500 11,250 10,000

2,100 2,280 2,660 3,200 3,750 4,300 4,880 5,700 6,550 7,130 11,000 9,750 8,800

32,000 33,000 34,200 36,400 37,600 39,200 42,000 43,700 45,600 48,000 54,800 47,000 43,200

103 109 117 126 138 151 155 165 175 179 228 239 231

400 540 600 700 800 950 1,300 1,780 2,350 2,950 3,700 4,100 2,800

75 110 135 175 220 300 600 1,000 1,500 2,000 2,600 3,000 1,900

5,000 5,000 5,500 6,200 7,000 7,700 8,500 10,000 11,500 13,000 14,500 15,000 13,200

80 108 109 113 114 123 153 178 204 227 255 273 212

Source: Israel Association of Electronic. Note: Sales are in thousands of US$.


Yozma’s Success: Lessons for Other Countries The Context/Timing of Targeted VC Policies For cases like Israel’s where VC emergence was a policy-led process, the existence of favorable demand conditions was necessary but not sufficient for the successful VC emergence in phase 3. A timely and suitably designed targeted policy was also a requirement. Right timing was important due to internal and external reasons. The earlier the timing of the targeted policy the greater the risk that domestic demand (for the services of the future VC industry) would not have had enough time to build up to the level which, in conjunction with the policy-induced increases in supply, would trigger a cumulative process of VC emergence. On the other hand, the shorter the period between the initiation of such a process and the next downturn of the world VC industry, the less time available for industry emergence and for a significant high-tech impact to materialize. At the same time, the success of Yozma was dependent on complementary policies. Transformation generally requires a portfolio of coordinated policies rather than a single program, action, or institutional change (Avnimelech and Teubal 2004a, 2004b, 2005b; Teubal 2002). One reason for this is the need to continue supporting demand at least for some time, until a virtuous coevolutionary process takes hold. This function was performed by horizontal grants through the business sector R&D program (disbursements of which continued to grow till the end of the 1990s, the technological incubators program, and indirectly by the Magnet program).

Causes for VC Policy Failure We have identified at least seven possible reasons why a VC emergence policy may fail. Those are: . Inappropriate VC program design and lack of pre-emergence R&D/ Innovation capabilities in the business sector. . Insufficient demand for VC industry services. . Inappropriate entrepreneurial culture and an unsupportive external environment. . Lack of capabilities critical to develop a VC industry, especially global networks and cross-border contacts.

187


. Strong and influential traditional financial institutions that may block or hamper VC industry emergence. . A self-reinforcing path-dependence process whereby early emergence VC industry develop competencies in the late and nontechnological private equity markets. . Weak access to effective exit markets. We will highlight some of these that we hold as particularly important.20 First, a common cause of VC policy failure is unfavorable domestic context, in particular a disregard of the importance of domestic demand for VC services. The causes can mainly be attributed to a lack of a critical pre-existing mass of start-up companies which are the demand agents for an early phase VC industry, and constraints associated with the cultural and institutional framework. In our view, a number of other countries within as well as outside the OECD suffer from cultural and institutional constraints that impede an emergence of technological entrepreneurship (Becker and Hellmann 2002; Black and Gilson 1999). In some cases unfavorable domestic background conditions are the result of an excessively early implementation of VC policies, that is if policies would have been implemented later the required pool of start-up firms and/or the underlying cultural and institutional constraints would have been overcome. This may have been the case of India who implemented VC policies very early (during the late 1980s)—even before the emergence of Bangalore as a center for software and IT services. As a general proposition it is important to point out that, at least during the last two decades, VC policies should not have been the initial or main thrust of a policy designed to generate large numbers of start-ups. Rather, they should have come after necessary R&D capabilities have been generated in the business sector and a minimum pool of innovative start-up firms are in place. We view with skepticism the simultaneity perspective as regards the development of VC and start-ups, which has been advocated in the literature (e.g. Black and Gilson 1998, 1999; Gilson 2003). In contrast, we strongly emphasize a sequential or phased strategy where—prior to the onset of a virtuous VC and start-up coevolutionary process can take over— a presence and critical mass of innovative start-up firms is required. Another frequent cause of VC policy failure is deficiencies in VC policy design. An example is the German program of the early 1980s which only promoted one intermediary (VC) rather than a number of new VC agents, as was the case with Yozma. Moreover, and as important, is the insufficient attention that was given to issues of organization, capabilities, and strategy

188


of the new agents. These and other design problems seem to have been the result of a VC as pool of money perspective, rather than VC as an industry perspective. Finally, it is also important to point out that the implementation of (even well-designed) VC policies may encounter obstacles, some predictable, others not. Hence, flexibility at the firm level as well as at the policy level is a necessary condition to develop a competent and sustainable VC industry.

Conclusions Israel’s policy-led experience with VC and high-tech cluster reconfiguration is unique in terms of the specifics of the context, the timing and the design of the high-impact policies (particularly the targeted VC policy— Yozma) implemented. It seems to have been one of the few policy schemes, which implicitly followed an evolutionary perspective, and it was a resounding success in terms of successful VC emergence. Although consolidation is not yet assured there are very signs starting in 2004 (Avnimelech and Teubal 2005a). Moreover, because of the fertile soil for VC recovery change was created because of the targeted policy, it seems far-fetched to state that the geographic location of the innovative hightechnology clusters in Israel was random. While other countries cannot in detail copy the Israeli experience, there are a number of useful analytical lessons learned. First, this is an outcome of a more generic three phase model of successful evolution of an innovation and technology policy oriented to business sector R&D, VC, and high-tech industries, where a clearly discernible element is strong coevolution between the business sector and innovation/technology policy. There are strong reasons to believe that an analysis of the Israeli-specific variant may clarify the policy options available to other countries aiming at developing VC industries and high-tech clusters. A related conclusion of our analysis is the complexity of the design of the targeted VC-directed program. In Israel it included an explicit government venture investment component without falling into the trap of a government VC company (the fund of fund function was crucial) and a highly original set of upside incentives. Thus, the government fulfilled an apparent signaling function, underlining the strategic role of the VC industry in developing the high-tech sector. In addition, a crucial multidimensional coordination process (particularly related to accessing world-class

189


intelligent investors in the area) was pursued, inserting selection mechanisms that involved VC company teams, VC organizations, and VC strategies. Moreover, implementation assured a process of collective learning as well as investment coordination among the various VC funds created under the sponsorship of Yozma. The upshot would seem to be that there is a critical mass of effort that should go into the policymaking process and to develop high-level capabilities. These requirements might separate those countries capable and willing to go through the required lengthy policy process and policy cycle culminating in VC policies from those who are not willing or capable to do so.

Notes 1. A number of papers (Avnimelech 2004; Avnimelech and Teubal 2004, 2005a) analyze the process of emergence and development of Israel’s VC industry. 2. The SBIC program was not a VC directed but a VC-related program. Unintended effects of this program played a role in what was essentially a market-led process of VC emergence (further empirical research is required to ascertain statement). 3. The Israeli case also suggests the need for a well-developed Innovation and Technology Policy (ITP) infrastructure of capabilities and institutions. 4. This is a feature of young markets. VC cooperation involves collective learning, syndication, and so on. 5. See Avnimelech and Teubal (2004, 2005). 6. Lerner (1999) has shown that pension fund deregulation played a more important role than reductions in capital gain taxation in the recovery of the USA’s VC industry during the 1970s (Israel’s restructuring phase, see also Avnimelech, Kenney, and Teubal 2005). Moreover, he shows the quantitative and qualitative importance of the SBIR program’s subsidies after 1982 (i.e. during consolidation of the US industry). 7. This type of R&D could be termed regular, or classic, R&D to differentiate it from generic and cooperative R&D, which is of a more infrastructural type (the Magnet program—implemented in phase 2—is an example of a generic, cooperative R&D program). The latter’s objective is to generate knowledge, capabilities, and components rather than directly marketable outputs. The output of generic R&D would facilitate (or become inputs) to a subsequent regular R&D activity. 8. Learning, including experience-based learning triggered by increased R&D in the business sector, is the main factor leading to enhanced R&D/Innovation capabilities.

190

The Emergence of Israel’s Venture Capital Industry 9. This would ensure that the traditional financial institutions would not block the development of this new industry (as was the case of Germany in the 1970s and 1980s, see Fiedler and Hellmann 2001). 10. In choosing among candidate management teams for Yozma Funds the government also took into account the high-tech experience of team members. 11. To this must be added the US$ 20 million of the government-owned Yozma Venture Fund, and the fact that two of the ten Yozma funds totaled US$ 35 million each (rather than US$ 20 million). Thus total funds directly raised by the Yozma program were about US$ 250 million. 12. It did not provide guarantees or tax benefits, nor was it accompanied by new regulation rules for pension funds or corporate law. 13. The data on capital raised for each one of these VCs refers to their Yozma Fund only, not to other funds raised by these companies (e.g. follow up funds). 14. This means that the R&D additionally criterion was increasingly perceived as being irrelevant if not accompanied by additional criteria pertaining to the economic impact of business sector R&D. 15. Today all of them are held by one holding Company, that is Green Technology. 16. Inbal played two additional roles. First it was an important part of the variation process preceding selection of the right configuration of VC; second it also contributes to signal the government of Israel’s determination to create a VC industry. 17. Another ex post indicator of the high quality of start-up firms in the early emergence phase is the fact that most Yozma VCs had very high returns (5 out of the 11 Yozma funds had an annual rate of return of more than 100 percent). 18. This also meant that the market was already pointing the way to activities typical of VC and start-ups. 19. For further analysis, see Avnimelech and Teubal (2005b). 20. For an extended analysis of VC emergence and policy failures, see Avnimelech and Teubal (2005a, 2005b).

191


Part III Crafting Cluster and Economic Development Policy


10 Clusters and Clustering: Stylized Facts, Issues, and Theories Luigi Orsenigo

The notion of clusters has spawned an enormous literature, which includes a variety of methodologies and disciplines. While these contributions have significantly improved our knowledge of clustering, several fundamental issues remain controversial and poorly understood. Among them are the relative role of alternative mechanisms that have been proposed as possible explanations of the observed concentration of productive and innovative activities, the nature of the dynamic processes as distinct from the ingredients that might produce clustering and finally the appropriate role of policy. Yet it has become increasingly questioned how precisely space influences innovation and what are the specific mechanisms leading to agglomeration. Some authors even come to the conclusion that the role of geography might have been vastly exaggerated. This chapter discusses the mechanisms and processes that generate geographic concentration of innovative activities. Rather than addressing the whole literature on clusters, this chapter concentrates on the relationships between clusters and innovation as the source of regional competitive advantage. The intention is to provide a more nuanced view to the factors behind the origin and advantage of clusters. First, the relevance of the spatial dimension is a reflection of the concentration of cutting edge research in selected universities and public research organizations, and specific to the level of research groups, technological applications, or product markets. Second, the relevance of localized processes of creation and transmission of knowledge, which are only partially linked to knowledge spillovers, has more to do with active processes of construction of supporting institutions. Third, academic research is one source for

195

Clusters and Clustering

technology and for attempts to privately appropriate the economic benefits, but there are also other less formal mechanisms for the creation, diffusion, and economic exploitation of knowledge. A broad, but brief, overview of some key issues in the literature of clusters—to set the stage for the subsequent policy discussion—is presented.

Clusters of Innovative Activities: The Spatial Embeddedness of Innovation Recently a variety of intertwining factors have focused attention on the phenomenon of clusters of innovation as distinct from clusters of production activities. A primary reason is the increasing recognition of the role of innovation and more generally of knowledge as an engine of economic growth. Thus, academics as well as policymakers have started to consider how innovation can be generated and sustained in specific geographic areas and regions. Three findings have sparked new research in the field of the geography of innovation. First, it has been observed that innovation is more spatially concentrated than production activities (Audretsch and Feldman 1996). Second, concentration of innovative activities that drive economic growth occurs at spatial dimensions that are much smaller than nationstates, which are the typical units that make economic policy. Third, firms located in innovative clusters tend to be systematically more innovative than firms located elsewhere. Jointly, these observations have led to the notion that innovative processes are intimately embedded in space and have opened a set of questions concerning the specific characteristics of particular areas that make them conducive to innovation. There seems to be a wide consensus that clustering in innovative activities cannot be simply explained by some sort of given and immobile endowments, but that powerful agglomeration forces must be at work in order to produce geographic concentration. However, what these forces are precisely and what is their relative explanatory role remain unclear. In some cases, the doubt can legitimately arise that perhaps some of these agglomeration mechanisms could be interpreted more simply as endowments or vice versa, as it will be argued at more length later on. In fact, developments in the geography of innovation have focused their explanations essentially on various reformulations of the three fundamental sources of agglomeration externalities originally suggested by Marshall (Henderson 1986; Krugman 1991a). First, economies of

196


intraindustry specialization: a localized industry can support a greater number of specialized local suppliers of industry-specific intermediate inputs and services, thus obtaining a greater variety at a lower cost. Second, labor market economies: a localized industry attracts and creates a pool of workers with similar skills, which benefits both the workers and their employers. Third, ease of communication among firms: information about new technologies, goods, and processes seem to flow more easily among agents located within the same area, thanks to social bonds that foster reciprocal trust and frequent face-to-face contacts. Therefore adoption, diffusion, and innovation seem faster and more intense in geographic clusters than in scattered locations. That is, some knowledge spillovers exist, which are geographically bounded.1 The notion of knowledge spillovers has taken a central role not only in scientific analysis but also in policymaking, where a key emphasis has been attributed to the idea that interventions should be directed primarily toward the attempt at facilitating such knowledge flows and spillovers, through, for example, offices for technology transfer and all kind of measures for strengthening university–industry relations. However, it has become increasingly acknowledged that the evidence supporting the role of knowledge spillovers is largely indirect and that it is quite difficult to clearly separate knowledge spillovers for other types of pecuniary externalities and more generally between Marshallian externalities and more classic urbanization externalities or even natural endowments (Breschi and Lissoni 2001; Ellison and Glaeser 1999; Glaeser et al. 1992; Henderson 1999). Similarly, knowledge within clusters in many cases does not appear to simply spillover. Rather, access to such knowledge seems to require deep involvement in the research process and bench-level scientific collaboration and the conscious investment of resources not simply to search for new knowledge, but to build the competencies to absorb the knowledge developed by others (Almeida and Kogut 1999). Finally, in other cases, knowledge flows occur via (localized) mobility of researchers and the workforce and are mediated by market transactions or other institutionalized or quasi-institutionalized mechanisms involving not simply mutual trust and face-to face contacts, but highly complex economic and social structures (Zucker et al. 1998). In general, the very origin of knowledge spillovers seems to be embodied in a specialized workforce exemplified by social cohesion and mutually beneficial incentives to share information.

197


Other studies suggest that local sources of knowledge are key to determining success in the development of new products and processes only in areas with a large accumulation of knowledge (e.g. Echeverri-Carroll and Brennan 1999; Giuliani 2004; Lyons 1995). Innovation in firms located in cities with a relatively small accumulation of knowledge depend on the relationships with universities and other high-technology firms (suppliers and customers) located elsewhere, especially in higher-order urban centers. The most dynamic and innovative firms look for knowledge embodied in engineers and scientists wherever they are available, and are not necessarily constrained in this by geographic barriers. Local knowledge sources appear to be relatively less important for firms located in lower-order regions. For these firms, local universities are viewed as suppliers of skilled workforce, rather than loci of innovations or sources of product ideas or spillover effects. In order to sustain high rates of innovation they must develop linkages with actors located in higher-order regions.

Clusters and Networks The issue becomes even more tangled whenever the cluster is directly assimilated by the concept of networks. Indeed, almost all the available analyses of clusters rely—directly or metaphorically—on the idea that the agglomeration forces that keep innovative activities localized are expressions of networks of localized relationships among agents. However, the concept of networks varies drastically. In some cases, for example networks are considered as a specific new form of organization of innovative activities, which is substituting the traditional model based on internal R&D (Powell 1996). According to this kind of interpretation, innovative activities tend to cluster because such kinds of networks exist: the cluster is the outcome of the existence and development of local networks, rather than networks emerging as a result of specific localization factors. Partly overlapping with these views, in other cases, the notion of networks is strictly associated to the notion of collaboration. According to this perspective, cooperation in innovative activities and interactive learning are the distinct property that defines and identifies (successful) clusters (Cooke 2002; Maskell 2001). A network-like structure is a typical property of clusters and networks should be viewed as the proper unit of analysis to investigate innovation: knowledge resides in the network and not simply in each of its constituent nodes. Within innovative clusters, firms learn

198


through a variety of types of interactions, ranging from user–producer relationships, formal and informal collaborations, interfirm mobility of skilled workers and the spin-off of new firms from existing firms, universities and public research centers. Embeddedness of local firms in a thick network of knowledge sharing is supported by close social interactions and by formal and informal institutions that promote the development of trust among participants in the network. However, the concepts of networks and clusters do not coincide. More important—and quite obviously—networks do not necessarily imply cooperation, trust, and distributed knowledge. As such, networks are just a language (in this sense, having the same methodological status of game theory) to represent and analyze interactions among agents. And in its most developed form—social network analysis—this approach is firmly grounded in sociology, its main object of analysis being social relations. Thus, there are many different types of networks and many different principles that determine their structure and evolution. Thus, formal applications of network analysis to the study of innovative clusters as well as an exploding number of case studies have highlighted the extreme diversity in their structure, logic, and dynamics. In some cases, highly hierarchical structures are observed (Orsenigo, Pammolli, and Riccaboni 2001). More generally, these studies had the great merit to show that clusters are characterized by different highly structured patterns of knowledge diffusion and generation, produced by the interaction of a host of overlapping social and economic relations. This observation further weakens any simplistic description of the advantages of colocalization as the result of some kind of local externalities. If anything, the ability to tap into the local knowledge base and to exploit the other possible agglomeration externalities appears not to be unrestricted, but structured and mediated through specific social, organizational, and economic mechanisms. Hence, clusters might emerge as the outcome—or as a specific subnetwork—of sets of relations which are not necessarily based on spatial proximity, but on other forms or contiguity, like organizational proximity, epistemic communities, or communities of practice (Rallet and Torre 1999). Conversely, clusters might well result as the effect of the coalescence of different networks. One particularly important result of this stream of literature is that successful clusters are much more open and outward oriented than the conventional interpretations would suggest. Rather than simply leveraging the knowledge available within a cluster, the existence of weak and strong ties with agents located outside the cluster itself increasingly

199


appears to be a crucial feature of the more innovative clusters (Bresnahan, Gambardella, and Saxenian 2001). As such, this finding does not imply by any means that local sources of knowledge are unimportant. On the contrary, it might be interpreted as confirming that the ability to tap external sources of knowledge—absorptive capacities—is itself a function of the local technological capabilities. However, in more general terms, this observation links with the broader question concerning the existence an ideal knowledge network structure that can be associated with high-innovative performances. In some respects, the debate about the advantages of Silicon Valley versus Route 128 might be considered as a particular version of this direction of research. In this respect increasing attention has been devoted to the studies of dense networks (Coleman 1988), to the importance of structural holes (Burt 1992) and to small world type of networks (Milgram 1967; Watts and Strogatz 1998). As discussed in Ahuja (2000) and Giuliani (2004), most likely an optimal network structure for innovation does not exist; rather, different structures are likely to show different types of advantages and disadvantages. For example, dense networks tend to favor the formation of trust, which in turn limits opportunistic behavior (Coleman 1988) and encourage cooperation and diffusion of more high quality, fine-grained knowledge (Uzzi 1997). Networks characterized by structural holes, instead, allow firms to expand the diversity of knowledge they can have access to (Ahuja 2000) and reduce the probability of negative lock-ins (cf. Gargiulo and Benassi 2000). Similarly, small worlds are considered a more efficient network structure as compared to a randomly interconnected one, to the extent that fewer but more distant linkages increase the probability of accessing diverse knowledge and allow efficiency gains in the processes of knowledge diffusion (Cowan and Jonard 2004).

Forms of the Organization of the Cluster and the Nature of Technology In some respects, the debate about the advantages of Silicon Valley versus Route 128 might be considered as a particular version of this discussion on the relative efficiency of different cluster or network structures. Saxenian (1994) proposed the extremely influential argument that the superior performance of Silicon Valley was related to the particular form of organization of innovative activities that had been developing over time, based indeed on network-like structures of interactions among entrepreneurial

200


agents. On the contrary, Route 128 fared worse as a consequence of the dominance of a more traditional organizational structure, based on large vertically integrated firms. Kenney and von Burg (2001) object to this interpretation and argue that the main difference resided not so much in the organizational structure but in the economic success of the relevant technology or set of technologies and related products of the cluster. Personal computers were a better product platform for firm, industry, and regional economic growth. Relatively little attention has been paid to the relationships between the structure and performance of a cluster and the nature of the knowledge base and the market opportunities on which the cluster is based. Yet the properties of the relevant technology or technologies are likely to have an important role in shaping not only the critical resources or endowments that might determine the success or failure of clusters, but also the relative importance of alternative mechanisms of agglomeration and, of course, the relative efficiency of alternative network structures. In fact, the debate on knowledge spillovers has concentrated on one particular property of knowledge, that is tacitness. However, while the importance of tacitness cannot be overemphasized, knowledge and technologies are characterized by other dimensions that have a profound impact on the ways through which innovation takes place (Malerba and Orsenigo 2000). The nature of technological regimes bears a close relationship to the patterns of innovative activities (Breschi, Malerba, and Orsenigo 2000; Malerba and Orsenigo 1997). Thus, for example the nature of technological opportunities and the space, in which search for new products and processes occurs, implies specific incentives and possibilities for the entry of new innovators and more generally, for the speed of technological change and for the organization of innovative activities. High opportunities in technologies offer ample possibilities of invention along diverse trajectories and over a wide range of products, processes, and market segments. These are likely to be associated not only to high rates of innovation, but also by a large and rapidly changing population of innovators, who explore different avenues and open new market niches. Similarly, technologies which involve the access to and the mastery of differentiated fragments of knowledge open up possibilities for vertical and horizontal specialization and the development of adequate organizational devices for the integration of the relevant knowledge. To some extent, the discussion about the role of specialization versus diversity might be interpreted also along these lines. Thus, Feldman and

201


Audretsch (1999), for instance support the view that diversity matters more than specialization, finding that the number of innovations owes more to the presence of other industries whose science base is related to that industry rather than to the industrial specialization. In a similar vein, Bresnahan, Gambardella, and Saxenian (2001) suggest that clusters tend to form taking advantage of new technological and market opportunities that have not already been exploited. The degree of cumulativeness and the appropriability conditions of technological advances are also likely to influence the structure of the cluster and the processes leading to the cluster. Strong cumulativeness and a tight appropriability regime make the circulation of knowledge more difficult and tend to strengthen and reproduce over time the advantages of early innovators. This might lead to stronger and persistent concentration of innovative activities in few firms to more hierarchical network structures and under some further conditions to higher degrees and persistence of spatial concentration (Breschi and Malerba 2001; Malerba and Orsenigo 1997). The combination of these and other properties of knowledge can therefore give rise to a large variety of different processes and structures of clusterization.2 In turn, alternative characteristics of the clusters—in terms of their endowments, patterns of internal interactions, mechanisms of agglomeration—may display different degrees of fit with the relevant technological regime. More systematic attempts to map and taxonomize the properties of technologies and the structures of clusters might turn out to be a very promising avenue for future research (Casper and Kettler 2001; Cooke 2002; Hall and Soskice 2001).

The Dynamics of Clusters: Genesis, Evolution, and Innovation In a recent paper examining the reasons why Detroit emerged as the capital city of the automobile industry and maintained its leadership over time, Klepper (2002a, 2002b) proposes empirically and tests the hypothesis that this process was the outcome of a combination of chance (the presence in Detroit of a few key individuals who were not, however, closely related to each other) and processes of spin-offs of new firms by incumbent companies. On the basis of the key assumption that the best firms generate a larger number and more efficient spin-offs, the model produces patterns of agglomeration without relying on any standard notions of externalities. First, Klepper’s argument does not rely on the

202


standard notions of externalities, but introduces the idea that other forms of dynamic increasing returns might lead to clusterization, in particular as it concerns the cumulativeness of innovative processes. Cumulativeness plays indeed a fundamental role in the explanation of the processes of concentration at the industry level and constitutes a plausible further candidate explanatory variable for concentration at the spatial level (Dosi et al. 1995; Malerba et al. 1999; Sutton 1998). As long as innovation is cumulative, early innovators have a higher probability to continue to innovate in the future, at least until the emergence of new technological paradigms makes the relevant knowledge and competencies obsolete. Shifting this argument from individual firms to clusters requires some additional assumptions about the mechanisms through which knowledge and competencies are replicated outside the boundaries of the original innovator(s). Spin-off processes from incumbent firms (or other organizations like universities) are certainly a plausible, but certainly not the only mechanism, which does not necessarily rely on externalities (see Chapter 3). Second, Klepper’s approach somehow raises the question about the direction of causation between innovation and clusters. Whereas most of the literature discussed so far focuses on the conditions that make a particular area conducive to innovation—that is on the idea that clusters promote innovation—the opposite nexus of causation might turn out to be at least equally important: it is an original innovation that creates clusters. Quite obviously, it remains necessary to specify the mechanisms that tend to keep subsequent innovations within the clusters rather than spreading outside. In this respect, the degree of overlapping between these two extreme approaches is very high and it might be difficult to meaningfully separate them. In addition, Klepper’s argument suggests that perhaps more emphasis should be attributed to specific characteristics of the firms and the other key agents active within a cluster. In the domain of cluster studies, relatively few contributions have emphasized the relationships between firms’ characteristics and the innovative outputs of clusters or regionally bounded areas (Beaudry and Breschi 2003; Harrison, Kelley, and Gant 1996). As suggested by Caniels and Romijn (2003: 1253), the regional agglomeration studies emphasize the favorable impact of geographical proximity on regional economic performance; but the firms that constitute those agglomerations largely remain black boxes. In contrast studies dealing with technological learning explain economic performance at firm level without systematically taking accounts of the effects of geographical proximity.

203


Similarly, in network studies, recent contributions have stressed that ‘the bulk of network research has been concerned with the consequences of networks’ (Borgatti and Foster 2003: 1000) and that ‘ . . . limited attention has been paid thus so far to how important non-structural features—such as the characteristics of the organizations that represent nodes in a network, geographic location, or the institutional underpinnings of the larger structure—alter the character of information flows’ (Owen-Smith and Powell 2004: 5). Finally, emphasis on reverse causation has the merit to recast the issue in an explicit dynamic way and therefore to reframe the initial question about endowments and agglomeration forces in quite a different perspective: clusters may well emerge as a sheer outcome of chance and develop through the working of a variety of self-reinforcing processes, irrespective of any initially favorable endowment. Indeed, one does not necessarily rely on chance alone in order to explain the genesis and the development of clusters. In many cases, particular sets of initial conditions make the onset of clusters more likely. For example, Bresnahan, Gambardella, and Saxenian (2001) suggest that emerging clusters tend to share some common features: existence of unexploited technological and market opportunities, highly educated skilled labor, firm- and marketbuilding capabilities by pioneering firms, connection to markets, and luck. Another recurrent factor is the availability and concentration of state-of the-art knowledge in key agents. Interestingly, variables like the presence of supporting institutions (e.g. VC) diffusion of particular social attitudes (e.g. entrepreneurship), appear to play a much lesser role in nascent clusters and if anything, they tend to develop later on as a product of the agents’ activities (Feldman 2001). Thus, in a dynamic perspective, considering these initial conditions as endowments may be deeply misleading, insofar as they are not given, but they are themselves the result of previous processes of construction of competencies and institutions. This interpretation resonates with all those—mainly appreciative— accounts which highlight how the development of clusters involved complex processes of construction of competencies, supporting institutions, organizational structures unfolding in time by heterogeneous agents who cannot perfectly understand the environment in which they act (Bresnahan, Gambardella, and Saxenian 2001; Feldman 2001; Kenney and Von Burg 2001). Furthermore, a now significant stream of theoretical literature has been indeed analyzing the nature of economies and diseconomies of agglomeration in a truly dynamic framework in which persistent spatiotemporal patterns are conceived as emerging out of direct interactions

204


among boundedly rational, heterogeneous economic agents (cf. Arthur 1994: Chapters 4 and 6; Cowan and Cowan 1998; David, Foray, and Dalle 1998). In explicitly dynamic settings, it becomes possible to appreciate: the complex trade-offs become more clear between purely random factors and more systematic, historical forces (or, put it differently, the issue of necessity vs. chance) underlying the emergence of spatially ordered structures (Bottazzi, Fagiolo, and Dosi 2002: 9).

The debate on the nature of the externalities that generate and sustain clusters provides—as it often happens in academia—perhaps too many possible hypotheses for explaining the phenomenon under consideration, and it is intrinsically difficult to satisfactorily discriminate among them. To some extent, this should come as not too big a surprise, when one considers the high degree of heterogeneity in the structure of clusters and in the nature of the relevant technologies. In a slightly different perspective, the concept of clusters bears a worrying resemblance with other stylized facts, which are found in industrial dynamics, like the persistence of skewed distribution of firms’ sizes. The facts are often what statisticians call unconditional objects (Brock 1999), that is they can be generated by an enormous variety of alternative processes so that empirical testing becomes exceedingly difficult. Under these circumstances, it becomes necessary to impose much stronger restrictions both on the types of phenomena that are to be explained and on the alternative theories. Thus, a theory should be able to account for a larger set of phenomena at the same time and these phenomena should be specified in much closer detail. A final point deserves mention. The literature on clusters has certainly provided strong evidence that innovation often tends to be spatially localized and an extremely rich variety of explanations for this phenomenon. Yet less attention appears to have been attributed to the exploration of the null hypotheses such as are clusters a necessary characteristic of innovation processes? And are clusters unequivocally beneficial to innovation? Certainly more work remains to be done.

The Ingredients of Innovative Clusters Almost all of the studies on biotechnology concentrate their attention on a set of ingredients that are usually thought to be important constituents of clusters: the scientific base; entrepreneurship, VC, and a favorable

205


intellectual property (IP) regime; linkages with large firms and other industries; institutions, policies, and other infrastructures that support and promote entrepreneurship. Each will be discussed in turn. A strong academic base seems to be a recurrent ingredient.3 However, other types of research institutions are often mentioned to play a significant role: national laboratories, hospitals and medical schools, and laboratories of large corporations. Indeed, there is significant variation in this respect. In Europe—especially Continental Europe—the role of research organizations appears to be relatively more prominent as public laboratories, like the Max Planck Gesellschaft or the Institute Pasteur, rather than universities have been traditionally the main performers of basic research. Rather than the specific nature of the relevant research organizations, what might turn out to be more important is the absolute level of excellence in research, the degree of variety in research and the interactions among organizations. Zucker and Darby (1995) show that academic research as such explains little of new firms’ formation and their performance, but it is rather the presence of star scientists that leads to start-ups and agglomeration. Further, variety and diversification of scientific research within the cluster appear to play a significant role (Audretsch and Feldman 1996; Owen-Smith et al. 2002). In contrast, laggard European clusters appear to be much more specialized and less integrated, both along the horizontal and along the vertical dimensions. Along the same lines, links between science base and companies are supplemented by links between industrial sectors (Prevezer 2001; Swann and Prevezer 1996). In general, firms with complementary rather than similar specializations are attracted to each other and tend to enter a cluster together. Partially different results are obtained however by Aharonson, Baum, and Feldman (2004a, 2004b), who provide evidence that firms located in clusters were strong in their technological specialization: entrepreneurs appear to seek out firms that are similar to themselves. A strong scientific base is a fundamental ingredient of clustering, it is also clear that it is not a sufficient condition. Examples abound of cases of areas where a strong scientific base has failed to spawn dynamic clusters (see Feldman 1994; Orsenigo 2001). The willingness and ability to exploit such knowledge for economic purposes have been consistently identified as crucial ingredients of the clustering process. As a general proposition, this assumption is dangerously close to a truism: if a cluster is defined in terms of agglomeration of economic activities, then it is obvious that economic considerations are crucial for the explanation. The proposition becomes much more interesting and controversial as soon as the specific

206


factors leading to and/or allowing for the decision to commercially exploit knowledge are examined. Entrepreneurship is again quite obviously a fundamental characteristic of clusters. In many instances, the propensity to entrepreneurship is considered as a psychological and/or a social characteristic. Especially in this second interpretation, the explanation of differentials in entrepreneurial attitudes across space plays a significant role in the debate. Clusters are indeed often defined and explained as geographic areas where entrepreneurial attitudes are particularly developed, for various historic reasons, and continuously reproduced by the cluster itself (Becattini 1990; Saxenian 1994). Without denying the crucial relevance of these factors, a discussion of entrepreneurship falls beyond the scope of this chapter. It might be just worthwhile mentioning—with specific reference to the case of biotechnology—that attitudes toward entrepreneurship cannot account as such for the underdevelopment of biotechnology in Italy, a country which exhibits very high rates of new firm formation in sectors other than biotechnology. In a more subtle interpretation, entrepreneurship is not considered simply as a psychological and social attitude, but as a specific form of organization of innovative activities, typically contrasted with the traditional picture of vertically and horizontally integrated large corporations. This is indeed an important argument and it plays undoubtedly a crucial role in the case of biotechnology, especially as it concerns the differences between the performances of the USA and Europe. In any case, it is perhaps more fruitful to concentrate attention first on the factors that might facilitate or hinder entrepreneurship and the development of decentralized, network-like forms of organization of innovative activities, assuming for the sake of discussion that individual and social propensities are evenly distributed across space. Rules governing academic involvement in commercial activities, IPRs and VC are in fact commonly referred to as essential factors that might account for the differential performance of the USA—the distinct character of the American way of innovation. This system is typically described as being based on three interconnected pillars: university–industry interactions, a strong IPRs regime (which favors the commercialization of scientific research), and VC. The key role of scientific knowledge for technological innovation has indeed manifested itself in an unprecedented intensification of both industry–university ties and in the direct involvement of academic institutions and scientists in commercial activities. Both phenomena are

207


certainly not new in the USA (Mowery et al. 2001; Rosenberg and Nelson 1994). However, since the mid-1970s and over the whole evolution of the biotechnology industry, the drive toward an increasing commercialization of the results of research accelerated dramatically. Universities’ patenting and licensing and the creation of academic spin-offs has become a distinct and crucial phenomenon of the American academic system. The emergence of the entrepreneurial university and the specific forms in the USA are strictly linked to some basic characteristics of the US academic system. American universities had traditionally been highly responsive to the needs of the local communities and industries. Also the organization of research and teaching had characteristics that facilitated both the production of high-quality research and high degrees of mobility between academia and the commercial world. Specifically, departments have long been the main organizational entities as opposed to the European institutes, dominated by a single professor, far less interdisciplinary in nature and with feudal-like career paths (Ben-David 1977; Clark 1995). Moreover, in the US high degrees of integration between teaching and learning have been achieved through the sharp separation between undergraduate and postgraduate levels. The creation of researchoriented postgraduate studies entailed, in fact, a number of important consequences. In particular, postgraduate students are typically exposed and trained to the practice of scientific research within research teams composed of students and professors within departmental organizations. This arrangement does not only tend to free resources for scientific research, but provides also a fundamental experience to participating in and managing relatively complex organizations. In other words, it constitutes an essential source for the development of organizational capabilities. Moreover, the career of young research scientists after graduate studies has—under various perspectives—entrepreneurial characteristics. For instance, post-docs have to raise funds for their own research in a highly competitive environment, where performance is judged on the basis of a track record and the ability to set an independent research agenda (Gittelman 2000). Finally, graduate students joining the industrial world after the completion of their studies constitute an essential source of skilled demand for academic research. In Continental Europe, the integration of teaching with research has progressed far less than in the USA (and to some extent than in the UK). Clearly, enormous differences in education systems, especially on the higher education level, exist across Continental European countries and they certainly should not be overlooked. For example, as it was mentioned previously, in France, universities have

208


never been the main center of both scientific research—which has been essentially conducted within the national laboratories and coordinated by the CNRS (National Center for Scientific Research)—and the education of the elites, monopolized by the system of the grandes ećoles. In Germany, the institute—dominated by an individual professor—has been the main organizational unit coordinating teaching and research. Despite these enormous differences, the structure of the academic systems of many European countries shares some important common features, as compared to the Anglo-Saxon systems. Doctorate students are a relatively recent innovation in many Continental European countries and they remain far less professionally orientated than in the USA. Departmental structures are also relatively new and in many cases institutes continue to be a very important organizational entity. In general, research has tended to be far more removed from teaching than in the USA. And in fact, in many Continental European countries, research has been to a large extent separated from universities and concentrated in specialized institutions (Ben-David 1977; Clark 1995). In Europe a model has been emerging based on high degrees of division of labor and specialization between teaching and research institutions, whereas in the USA the dominant model has been an integrated one (Bruno and Orsenigo 2003). It is possible to speculate that this separation might have had negative effects on both the quality of research and the ability of academic institutions to interact with industry. Integration of research and teaching and collaboration with industry has been relatively more developed and frequent in the case of engineering schools (the Continental European polytechnics) and in some selected disciplines in particular countries (chemistry in Germany). In the 1960s–70s, however, with the development of mass academic education, the scientific revolutions linked mainly to microelectronics and molecular biology (developed mainly in the USA) and the crisis of the traditional industries mainly connected with the polytechnics, in many cases industry–university interaction has further weakened. To remedy this gap, a reproposition of the specialized model has tended to spread in more recent years for the management of the interactions between research and industry and technology transfer. Differently from the US case, where universities have gradually extended their functions (an integrated model centered on universities), one observes in Continental Europe the development of various types of specialized institutions for technology transfer, who act as intermediaries between research and industry (an institutional specialization model).

209


The coupling between scientific, organizational, and entrepreneurial capabilities thus constitutes an essential precondition for subsequent developments in industry–university relations. However, it is also important to notice that such developments are to some extent to be considered as part of a much more general tendency toward the diffusion of an increasingly favorable attitude toward the establishment and enforcement of strong IPRs. The establishment of clearly defined property rights played indeed an important role in making possible the explosion of new firm foundings in the USA, since the new firms, by definition, had few complementary assets that would have enabled them to appropriate returns from the new science in the absence of strong patent rights (Teece 1986). In the early years of biotechnology considerable confusion surrounded the conditions under which patents could be obtained. In the first place, research in genetic engineering was on the borderline between basic and applied science. Much of it was conducted in universities or otherwise publicly funded, and the degree to which it was appropriate to patent the results of such research became almost immediately the subject of bitter debate. Similarly a growing tension emerged between publishing research results versus patenting them. While the norms of the scientific community and the search for professional recognition had long stressed rapid publication, patent laws prohibited the granting of a patent to an already published discovery (Kenney 1986; Merton 1973). In the second place the law surrounding the possibility of patenting life-formats and procedures relating to the modification of life-forms was not defined. This issue involved a variety of problems, but it essentially boiled down first to the question of whether living things could be patented at all and second to the scope of the claims that could be granted to such a patent (Mazzoleni and Nelson 1998; Merges and Nelson 1994). In fact, these trends were partly spurred by a growing concern about how to exploit more efficiently academic research and by the need to put some order in the system that governed the conditions at which universities could obtain patents—and therefore income—on the results of publicly funded research. The Bayh–Dole Act in 1980 sanctioned these attitudes, by greatly facilitating university patenting and licensing. Parallel to Bayh–Dole, a series of judicial and Congress decisions further strengthened the appropriability regime of the emerging sectoral system. In 1980, the US Supreme Court ruled in favor of granting patent protection to living things (Diamond vs. Chakrabarty), and in the same year the second reformulation of the Cohen and Boyer patent for the DNA process

210


was approved. In the subsequent years, a number of patents were granted establishing the right for very broad claims (Merges and Nelson 1994). Finally, a one-year grace period was introduced for filing a patent after the publication of the invention. The third pillar of this emerging system was, of course, VC. Once again, VC was a long-standing institution in the American financial and innovative system. It was already active—in various forms—ever since the 1920s (or even before) and emerged as a vibrant industry with the electronic revolution in the 1960s. Most likely, the existence of a vibrant VC industry which developed with the ICT revolution was a factor that favored in different ways the emergence of the Bay Area and Boston as leading biotechnology clusters (Audretsch 2001; Niosi and Bas 2001; Prevezer 2001). However, it has to be emphasized that these factors—industry–university relations, IPRs, and VC—were not simply preexisting at the onset of the biotechnology industry, but they coevolved over time. As Mowery et al. (2001) have shown the emergence of the industry–university complex (Kenney 1986) and the entrepreneurial university predates Bayh–Dole. Similarly, other studies (e.g. Kortum and Lerner 1998) concur in suggesting that the growth of patenting activities in the USA over the past 10–20 years does not seem to have been spurred so much by a tighter appropriability regime, but rather on the the rise of the two main technological revolutions of the second half of the century, microelectronics and especially biotechnology. Thus, the relationship between patent protection, innovation, and commercialization of academic research appears to be much more complex than it would appear at first sight. New technological opportunities have provided new opportunities for invention and for its economic exploitation as well as improvements in the productivity of R&D. Stronger and wider IPRs have certainly incentivized and facilitated the commercialization of academic research, especially in the form of academic spin-offs, but the extent to which they have also directly spurred invention remains more doubtful. To the extent that the development of biotechnology rests primarily on the absolute excellence of scientific research, the influence of IPRs on innovative activities in biotechnology appears mediated by this latter factor. Moreover, the IPR regime cannot account for geographic differentials in performance within countries, both as it concerns the USA and Europe, being a nationwide, rather than a cluster-specific factor. Even in Europe, no simple and direct relationship can be found between the development of biotechnology—at the nation-state level, let alone at the clusters

211


level—and IPRs regimes. If anything, the explanation might then turn on the observation of more aggressive attitudes and better organizational structures of specific universities toward IPRs and technology transfer in different clusters. The case-studies literature does indeed provide evidence of this (Audretsch 2001; Niosi and Bas 2001). A somewhat different, but germane interpretation is provided by Feldman (2001) in her analysis of the case of the Capitol region, where entrepreneurship was unleashed by, on the one hand, exogenous events like the downsizing in federal employment and increasing reliance on outsourcing and, on the other hand, by a mix of government policies that created demand for ICTandbiotechnology services, as incentives and pressures for the appropriation and commercialization of public research, despite the feeble commitment of some important research institutions (e.g. Johns Hopkins) toward academic entrepreneurship. Perhaps more interestingly, it is important to stress again that changes in IPRs regimes and greater entrepreneurial attitudes on the part of academics have themselves been spurred by the initiative of individuals, firms, industry organizations, etc. in knowledge-rich environments which offered the opportunities for innovation and new firms’ creation. Similarly, while there is ample evidence that local availability of VC facilitates the development of new biotechnology firms, much less evidence is available that this was a necessary precondition for the take-off of the industry. In many cases, in Europe new firms have been founded resorting to capital located outside the cluster and even outside the country and clusters have developed attracting VC rather than finding it already present within the location (Allansdottir et al. 2001; Orsenigo 2001).

Local Demand, Institutions, Infrastructure, and Policies Analogous remarks can be suggested as it concerns the other factors that are usually claimed to be important ingredients of a cluster. While there is little question that these factors are usually present in a successful cluster, there is much less evidence about their specific nature and that they are a necessary precondition for the development of a cluster. Thus, for example, Niosi and Bas (2001) find that biotechnology clusters in Canada are located in large metropolitan areas, comprising high-level research conducted in prestigious institutions, an ample pool of VC, large and research-oriented hospitals, etc. Aharonson, Baum, and Feldman (2004) also support the finding that urbanization economies are an important factor in the Canadian biotechnology industry. Similar results

212


are found for some European clusters (e.g. the Stockholm area), although in general clusters appear to form around centers of academic excellence rather than cities as such (Allansdottir et al. 2001; Pammolli and Riccaboni 2001). Similarly, the local presence of large corporations does not systematically emerge as a fundamental ingredient of clusters. Large companies are frequently located outside the cluster, both in the USA and in Europe, although linkages between local actors and big corporations are invariably a crucial component of the successful development of a biotechnology industry. Moreover, in some European cases, there are instances where new firms are created as spin-offs of large firms undergoing processes of restructuring, typically after episodes of mergers and acquisitions (Orsenigo 2001). As already mentioned, the existence of related industries and the relations between the various segments of the biotechnology industry appears in some studies as an important determinant of the (local) development of biotechnology (Prevezer 2001; Swann and Prevezer 1996). However, such relations are quite intricate and appear to vary significantly across different clusters and types of firms (Lemarie`, Mangematin, and Torre 2001; Mangematin et al. 2003). Finally, the degree of variability is even higher in terms of policies. Reviewing the literature in this case is tremendously difficult. Two main general (and generic) conclusions in this respect might simply be that: especially but not exclusively in the American cases, early successful clusters developed almost spontaneously without any direct intervention, although even in these instances, the initiative of particular (and diverse) institutions plays an important role in the cluster; especially in Europe, one observes a bewildering variety of policies at the nation and regional level to foster the development of biotechnology. A fair assessment would seem to be that it is almost impossible to draw any kind of generalization, both as it concerns experiences of success and failure. The development of successful clusters has been achieved through wildly different approaches: surely, there is no single way, let alone recipe, to success. On the other hand, failure is pervasive, irrespective of the policies that have been adopted.

Mechanisms of Agglomeration, Networks, and Network Dynamics Clusters are not simply born with all the necessary ingredients in place. Thus, it becomes important to examine the nature of the underlying

213


processes. The previous discussion on ingredients seems to suggest that classic Marshallian externalities play an important role in generating agglomeration. However, almost all the case studies and econometric emphasize that proximity to universities and/or other sources of knowledge is important and the usual explanation is that it makes the circulation of tacit knowledge easier. Similar arguments are provided as it concerns VC: detailed knowledge of the opportunities offered by the knowledge-creating institutions, personal acquaintance with the scientists, continuous monitoring of companies are fundamental aspects of VC and this knowledge is much easier to be acquired at the local level. Second, the relevance of soft institutions might also have been overestimated. Knowledge flows appear to be channelled through market transactions and interorganizational rules (Arora et al. 2004) than the classic notion of the industrial district. In between these two extremes, knowledge flows appear to be structured in a fundamental way by the social and technical rules of interaction that govern scientific and technological research, as will be discussed at some more length in the next paragraph. Third, this interpretation is consonant with—albeit not identical to— the model proposed by Klepper (2002) for the automobile industry. According to this view, as discussed in Section 2, clustering results from processes of spin-off—as distinct from spillovers—from knowledge-rich organizations: companies in the case of automobiles, universities in the case of biotechnology. These organizations perform as incubators of capabilities which are then exploited in the form of new firms and more generally of the formation of a pool of highly skilled labor force. Going back to the discussion in Section 2, it would appear that increasing returns, in addition to externalities, are fundamental movers of the processes of agglomeration. The study by Romanelli and Feldman in Chapter 5 constitutes in this perspective a first and important attempt to disentangle empirically the mechanisms through which this type of increasing returns work, at least in the case of biotechnology. In particular, they show that around 75 percent of the human biotherapeutics firms founded in the USA over the period of 1976–2002 identified in their analysis have a local origin, that is they spun off from institutions, especially universities, but also other companies, located in the same geographic area. Moreover, successful clusters continue to exhibit high rates of internal firms’ formation, whereas weaker clusters are characterized by both lower domestic profilicness and higher propensity to migrate. Attraction of entrepreneurship from outside the clusters appears also to be a significant phenomenon. In other words, two interlinked and self-reinforcing processes seem

214


to drive the evolution of clusters. First, a process of spin-off from local institutions and organizations—similar to Klepper’s account—originates and sustains the cluster. Second, on these bases, immigration of new entrepreneurs and firms having an origin from outside the cluster is set in motion which strengthens and overlap with the former. Fourth, the relevance of the local knowledge base is likely to change over time, as clusters evolve and mature; and across different typologies of firms. Moreover, local knowledge has a different impact on the entry and the growth of firms. As already mentioned, local sources of knowledge appear to be fundamental in the early stages of the development of a cluster and for new, highly specialized firms. Access and ability to use (and integrate) external knowledge becomes increasingly important for growth and diversification (Lemarie`, Mangematin, and Torre 2001; Mangematin et al. 2003; Prevezer 2001; Swann and Prevezer 1996). Fifth, network analysis provides further insights into this discussion, particularly as it concerns the role of the local knowledge base, the mechanisms that regulate the flows of knowledge and their dynamics over time. One finds in the literature widely different interpretations of the nature, motivations, structure and functions of these networks, ranging from more sociologically oriented approaches to economic explanations based on (various mixes of) alternative theoretical backgrounds, for example transaction costs, contract theories, game theory, and competence-based accounts of firms’ organization. According to an influential interpretation, collaborations represent a new form of organization of innovative activities, which are emerging in response to the increasingly codified and abstract nature of the knowledge bases on which innovations draw (Arora and Gambardella 1994; Gambardella 1995). To be sure, substantial market failures exist in the exchange of information. However, the abstract and codified nature of science, coupled with the establishment of property rights, makes it possible, in principle, to separate the innovative process in different vertical stages. Different types of institutions tend to specialize in the stage of the innovative process in which they are more efficient: universities in the first stage, small firms in the second, big-established firms in the third. In this view, then, a network of ties between these actors can provide the necessary coordination of the innovative process in a division of innovative labor. In this perspective, it has also been suggested that the locus of innovation (and the proper unit of analysis) is no longer a firm, but the network itself (see Powell, Koput, and Smith-Doerr 1996). In this case, the direction of causation is reversed: it is the structure of the network and the position of agents within it that fundamentally

215


determine agents’ access to relevant sources of scientific and technological knowledge and therefore innovative activities and performances (see also Walker, Kogut, and Shan 1997). Finally, these partial and not entirely congruent results might also begin to shed some light on the dynamic mechanisms that determine the advantages of agglomeration. Once again, rather than relying on spillovers as such, clusterization might be interpreted as the outcome of the intrinsic difficulties involved in the reproduction of knowledge outside their specific original contexts. Thus, spin-offs and start-ups tend to locate close to their parents and relatives also because what they know derives from region-specific practices, ways of doing things, etc., and their replication occurs with higher probability and ease within the same context. Clearly, much more research is needed here (e.g. looking at differential patterns of growth and performances of local vs. immigrated firms).

Conclusions The evidence strongly supports the view that there are indeed advantages to agglomeration and powerful forces leading to the development of clusters. Yet the strength of these clustering tendencies differs according to the indicators that are used. Moreover, what exactly those advantages and processes might be remains much less clear. They appear to be related mainly to the stock of accumulated scientific and R&D capabilities within a geographic area. Adequate incentives for the use of such knowledge for commercial purposes are also obviously important, although their nature and effects may differ substantially across space and time. Moreover, it is hard and probably misleading to distinguish between the ingredients and the processes that underlie cluster formation and development. The processes do depend on and are shaped by initial conditions. Thus, different agglomeration forces may play a role under alternative initial circumstances. Indeed, the empirical evidence does not provide clear and homogeneous indications in this respect. If anything, the evidence seems to support a picture whereby the spatial concentration of innovative activities derives mainly from processes of spin-off from highly capable universities and research centers. However, classic Marshallian externalities and urbanization economies play a significant role, but coupled with (and perhaps triggered by) processes driven by the increasing returns intrinsic in the generation of knowledge. In any case, both case studies and econometric evidence tell different stories, from which is hard

216


to draw strong generalizations. Similarly, little evidence is available on the relative role of the processes of firms’ formation originating within the cluster and processes of attraction of entrepreneurs initially located outside the cluster. In the same vein, the evidence concerning localized knowledge spillovers is mixed. On the one hand, a considerable body of results—both econometric studies and network studies—strongly suggest that knowledge is not simply in the air within a cluster, but knowledge flows are heavily structured by a variety of economic and social factors which in many instances do not have any clear geographic connotation. Yet, on the other hand, there are also overwhelming suggestions that indeed knowledge is spatially sticky and intense knowledge flows are an intrinsic characteristic of clusters. Most of this evidence is provided by case studies. Thus, the question might be asked if such knowledge flows are and can be captured by quantitative, econometric studies. But rather than simply dismiss the qualitative evidence, more effort should be devoted to devise better measures and indicators. The evidence forcefully points to the observation that—as much as agglomeration forces are influenced by structural initial conditions— processes are the essence of what clusters are made of. The factors that lead to the genesis of a cluster are different from those that later sustain the cluster itself. The factors that influence firms’ entry in a cluster are at least partly different from those which promote firms’ growth. Again, it is hard to identify invariant processes across clusters. Yet most of the case studies and of the dynamic network studies (but also some econometric results) suggest that clustering is the outcome of processes of construction and coevolution of the conditions that allow clusters to exist rather than the automatic effect of specific preconditions and agglomeration factors. In this sense, innovation generates clusters at least as clusters create innovation.

Notes 1. A significant fraction of the theoretical literature on clusters explains different spatial agglomeration patterns (from concentration of economic activities in few locations to hierarchical structures) as the solution of static trade-offs between agglomeration and dispersion forces, particularly combinations of static externalities, transport costs, and economies of scale (Fujita 1988, 1989; Henderson 1974; Papageorgiou and Smith 1983). The New Economic Geography

217

Clusters and Clustering focuses instead on various forms of increasing returns to scale (or indivisibilities) at the level of individual agents as the main mechanisms leading to agglomeration and to its persistence (Fujita, Krugman, and Venables 1999; Fujita and Thisse 1996; Krugman 1996, 1991a, 1991b, 1991c, 1993; Krugman and Venables 1995a, 1995b, 1996; Ottaviano and Puga 1998; Ottaviano and Thisse 1998; Puga and Venables 1996; Venables 1996). However, and perhaps not too surprizingly, given that innovation is the subject of the analysis of this stream of research, the role of knowledge spillovers has attracted the attention of scholars. The localized nature of such spillovers is usually associated to the tacit nature of knowledge. Thus, communication occurs more easily through face-to-face contacts and interactions. Different types of methodologies include estimation of knowledge production functions as in Jaffe (1989), Acs, Audretsch, and Feldman (1992, 1994), Audretsch and Feldman (1996), Feldman and Audretsch (1999), and Feldman and Florida (1994). Alternatively, patent citations have been used to track direct knowledge flows from academic research into corporate R&D (Almeida and Kogut 1999; Jaffe, Trajtenberg, and Henderson 1993). Moreover, an immense set of empirical case studies and narratives confirm that indeed important localization effects exist in innovative activities. 2. For some preliminary discussions and empirical evidence, see Bottazzi, Dosi, and Fagiolo (2002). 3. There is little question that all biotechnology clusters are found in locations where there exists a strong concentration of scientific capabilities and institutions. Both case studies and econometric evidence strongly support this view. However, beyond this observation, several further questions remain open (Audretsch and Stephan 1996; Prevezer 1997, 2001, 2003; Swann and Prevezer 1996; Swann, Prevezer, and Stout 1998; Zucker, Darby, and Brewer 1998). In particular, it is less clear what kind of institutions are important; what kind of scientific research is really important; and what exactly is the shape of the relation between the strength of the knowledge base and the performance of the cluster.

218

11 Mors tua, vita mea? The Rise and Fall of Innovative Industrial Clusters Mario A. Maggioni

I will tell the story as I go along of small cities no less than of great. Most of those which were great once are small today; and those which in my own lifetime have growth to greatness were small enough in the old days Herodotus (440 bc), The History, quoted in J. Jacobs (1969)

History demonstrates that when a major technological innovation is created, new clusters appear that become the locus of the new activity. These new clusters often develop and grow at the expenses of older, more mature historic sites. Of course, as time marches on, new industries and their locations, as a result of interactions between agglomeration economies and diseconomies on the one hand, coupled with incremental versus radical innovation on the other, may lose their advantage. Thus technological and regional dynamics go hand in hand and are mutually determined in a complex web of circular cumulative causation with both positive and negative feedback. The most frequently cited forces that underpin and propel cluster have been analyzed and discussed at length in the previous chapters. The empirical analyses are frequently based on case studies due to the lack of systematic data on clusters. The dominant explanations refer to spin-offs and imitation, anchor tenants, leader–supplier relationships, knowledge and information diffusion, institutional processes and social legitimacy, signaling, and, least but not last, agglomeration forces.1 This chapter will focus on the last mechanisms in order to model the trade-off between

219

The Rise and Fall of Industrial Clusters

forces that serve to foster spatial concentration as compared to forces leading to more geographically fragmented production structures. Building on Brezis and Krugman (1997), this chapter makes use of formal analysis in order to illustrate the different possible evolutions of a dynamic system composed of two clusters. It is shown how different clusters’ evolution, often told as separated stories, are part of a wider picture in which technological and spatial interactions between emerging and declining clusters play a decisive role. The process can be easily described within a population ecology theoretical framework where both technological dynamics within the cluster and spatial interactions between clusters determine the life cycle of a cluster. Focus is on the logic of this evolution and the policy responses. All technical details are provided in the Appendix. A final section draws some policy suggestions for public authorities and regional planners dealing with the development of an innovative cluster.

Development of an Industrial Cluster: Locational Benefits Versus Costs Firms decide to settle in a cluster on the basis of the expected profitability of being located there. This profitability depends on net locational benefits—obtained as the difference between gross locational benefits and costs—which, in turn, are based on both observable and unobservable elements. Both locational costs and gross benefits are nonmonotonic functions of the local industrial mass. As far as costs are concerned, they are U-shaped due to the classical combination of a decreasing average fixed cost schedule and an increasing variable fixed cost schedule; as far as gross benefits are concerned they have an inverted U-shape due to the interaction of agglomeration economies and congestion phenomena over a limited amount of land and infrastructures.2 Net locational benefits are described therefore by an inverted U-shape function of the number of located firms that is often quoted as the indirect microeconomic foundation of an S-shaped development path of the cluster. For simplicity it can be assumed that in a world of limited information regarding local costs and revenues available to the outsiders, profitability expectations for any particular location is based solely on the number of firms already located there (the number of previous locations being the only observable variable). Moreover, think of locational gross benefits for a firm locating in a given cluster as composed of geographic and

220


agglomeration benefits (Maggioni 2002a).3 Geographic benefits thus depend on the intrinsic features of the site (such as the quality of local supply of capital and labor, efficiency and quality of the local network of specialized suppliers and business service firms, and the quality of urban and industrial infrastructures). Agglomeration benefits diminish after they have reached a certain level, that is agglomeration is a concave non-monotonic function of the number of firms already established in cluster (q). Hence, as the number of firms located in the cluster increases, gross benefits (i.e. the sum of geographic and agglomeration benefits) first increase because of agglomeration economies (due to productive specialization, knowledge spillover, access to specialized labor, reduced transport and transaction costs, etc.), then decrease when the negative effects of increased congestion exceeds the agglomeration economies. Locational costs are constructed in a similar way. They also include two components: geographic costs (locally prevailing wage, rental cost, tax rate, etc.), and agglomeration costs (which are assumed to be a convex nonmonotonic function of the number of incumbents). This means that, as the number of firms located in the cluster increases, locational costs initially decrease until some optimal number of users for a given set of urban, industrial and environmental infrastructures and resources is reached. Then they increase due to the competition, between a larger number of firms, for a limited pool of local inputs (i.e. capital, labor, business services, land, and public infrastructures) which raises their prices.4 Net locational benefits are equal to the difference between gross locational benefits (a concave function) and locational costs (a convex function); therefore each marginal firm, which enters the cluster, increases the net average profitability of locating in the cluster only up to a threshold. After that point, any new entrant lowers the average net benefits available to each resident firm and new entrant.5 This formulation is concomitant to some general results obtained in the industrial location and urban economics literatures, which demonstrated the existence of an optimal dimension of a spatial agglomeration of firms and/or households.6 These relationships are illustrated in Figure 11.1. A first observation is that several optimal sizes of the cluster may exist. Second, the maximum dimension of the cluster (K) is endogenously determined by the structure of locational benefits and costs functions.7 To simplify, we assume that location by firms outside the region are exposed to exactly the same benefits and costs from colocation, implying that the analysis can be pursued by studying the behavior of a representative

221


Agglomeration benefits and costs

Average net benefits Marginal net benefits Average costs Marginal costs Average gross benefits Marginal gross benefits

O A

B C B' X Number of incumbents n(t)

K

Figure 11.1. Agglomeration costs and benefits for incumbents, with critical sizes of a cluster Source: Maggioni (2002a).

firm.8 At point A in Figure 11.1 the minimal sustainable dimension of the cluster is attained (i.e. where agglomeration net benefits start to be positive and a spontaneous entry process develops). Prior to A no firm will spontaneously enter the cluster because average costs are higher than average benefits. A can be called the critical mass of the cluster. A can be reached only by a group of cocoordinated firms (n) entering together, or by direct intervention of a public authority aimed at subsidizing entries until A is reached. B is the dimension where average agglomeration costs are minimized, while B’ represents the cluster dimension that maximizes gross average agglomeration benefits. B and B’ underline the importance of analyzing both costs and benefits of location to avoid harmful misrepresentation of the economic reality, derived from the a-critical application of some early contributions of location theory.9 Obviously, it could also be the case that B’ < B. At C maximum average net benefits are obtained. Up to C every new entrant increases (by its very entry) the average benefits of all incumbents; after C the average benefits decrease. C is therefore the optimal cluster size. However, it is neither the social efficient outcome (given that marginal benefits are still greater than marginal costs) nor the maximum possible dimension (average benefits are still positive). At C, several firms outside the cluster might still want to enter, while firms already in the cluster would like to deter further entries. Thus, at this point a conflict of interests appears between incumbents, outsiders, and public authorities, since the maximization of social welfare differs from private profit maximization. From a social point of view the optimal solution is at point X where

222


marginal costs equal marginal benefits, hence, the total benefits (number of firms times per capita benefits) are maximized. However, as the average benefits at X are still positive, some outsiders would still like to enter. Such entries would reduce the total amount of benefits available to incumbents. K is the maximum dimension of the region (in terms of economic mass) since gross locational benefits equal locational cost, that is net locational benefits are zero. After this point, average benefits are negative and therefore there are no incentives to enter. However, new entries are still possible but these would be at the expense of some incumbents who would be driven out of the cluster.10

The Life Cycle of Clusters: An Ecological Model The simplest growth model for an industrial cluster—which stresses the relevance of firms’ spatial interactions—can be expressed in the following format: ‘The rate of growth of the industrial mass equals the product of the individual firm’s contribution to the regional population’s growth and the number of firms already in the cluster’ (Maggioni 1993).11 If only positive feedbacks mechanism (such as spin-off, agglomeration economies, and knowledge spillovers) are taken into account (and these are assumed to be constant), then each individual firm’s contribution to the level of average locational benefits and, consequently, to the growth of cluster, would be equal to a constant. In this case cluster industrial growth would follow an explosive exponential path.12 On the other hand, if negative feedbacks (such as congestion and competition effects) are included, then some modifications to this simple model are required to allow for some density dependent factors to progressively depress the level of locational benefits and to slow down the process of industrial growth of the cluster as its size approaches the ceiling level K. In this case, due to the counteracting roles played by r and K, the development of a cluster (i.e. the number of firms located in the cluster in each moment of time) follows an S-shaped curve. Both K and r play a major role in shaping a logistic growth path: the greater r the steeper the S-shaped curve, the larger K the higher the ceiling level of the function (and the equilibrium size of the cluster). In the ecological literature r, the cluster’s incipient (or intrinsic) growth rate is calculated as the difference between the birth and mortality rates of a population. This observation can be translated into the economic framework when net entry—and consequently the intrinsic rate of cluster growth—is calculated as the difference between total entries and exits.

223


The same value can therefore correspond to very different situations: steady-growing clusters where few new firms enter and no one exits, and a perturbed cluster where a high birth rate is almost compensated for by a high death rate. r is a composite index that describes the cluster growth potential and the probability that firms, once entered, survive in the cluster. A region’s industrial carrying capacity is defined by K: the maximum number of profitable firms the cluster can sustain (where we disregard interregional interactions). It depends upon two factors: first, the finite quantity of geographic benefits (which is related to the limited availability of local resources such as labor, capital, land, intermediate inputs, and infrastructures); and second, the decreasing part of the agglomeration benefits function (which depends on competition, congestion, and lobbying of incumbents). Therefore in the long run K may change as a result of the inflow of additional skilled workers, the provision of new advanced public infrastructure, and the diffusion of (technical, organizational) innovations. For a given cluster and a given population of outsider firms, therefore, we can postulate that the carrying capacity of a cluster act as an upper limit to growth (K#N). The number of entries therefore depends both on the actual number of potential entrants—the number of outsider firms that can enter and still make profits—and on the number of firms already located there n(t). Consequently, at any period t, K and n(t) determine the level of average locational net benefits available to incumbent firms in each period. From this basic structure it follows that the number of firms in the cluster directly generates (through agglomeration dynamics) the level of locational benefits: since the entry rate is assumed to be proportional to the level of locational benefits, it also indirectly determines the location of new firms into the cluster. This simple model can be used to empirically estimate key parameters of the location path of different clusters.13 The S-shaped pattern depicted in Figure 11.2 (showing the growth of nq (t) as time passes) can be derived from several alternative relationships existing between the process of firms formation in the cluster (or entry) n_ q and the stock of firms already located (operating) in the cluster nq (t).14 In this chapter we will focus the attention on two alternative versions of the original logistic equation; the quadratic and the cubic. These models are shown in Figures 11.3 and 11.4 and contain two crucial dimensions of industrial clusters. First, the minimum dimension for a self-sustaining cluster (or critical mass, A) and, second, the maximum number of profitable firms the cluster can sustain in isolation (the carrying capacity, K). These models are useful in describing the evolution of each cluster in

224


Number of incumbents (n)

Maximum dimension (K ) Maturity

Golden age Birth/take-off

Time

Figure 11.2. The development of an industrial cluster, in isolation Source: Maggioni (2002a).

Cluster’s growth rate

Firms’ entry Cluster’s growth n

A

Firms’ existing stock Cluster’s size (n)

K

Firms’ existing stock

Figure 11.3. The development of the old industrial cluster, stock–flows relation Source: adapted from Maggioni 2002b.

isolation.15 The next section is devoted to the analysis of the interactions between an already developed cluster and a newly emerging one, based on the rise of a new technology.

225


Cluster’s growth rate

Firms’ entry Cluster’s growth n

A

Firms’ existing stock Cluster’s size (n)

K

Firms’ existing stock

Figure 11.4. The development of the new industrial cluster, stock–flows relation

New Technology and the Life Cycle of Clusters Let us assume, as in Section 2, that the development of an established cluster in isolation (cluster i) is described by Figure 11.3. Cluster i is specialized in the production of a well-known product and uses an established technology. Now—basically following Brezis and Krugman (1997)— assume that a new technology (t) is introduced. The new technology may represent a new way to produce the same product (process innovation) or a new kind of the same product (product innovation). The critical hypotheses on the relationships between the two technologies are the following: . Each technology follows a learning curve so that productivity is an increasing function of the cumulative experience (i.e. the sum of all previous output) within the cluster. . Past experience with the old technology is irrelevant for the new technology (i.e. cumulative output produced by using the old technology has no effect on the new technology’s learning curve). . The new technology is potentially superior (i.e. for a given amount of cumulative output the new technology is more productive than the older one or, in other words, learning effects are greater for the new technology). . Despite a potential advantage, the new technology is initially inferior to the old in an established cluster, given that no cumulative output exists for the new technology.

226


. While in general inferior to the old technology, the new technology is superior in a newly established cluster (with no cumulative output for any technology) and leads to higher productivity and higher locational benefits. The locational decisions by firms depend on the available benefits. The shape of locational benefits functions (and, consequently, the shape of the cluster’s growth function) is crucially dependent on the technology used in the cluster. As shown in Figure 11.3, the growth rate of the established cluster follows a classic logistic function (initially large, then linearly decreasing with the cluster’s size). The growth of the new cluster is described in Figure 11.4 where a cubic logistic function, implying a bellshaped growth pattern, that is the growth rate is low for both very small and very large sizes of the cluster. If firms relocate, they absorb the locational benefits of the respective cluster (technological and knowledge spillovers).16 In particular, when firms relocate from the old cluster to the new one, they are able to exploit the knowledge spillovers, to poach already trained workers and thus they become able to use the new technology.17 The results obtained by Brezis and Krugman may be replicated in this adapted framework and are rather sharp: When the new technology becomes available, firms in the established center (cluster) do not adopt it, because given their experience they remain more productive with the old technology. A new smaller center (cluster) comes into being, however because the new technology is good enough to compete with the old (in a newly established cluster), as the new technology matures through learning, both the new technology and the new city-region (cluster) that is based upon it, take over from the established region (cluster) (Brezis and Krugman 1997: 380).

Our story is based on the assumption that the key external economies that support the development of the cluster are learning effects associated with the geographic concentration of an industry in a cluster. As long as the technology undergoes normal progress (i.e. follows a technological trajectory) the interchange of knowledge within the established cluster will tend to preserve its leadership. When new technologies arrive that are discontinuous with those that came before (i.e. change the technological paradigm), existing industry concentration may be of little value. The result then is that new technologies tend to be exploited in new clusters that do not suffer the diseconomies associated with an established cluster. The relation between the introduction of a technological innovation and the emergence of a new cluster is crucially dependent on the fact that

227


learning economies are greater—and can be identified—for the new technology in a two-period framework (Figure 11.5). The sizes of the clusters are measured along the same horizontal axes so that ni þ nj ¼ N.18 In the first period (denoted by superscript 1) cluster i has a larger carrying capacity Ki (because of the established technology, a large market and the cumulated output); while cluster j, which is based on an innovative technology (and a new product with an initially smaller market), is characterized by a smaller one, Kj .19 Given that firms choose in which cluster to locate on the basis of the level of locational benefits, the only stable equilibrium in the first period is E1 where the old cluster locational benefits function (B1i ) intersects with the new cluster one (B1j ) and ni >> nj given that: . for ni > E1 , B1j > B1i , therefore firms will leave cluster i for cluster j, reducing ni . . for ni < E1 , B1j < B1i , thus firms will leave cluster j for cluster i, increasing ni . . In the second period (denoted by superscript 2) both functions of locational benefits have shifted up because of learning economies. However, given the diminishing returns to experience and the characteristics of the new technology, function Bj rises more than Bi (Kj increases more than Ki ). Therefore the new equilibrium is established at E2 where B2i ¼ B2j and n2j >> n2i , and the new cluster leapfrogs the old one.

Bi

Bj

B 2j

B2i

B1j

B1i

nj

K2j

K2j

E2

E1

N

E2 n2j >> n 2i

E1 n1i >> n 1j

Figure 11.5. Learning effects and clusters leapfrogging

228

K1i

K2i

ni


The above results are based on a comparative static framework, where only the relationship between the size of the cluster and the level of locational benefits is explicitly modeled. The following section introduces a dynamic version of this model.

An Ecological Modeling of Clusters’ Technological Life Cycle Assume that after an improved new technology is discovered some time (t) after the birth of the old cluster (i). As shown above, the new technology will not instantaneously be adopted by the old cluster i due to its higher productivity that originates in the cumulated experience of the old technology. Thus, if a cluster (j) emerges that is based on the new technology it will arise in a previously deserted space, that is where there is no previous experience of any technology. Because of the superior performances of the new technology and the larger learning economies, the new cluster will attract a share l of the firms located in cluster i. Therefore, cluster i will experience an outflow of firms (i.e. a share of its firms’ stock) which is postulated to be proportional to the industrial mass of the new cluster j (which is growing overtime). That implies that the outflow of firms from cluster i to cluster j will initially rise (because cluster j grows faster than cluster i) and then decline (when the industrial mass of cluster i is strongly reduced). Figure 11.6 graphically displays the net growth of cluster i which depends on the difference between inflows Gi (as described in Figure 11.3) and outflows (Oi ). In each moment of time the dynamics of the cluster will be determined by the relative position of Gi and Oi . Every time inflows are larger than outflows (Gi > Oi ) the net number of firms (ni ) will increase, whereas a decrease will take place when Oi > Gi . The change in firms (n_ i ) will equal to zero only at the intercept with the horizontal axis and when Gi ¼ Oi . In general, three different dynamics patterns can be discerned in the old cluster: . If 0 # ni < n0 , the cluster will unravel until it disappears; . if n0 # ni < n00 , the cluster will grow and reach size n’’ (with n’’ < K); . if n00 # ni , the cluster will decrease until size n’’ is reached. Hence, there exist two stable equilibriums (0 and n’’) in Figure 11.6, while n’ is unstable. Should the outflow accelerate, a new situation with only two equilibrium points emerges, as depicted in Figure 11.7. First, if 0#ni < n the

229


cluster will unravel until it disappears. Second, if n > ni , the cluster will decrease until size n is reached. Should the rate of outflow increase even further (Oi > O0i ), the only stable equilibrium point is the origin and the cluster is doomed to extinction. The development path of the old cluster i—taken its structural parameters as given—will crucially depend on three parameters20 which refers to

Cluster’s growth n

Oi

Stable equilibrium Unstable equilibrium

Gi

K n⬙ Cluster’s size n

n⬘

Figure 11.6. The growth and depletion of an industrial cluster

Cluster’s growth

n O ⬘i

O⬙i

Stable equilibrium Saddle point Gi

A

K

n* Cluster’s size n

Figure 11.7. The growth and depletion of an industrial cluster

230


the new cluster j and on the time lag between the start of the old and the new cluster. Those are: . rj , the intrinsic growth rate of cluster j . Kj , the carrying capacity of cluster j . l, the share of migrant firms in cluster i.

Policy Implications Local policymakers as well as private agents interested in the development of cluster i—as the developers in Henderson (1977)—have two different policy alternatives. Either policies can be pursued that aim at maximizing the rate of cluster’s growth (ri ), or can policies strive at enlarging the longrun equilibrium size of the clusters carrying capacity (Ki ). This first policy (r-type) is designed to increase positive externalities endogenously generated by the location of a new firm in the cluster. The rate of growth (ri ) expresses the largest possible attraction and generation power of a given number of located firms and influences the growth rate of the cluster (Figure 11.8).21 An r-type policy explicitly supports agglomeration economies and knowledge spillover in the development process of a cluster. Moreover, it also expresses the difference between firms’ birth and mortality rates in the region. Hence, r-type policies may embrace both instruments to increase the birth rate and to decrease firms’ infant mortality rate (through innovation diffusion supporting policies, start-up incentives, provision of business planning services, diffusion of VC

n

n

n

time

Figure 11.8. R-type policy Source: Maggioni (2005a).

231


activities, and so on). An r-type policy is therefore a policy intervention to be used in order to foster the initial phase of development of a cluster in an initial hostile environment. A K-type of policy is designed to increase the regional carrying capacity which is a region’s ability to sustain a given number of profitable firms (Figure 11.9).22 A region’s carrying capacity is a function of the local endowment of resources (inputs and infrastructures) and of the average level of use of these resources made by resident firms. Thus, any public policy aimed at increasing the quantity and/or quality of local inputs and infrastructures, and at raising the efficiency of local firms can be defined as a K-type policy. Given that policymakers face some budget constraints, they must use some ordering criterion to make the best policy choice. In general, it has been shown by Maggioni (2002b) that the desirability of these different development policies is crucially dependent on the preferred target of the intervention, the chosen time framework for the implementation of the policy, the level of development of the targeted cluster and the state and variability of the relevant external macroeconomic conditions. As far as the target of the policy is concerned, r-type policies are mainly addressed to firms, while K-type policies are usually directed toward the economic environment and the productive and urban infrastructures of the local economic system. According to this taxonomy, r-type policies imply interventions, such as start-up incentives, fiscal allowances, information diffusion programs, and so on while the fostering of the local

n

n

n

Figure 11.9. K-type policy Source: Maggioni (2005a).

232

time


university and the strengthening of the regional network of transport and communication infrastructures can be defined as K-type policies. An alternative criterion relates to the time horizon needed for the implementation of the economic policy interventions. Usually r-type policies generate results in the short run, while K-type policy needs a longer time to be effective. On the other hand, while the first type merely influences the starting date and speed of development (without changing any structural conditions), K-type policies are the only ones capable of moving the cluster size from a lower equilibrium level to a higher one, thus ensuring higher sustainable long-run growth. The third criterion refers to the stage of development of the targeted region. An r-type intervention is perfectly suited to be implemented in a developing region where the main problem is the establishment and early survival of the seed firms within the cluster. On the other hand, K-type interventions should be implemented in industrially developed areas where competition on inputs and congestion of infrastructures are the main obstacles to the further development of the cluster. Finally, the state and variability (i.e. depth and frequency of exogenous shocks) of the relevant external macroeconomic environment may influence the choice of policy.23 A pure r-type or K-type policy would be called for when the environment is stable; an intermediate policy when shocks are frequent and limited; and a mixed policy (i.e. a weighted combination of pure r-type and pure K-type) when shocks are deep but infrequent.24

Policy Simulations Figures 11.10 and 11.11 illustrate two possible outcomes of the interaction between the old and the new cluster: either the two clusters coexist (even if the new one overcomes the old as far as the economic mass is concerned) or the cluster j attracts all firms and cluster i is driven to extinction. From the above analysis it is self-evident that cluster i is more likely to survive the lower the intrinsic growth rate, carrying capacity and attractiveness of cluster j. More specific and interesting results derive from the simulations performed in order to assess the optimality of different development policies. In order to perform a series of simulation, first it has been necessary to transform equation (6), described in the Appendix, into an equivalent difference equation version as follows:25

233

The Rise and Fall of Industrial Clusters .

.

.

Figure 11.10. Long-run equilibrium: two clusters coexist

.

.

.

Figure 11.11. Long-run equilibrium: cluster j drives cluster i to extinction

Ni ¼ Gi Oi ¼

Dn ni (t) ¼ ri n2i (t) 1 Lni Dt Ki

(7)

which describes the net growth of cluster i as the difference between the number of firms entering the cluster and those leaving the cluster in the discrete time interval Dt.

234


Second, specific values to all the parameters have been assigned, through a first set of simulations, in order to build a benchmark case (i.e. a case in which the development of cluster j does not hinder the survival of cluster i): Ki ¼ 100,

Kj ¼ 120,

t ¼ 20,

L ¼ 0:05,

ri ¼ 0:1, Dt ¼

rj ¼ 0:2,

ni (0) ¼ 1,

nj (0) ¼ 1,

1 : 5

Third, it has been chosen a value of the attractiveness coefficient (L ¼ 0:7) which a first set of simulations have shown to produce the extinction of cluster i. Fourth, different policies have been performed on cluster i to check whether they were able to sustain the growth of cluster i even in presence of an extremely attractive cluster j on the basis of a twofold experiment: . Experiment 1: it has been assumed that the old cluster’s policymaker could either double the intrinsic growth rate ri or the carrying capacity Ki of the old cluster. . Experiment 2: it has been assumed that the old cluster’s policy maker could find comparatively easier to overcome some structural constraints hindering the cluster’s growth (K-type policy) than to lower the general entry and relocation costs of all firms in an old-established cluster (r-type policy).26 Table 11.1, under the heading ‘Sensibility on . . . ’, shows in columns 3, 4, and 5 the highest value of the parameters (rj ,Kj ,L) describing the growth patterns and the attractiveness of cluster j still compatible with the existence of cluster i and ni , the largest possible size of cluster i in

Table 11.1. Simulation results 1

2 Policy interventions

3 Sensibility on rj

4 Sensibility on Kj

5 Sensibility on L

rj

ni

Kj

Ni

L

ni

Exp. 1

KiP ¼ 2Ki ¼ 200 riP ¼ 2ri ¼ 0:4

0.35 1.50

130 39

208 331

67 35

0.10 0.14

93 44

Exp. 2

KiP ¼ 2Ki ¼ 200 riP ¼ 1:5ri ¼ 0:3

0.35 0.39

130 52

208 224

67 35

0.10 0.11

93 39

235


that condition, given the specific policy intervention (described in column 2) performed. The higher the values of rj , Kj , L, the more effective (in terms of resilience) the policy interventions performed on cluster i, since this shows that the old cluster—strengthened by a specific policy intervention—is able to survive the competition of an overperforming cluster j. A somehow different meaning has to be attributed to the value of ni . Here, the higher the value, the more effective the policy interventions in terms of growth of the original old cluster. The simulations performed show that r-type policies outperform (in terms of resilience) K-type policies (when the only threat for the established cluster is the development of the new one.27 This result is further reinforced if one considers that in the economic reality r-type are also more efficient than K-type policies (i.e. that a dollar spent on r-type policies produces a larger change in the relative parameter than a dollar spent on K-type policies).28 However, one may note that the long-run dimension of cluster i is achieved (as expected) after performing a K-type policy. This may signal a trade-off between resilience and growth in the development policies of industrial clusters and confirms that r-type policies are better suited to foster the survival of an established cluster in a situation of technological and market turbulence.

Conclusions After the crisis of the mass production model, the emergence of successful innovative industrial clusters in the late 1970s suggested that the opportunity still existed for technological development and continued economic growth. Small and dynamic high-tech firms were regarded as the main, if not the sole, engine of economic development, and regional innovation policy became the target of every public authority all over the world dreaming about the creation of a Silicon Valley clone. At that time, the dream of a set of powerful regional innovation policies able to generate the development of any areas seemed justified, since the development of an innovative industrial cluster was promised simply by linking the existing pools of local resources to the dynamics of international supply and demand of innovative technologies, products, and services (Blakely 1989).

236


Some twenty-five years later, it has been widely established that it is rather difficult, if not almost impossible, for regional policies alone to radically change the local patterns of innovative production. The amenity of the place (for the peripheral regions) or the existence of an industrial tradition, for the old-industrialized regions (Swann, Prevezer, and Stout 1998) could not constitute the basis of successful implementation of a modified science-based version of the Perroux’s growth pole theory (1955). From the analyst’s perspective, the spatial agglomeration of a number of innovative firms, which causes the genesis of a new high-tech cluster, appears such a complex and serendipitous phenomenon that no theoretical approach or econometric estimation may pretend to fully explain it. This chapter has followed a different strategy, by adopting a population ecology approach, and has focused on the interactions existing between technological innovation and the life cycle of clusters. Thus, in a sense, the claim that industrial location patterns are created through ‘the process of growth rather than through a process of efficient allocation of plants across a static economic landscape’ (Storper and Walker 1989: 70) and that ‘industries produce regions (clusters) and are capable of creating their own geography’ (Storper and Walker 1989: 70), is confirmed. The results stress the importance of lateral thinking in development planning: both scholars and policymakers should in fact always look carefully at the industrial and technological dynamics which may swiftly transform a new cluster into an old one and reverse the direction of the cumulative causation process, thereby transforming a successful developing cluster into a decaying one. Thus the choice between alternative technologies is a risky one. Betting on the wrong technology may have catastrophic effects on the economic development of the area. Finally, the results contrast a common wisdom among European policymakers, which overemphasizes the importance of carrying capacity (universities, infrastructures, and so on) in determining the success of an innovative industrial cluster, at the expense of firm-based microlevel incentives aimed at increasing the endogenous growth of the cluster. The provision of scientific, logistic, and economic infrastructure, as well as the attraction of exogenous agents (such as large corporations’ branches), may well support the development of a cluster in an undisturbed external environment, but its long-run performance in a competitive and turbulent environment will be determined by generating a sustained formation rate of small endogenous new firms and by supporting their growth.

237


Acknowledgments This chapter has greatly benefited from fruitful discussions with S. Beretta, M. Riggi, and T. E. Uberti. I would like to thank P. Braunerhjelm, M. Feldman, and M. Prevezer for helpful comments. The usual caveats apply.

APPENDIX The growth of a cluster, when only agglomeration economies are considered, may be modeled as follows:

dnq ¼ n_ q ¼ rq nq (t) dt

(1)

where nq is the economic mass of the cluster (number of already established firms) and rq is the intrinsic growth rate. Equation (1) can also be solved for nq (t) as function of the exogenous initial industrial mass of the cluster nq (0):

nq (t) ¼ ert nq (0)

(2)

If negative feedbacks (such as congestion and competition effects, or epidemic dynamics) are included in the analysis, then the simplest dynamic model, which takes into account these features is the logistic equation,29 which can be written alternatively as (3a), in the quadratic case, and (3b) in the cubic one.

dnq ¼ n_ ¼ rq nq (t) dt dnq ¼ n_ ¼ rq n2q (t) dt

nq (t) 1 Kq

1

nq (t) Kq

(3a)

(3b)

In the quadratic version of the logistic function (which, in Section 3, describes the growth path of the established old cluster, i), the individual firm’s contribution to the cluster’ growth (and the cluster growth rate) is highest at the beginning of the development process and decreases as a linear function of the cluster’s population size:

ri n_ i ¼ ri ni (t) ni Ki

(4a)

In the cubic version of the logistic function (which describes the growth path of the new cluster, j), the individual firm’s contribution to the cluster’s growth (and the cluster growth rate) is a quadratic function of the cluster population size:

238


rj n_ j ¼ rj nj n2j (t) nj Kj

(4b)

Because of the superior performances of the new technology, for a given level of cumulated output, and the larger learning economies, the new cluster j will attract a share l of the firms located in cluster i. Therefore cluster i will experience an outflow of firms as described by equation (5a):

_ i ¼ lni Oi ¼ m

(5a)

According to equation (5a), the outflow from cluster i to cluster j is a constant share of cluster i industrial mass. However, it seems reasonable to suppose such an outflow to be also proportional to the industrial mass of cluster j. Equation (5a) can be modified to take into account this remark by substituting pffiffiffiffiffiffiffiffi L ¼ l ni nj =ni to the original value of the slope of the outflow function (l), thus _ i as a function of the product of the industrial masses of the two modeling m clusters.30

_ i ¼ Lni ¼ Oi ¼ m

pffiffiffiffiffiffiffiffi l ni nj pffiffiffiffiffiffiffiffi ni ¼ l ni nj ni

(5b)

The net growth of cluster i will depend on the difference between Gi and Oi :

ni (t) 2 _ Lni Ni ¼ Gi Oi ¼ ni ¼ ri ni (t) 1 Ki pffiffiffiffiffiffiffiffi ni (t) l ni nj ¼ ri n2i (t) 1 Ki

(6)

Notes 1. For a detailed discussion and an extended survey of different explanations for cluster formation, see also Maggioni (2004a, 2004b). 2. Even if no physical borders exist to the expansion of the cluster one must take into account the existence of organizational minimum efficient scale. 3. For analytical convenience, I split locational benefits in two classes: geographic and agglomeration benefits. The first class refers to those components which are unaffected by the number of incumbents; while the second refers to those components which depend on the number of incumbents. By adopting this formulation, however, I do not intend to state that agglomeration benefits refer only to spillovers of scientific and technological knowledge and know-how. On the contrary I am convinced that relevant agglomeration benefits derive also from external economies of scale in the use of local resources. The same variable

239


4.

5.

6. 7.

8.

9. 10. 11. 12. 13.

14. 15.

240

(i.e. labor productivity) has a fixed geographic component, which depends on the quality of local workers, and a variable agglomerative component which depends on the number of firms already located in the cluster. A similar reasoning applies to costs. For an analytical version (see also Arthur 1988a, 1988b, 1990). An alternative explanation for the convexity of the locational costs function for firm f runs as follows: the locational costs function is composed of a fixed and a variable component. The fixed part of the costs (geographic costs) decreases as the number of entrants increases; while the variable part increases (because of competition) as the number of entrants increases. The combination of these two effects produce a U-shaped (convex) cost curve as the interaction between fixed and variable costs of production in standard microeconomics textbooks. A symmetric reasoning may also explain the inverted U-shaped benefits function. This interpretation is surely more realistic than the one used in the chapter, however it is not as theoretically efficient as the other one since both components become dependent on the number of incumbents. However, as it is made graphically evident in Figure 11.1, because of the inverse U shape of the marginal benefits function, there is a range, within the number of incumbents, where marginal net benefits are already decreasing, but still higher than average ones, and average net benefits are still increasing. Weber (1929), Isard (1956), Henderson (1977), Richardson (1978), Papageorgiou (1979), Tauchen and Witte (1983), and Miyao and Kanemoto (1987). Marginal benefits and marginal costs refer to the variations in those variables induced by the entry of a further firm. Average refers to the actual size of benefits and costs borne by any firm in the cluster. By considering average functions I indirectly assume that some market mechanism is at work in the cluster and makes both benefits (and costs) equal for each incumbent. Here the reference is to the debate between least cost (a` la Weber) and demand ¨ sch and Hotelling) approaches. side (a` la Lo After K, new entries thus support a turnover process without causing relevant changes to the equilibrium level. This is in terms of changes in the average locational net benefits, due to the interaction of agglomeration economies and diseconomies. The higher the cluster’s growth rate (rq ), the faster is the growth process. The difference may refer to different industries in the same geographic area, or to different geographic sites in the same industry. These estimated parameters could also be used as dependent variables in cross-section analyses in order to assess the influence of different factors on the level of the intrinsic growth rate of a cluster or on its maximum dimension. Any unimodal function will fit the bill. In a series of papers Maggioni (1993, 2002a, 2005b) and Gambarotto and Maggioni (1998) extend this simple logistic equation to a system of differential


16. 17.

18. 19. 20. 21. 22.

23.

24. 25.

26.

27.

equations in order to take into account a series of different interindustry and intercluster bilateral interactions. Or the industrial atmosphere, to use the most famous Marshallian quote. Technology is assumed to behave like tacit knowledge (Polanyi 1966), which is embedded in individual and social experience and involves intangible factors which can hardly be communicated over distance. The inequality part of the expression ensure the existence of a nontrivial (i.e. when each cluster reaches its own K) interaction between the two clusters. In order to be consistent with Hypothesis H3, it must also be assumed that rj > ri . for a discussion of these parameters, see the Appendix. This encompasses the entry of firms that were located outside the cluster and the birth of new firms inside the cluster. The analysis makes use of the concept of representative or average firms to take into account the fact that, in reality, firms differ in size and that the growth of a high-tech cluster may imply either the increase in the number of established firms (i.e. the entry of new firms in the region) or the growth in size of a number of located firms. See Swann, Prevezer, and Stout (1998). For an open and internationally integrated region the relevant external environment may well be the world, for a closed and underdeveloped region the relevant environment is likely to be limited to the nation, for an intermediate type of region, the external macroenvironment is a group of countries (i.e. Europe for a European country). For a detailed analysis of the relevance of the external macroeconomic conditions, see Gambarotto and Maggioni (1998). The simulation package (Stella Research 7.02) transforms the original differential equations of the theoretical model into difference equations. Therefore the choice of Dt (the interval of time between calculations) is somehow crucial in order to avoid weird results (it must be remembered that a logistic equation in discrete time may even produce chaotic behaviors). It should also be noted that throughout the simulations time must be intended as logical time (i.e. runs) and not as historic time. In particular, in case 1 it has been exogenously assumed that an r-type policy would double the intrinsic growth rate of cluster i(riP ¼ 2ri ¼ 0:4); while a K-type policy would double the carrying capacity of cluster i(KiP ¼ 2Ki ¼ 200). Different experiments have been performed by attributing different values to different policy interventions. In particular, case 2 shows the paradoxical results for an r-type policy which raises the intrinsic growth rate by one and a half (riP ¼ 1:5ri ¼ 0:3). The sensibility analysis has been performed by testing the effectiveness of r-type and K-type policies r an k; the different policies have been tested against a longer series of different on rj ,Kj , and L.

241

The Rise and Fall of Industrial Clusters 28. The simulation results show that the superiority of r-type policies is such to be proofed even in the most unfavorable (and unrealistic) situation. 29. The logistic equation—first developed by Verhulst (1845) and Pearl and Reed (1920) for demographic studies, then adopted by the ecological literature since Lotka (1925): ‘is the simplest model containing negative density dependence interaction. Further, it is the first two terms in a power series expansion of a more general growth model where the growth is a function of the actual size of the population’ (Dendrinos and Mullally 1985: 38). 30. This is the usual way to model the interaction between two populations in the population ecology literature from Lotka (1925), to Maynard Smith (1974), to Roughgarden (1979). For an application to industrial and regional economics, see Dendrinos and Mullaly (1985) and Nijkamp and Reggiani (1992). For an explicit application to industrial clusters see Maggioni (2002a).

242

12 Local Antecedents and Trigger Events: Policy Implications of Path Dependence for Cluster Formation David A. Wolfe and Meric S. Gertler

A critical issue for any consideration of the genesis of clusters and the role for policy in their origin is the relative importance of chance events, or serendipity, as opposed to rational or intentional actions. As greater attention is focused on the promotion of clusters as an economic development policy tool, the question of whether, and how, they can be fostered assumes greater significance. Central to this debate is the role of path dependencies created by small, initial—often chance—events, as opposed to the role played by deliberate actions by both private actors and public sector agencies in contributing to the genesis of clusters. The concept of path dependency has been adopted by a wide range of disciplines to analyze and explain a broader range of social phenomena—sometimes in a rather deterministic fashion. The concepts of path dependency and lock-in as they have been developed in evolutionary economics are complex and somewhat counterintuitive in the sense that they set out to explain how structured patterns of development—across both space and time—can result from seemingly chance or contingent occurrences. They have proven effective in explaining why and how certain technologies prevail in the competitive setting of the marketplace, although they may not always be technologically superior. The challenge in applying the concept to other disciplines and problems—such as the genesis of clusters—lies in determining precisely what aspects of a developmental path or trajectory can be attributed to underlying causes or preconditions, and what aspects are the result of chance or contingent events.

243

Local Antecedents and Trigger Events

The evolutionary approach to economics reminds us that economic systems change over time, but in ways that are shaped and constrained by past decisions, chance events, and accidents of history—implying a certain degree of serendipity. Current decisions and events are not determined by past ones, but they are conditioned by them. As a result of past choices and events, certain possibilities are easier to pursue in the present, others less so. This is key to appreciating the implications of path dependency for locational theory and accounts of cluster formation. However, extending the concept of path dependency from the narrowly technological to the social and political dimension raises a series of problems— both for academic researchers and for active policymakers. The challenge lies in reconciling the role that chance events play in launching a specific technology on the path to market dominance or a particular region to enjoying a concentration of firms in a cluster with the scope for subsequent change in broader institutional structures and settings. As it has been applied to locational theories of cluster development, path dependency has downplayed the role of serendipity or chance occurrences in launching the initial genesis of individual clusters in specific locations and overemphasized the subsequent advantage enjoyed by these regions against potential competitors. It suggests that the trajectory of specific regions and localities is rooted in a series of economic, social, and cultural factors that influence their development over time. The presence, or absence, of key institutional elements of the local innovation system may affect both their innovative capacity and their potential to serve as nodes for cluster development. However, path dependency should also remind us that the confluence of these factors in a specific location may have initially resulted from a set of chance events or occurrences rather than the conscious designs of private or public agents. This poses a significant challenge for policymakers charged with the goal of promoting the emergence and development of clusters in their local or regional economy. The following chapter explores the relation of path dependence to previous theorizing in the fields of economic geography and locational analysis and its contemporary value for both understanding the genesis of clusters and the practical constraints on policy designs for promoting their development.

Path Dependence and the Origins and Growth of Clusters The concept of path dependence originates with the desire of evolutionary economists to account for the factors which determine the selection

244


mechanisms that exist within the process of technological choice and the natural trajectories that emerge from those patterns. Brian Arthur (1994) and Paul David (1997) used path dependence to explain how and why certain technologies emerged and prevailed over competing technologies in periods of rapid innovation when the marketplace was characterized by a number of alternative technological designs. Paul David defines a pathdependent sequence of economic changes as one in which important influences upon the eventual outcome can be exerted by temporally remote events, including those dominated by chance elements rather than systematic forces. He suggests that in a dynamic process, positive feedbacks are generated by strong technical complementarities on the supply side of markets, and/or the interdependence of customer preferences operating on the demand side. These may arise as well from learning effects and habituation associated with the sunk cost effects of new technologies—such as those involved in learning how to use a new program. But he also insists that the concept of path dependence does not mean that economic outcomes are predetermined. Instead he quotes approvingly from Douglas North to reinforce his point that ‘contingent probabilistic events have a place throughout the dynamic process’ (David 1997). North argues that, ‘At every step of the way, there were choices—political and economic—that provided real alternatives. Path dependence is a way to narrow conceptually the choice set and link decision-making through time. It is not a story of inevitability in which the past neatly predicts the future’ (North 1990: 98–9). There is a closely related idea within the evolutionary approach—that of increasing returns. It refers to a process in which, once a particular economic change occurs, it becomes self-reinforcing. Brian Arthur, who is equally credited with elaborating the concepts of path dependence and increasing returns, maintains that in many areas of economic activity, stabilizing forces do not seem to operate; rather, positive feedback amplifies the effects of small economic shifts. The presence of positive feedbacks and the phenomenon of increasing returns make possible many equilibrium points rather than the single equilibrium point posited by the neoclassical model based on the notion of diminishing marginal returns. Once a set of chance events or a series of small historic accidents push the technological trajectory of a new product or process onto a certain path, the prevailing technology may become locked-in regardless of the purely technical advantages of the competing alternatives. The initial advantage may be acquired through small, seemingly insignificant events and the triumphant variant is not necessarily the technically superior or more

245


efficient one. Its dominance may be based purely on the fact that it was the first to gain wider acceptance in the marketplace, and that many supplying businesses, distribution networks, supporting technologies and users, and a large community of users and developers, all converged on its design. However, once it establishes a lead, further technological development is locked into the trajectory or path set by the dominant products. Competing technologies that were available at the outset quickly fade from view and become little more than historic footnotes (Arthur 1988a, 1988b). Evolutionary economists, historic sociologists, and economic geographers have expanded on the original application of the concept. While the specifics of the application vary across this range of disciplines, social scientists suggest that path-dependent analysis shares several common features. In the first place, it involves the study of causal processes that are sensitive to a series of events which occurred in the early stages of the causal sequence. Events that occur early in the sequence tend to exert a disproportionate influence over the long-term development path of the sequence. Second, these early events involve a high degree of chance or contingency that cannot be explained purely on the basis of the starting conditions or initial factor endowments. Similar starting conditions may lead to a wide range of possible outcomes. This fact makes it particularly difficult to forecast patterns of development based on the initial conditions. Finally, once the chance events have occurred, the path-dependent sequence exhibits a more deterministic pattern, involving a large degree of irreversibility. In economic and geographic systems, the degree of irreversibility is strongly reinforced by the effects of increasing returns to scale (Mahoney 2000: 510–11). The complementary concepts of path dependence, increasing returns and lock-in have obvious relevance for understanding the historic paths taken by regional clusters. Once a regional cluster establishes itself as an early success in a particular set of production activities, its chances for continued growth tend to be high. While this may be to some extent reducible to the success of dominant lead firms in the region, the more interesting aspect of this process has to do with the collective processes and forces at work: local social and economic institutions and culture. By the same token, ailing places may also face great challenges in improving their fortunes, for the same reason. Once a path-dependent trajectory of decline becomes established, institutional and cultural lock-in will make deviation from this path a serious challenge. The rich geographic literature on path dependence, increasing returns and lock-in has its own distinctive evolutionary trajectory. Within this

246


literature, three broad approaches can be distinguished that emphasize different aspects of the dynamics in regional development. The first approach focuses on the way in which the initial economic activity in a region triggered by accidents of history tends to become self-reinforcing. The second approach, following in the tradition of Marshall (1923), places greater emphasis on the influence of agglomeration economies and supply side externalities. The concentration of critical factors of production in specific regions tends to reinforce the effects of increasing returns in the region. Finally, a more recent tradition focuses on the extent to which the benefits derived from externalities in the form of knowledge spillovers are frequently tied to ensembles of related capabilities. From this perspective, the economic advantages conferred by the institutional infrastructure of the region are a vital element in the supply architecture for learning and innovation. We build on these previous findings when addressing the issue of how to promote the growth of cluster-based development within the nexus of innovation, experimentation, and learning. There have been highly influential classic works by Myrdal (1957), Hirschman (1958), and Kaldor (1970) on disequilibrium models of regional economic development. These authors endeavored to show how initial economic activity, triggered by accidents of history and geography, become self-reinforcing over time and lead to growing geographic unevenness and inequality. In Kaldor’s version of the story, early growth in the core region sets in motion private and social dynamics based on increasing returns to scale. Myrdal and Hirschman similarly outline a process of circular and cumulative causation, defining an evolutionary path in which backwash or polarization effects (such as selective out-migration of skilled labor and the net outflow of capital from peripheral to core regions) outweigh spread or trickle-down effects so that initial growth in the core region begets further growth, and initial disadvantage in peripheral regions becomes amplified over time. In this manner, initial events trigger long-term processes of interregional divergence which are extremely difficult to reverse. Indeed, the primary motivation for all three of these authors was to justify why public sector intervention at the national level was necessary in order to overcome these powerful, increasing-returns dynamics exhibited at the regional level. One of the first to link the concept of increasing returns to the division of labor and, at least implicitly, the geography of production systems is Allyn Young (1928), who noted how the intricate set of interdependencies between firms in a well-developed social division of labor leads to increasing returns dynamics. Young’s early insights have stimulated a

247


more recent reexamination of the dynamics of growth from an explicitly geographic perspective (Scott 1988; Storper 1999). In recent years, economists, such as Paul Krugman (1991a, 1991b, 1991c, 1991d) and Brian Arthur (1994), have drawn upon this rich tradition of earlier ideas within economics and geography to fashion more formalized models of territorial development. Krugman’s intellectual debt (1991a, 1991b, 1991c, 1991d) to Kaldor and the other early adherents of increasing returns theories of economic development is especially clear. He builds on these ideas, as well as Alfred Marshall’s original thinking (1923) on the nature of agglomeration economies, by specifying the types of supply side externalities that generate localized increasing returns. The first source is the large, deep pool of specialized labor created by the concentration of firms within the similar industries in the same location. The second arises from the fact that a local concentration of firms in the same industry can also support a larger number of specialized local providers of intermediate inputs and services, and thus reduce the cost to firms. Finally, the colocation of similar firms in a region can generate positive technological externalities or spillovers that can flow more easily among the similar firms than over longer distances (although Krugman is more skeptical about this externality). Overall, Krugman endeavors to show that the phenomenon of increasing returns is a key aspect of the process of industrial clustering that leads to a pathway of increasing sectoral specialization in particular regions over time (Baptista 1998: 27–9; Krugman 1991a, 1991b, 1991c, 1991d ). While Krugman’s work focuses on the way in which scale economies and positive externalities can feed the process of industrial clustering, Brian Arthur’s work focuses more specifically on the way in which agglomeration externalities contribute to the concentration of firms in specific regions. More recently, Maskell and Malmberg (1999) have argued that the competitive success of firms depends on their ability to develop sustainable, distinctive capabilities. These capabilities are most likely to arise from nonubiquitous and tacit forms of knowledge related to products, processes, and organizational routines within the individual firm. However, they will also arise from socially organized assets, such as localized, learning-based, interfirm relationships, that are not easily replicated by (groups of) firms elsewhere. Maskell and Malmberg (1999: 173) argue still further that a region’s institutional architecture accumulates and changes incrementally over time, and ‘thus represents the intricate contemporary interaction between elements of different ages . . . from the very old (religion, beliefs, and values) to the recent/current (contemporary

248


industry standards, current regulations, etc.)’. Because of these properties, this institutional endowment can become a key part of a region’s nonreplicable asset base, thereby reinforcing durable local competitive advantages that are difficult for competitor regions to emulate. Gertler (2004) develops this idea more fully, showing how this regional institutional architecture consists of social structures that shape the attitudes, norms, expectations, conventions and—ultimately—the practices of individuals and firms in the region through informal and formal means of regulation. These factors together determine the technological specializations of individual countries and regions; a pattern of specialization that may actually be increasing, despite the increasing reaches of globalization. They also provide an important clue for our understanding of how the trajectories of development for particular regions or local economies may be conditioned by the preexisting conditions—in turns of productive competencies in older technologies and products that firms located in the region enjoyed. In certain instances, this can help explain how a series of small chance events were able to take hold more successfully in the fertile soil of one region rather than another and launch it on a new path of development. However, one danger with these interlinked concepts is that they can serve as a double-edged sword—both to explain the social and technical bases of success for certain regions, but also to suggest the existence of constraints on the potential for others.

The Origin of Clusters: Theoretical Foundations and Empirical Findings While policymakers seem intent on finding policy-relevant solutions to this problem, the academic literature has been less successful in formulating a clear and consistent set of answers to this question, leaving the field to a host of consulting firms that have emerged to guide municipal and regional ¨ lvell, governments through an increasing array of cluster initiatives (So Lindqvist, and Ketels 2003). Still, Michael Porter, widely recognized as one of the leading authorities on cluster research and policy, is surprizingly clear on the factors that contribute to cluster formation and equally clear on the potential role that policy can play in their formation. He does not phrase ‘this reference’ in terms of path dependence and increasing returns; rather, he traces the roots of a cluster to his well-known diamond model of competitive dynamics. That is, cluster emergence depends on the local conditions for factor input, demand, firm strategy and rivalry, the

249


presence of related and supporting industries and, finally, historic circumstances (serendipity) and policy. Although in principle he affords all four corners of the diamond equal weight as factors contributing to the seeding of clusters, he clearly privileges the role of factor input conditions such as specialized skills and talent, specific areas of expertise in the research infrastructure, an attractive physical location, and especially supportive infrastructure (Porter 1998: 237). The key assets that determine the viability of a cluster are firm based. Of particular importance is the emergence of a core or anchor firm for the cluster. Whole clusters can develop out of the formation of one or two critical firms that feed the growth of numerous smaller ones. Examples of the role played by this kind of anchor firm can be found in the case of Medtronic in Minneapolis, or MCI and AOL in Washington, DC, or in the Canadian case, by the role of Northern Electric (now Nortel) in the genesis of the Ottawa telecom cluster. Once a cluster is launched by this combination of locational assets, chance events, and entrepreneurial dynamism, Porter affords a strong degree of importance to the role of increasing returns and feedback. The emergence of a major anchor firm in the cluster acts as a magnet for the local cluster, attracting both allies and rivals to locate in the region to monitor the activities of the dominant firm. This is the case with San Diego, where Nokia, Ericsson, and Motorola all located their CDMA wireless research efforts to benefit from Qualcomm’s leadership in the field, or in Ottawa, where Cisco and Alcatel both acquired local firms to benefit from the high degree of optical and telecommunications expertise in the region, largely spun out of Nortel, the cluster’s anchor firm. This raises the critical issue for policy analysts—what precisely is the relationship between the local antecedents that formed the basis for the genesis of the cluster and the specific events that triggered its emergence? And which of the two elements is most amenable to policy influence and which is the product of broader economic factors less likely to respond to policy stimuli?

The Knowledge–Entrepreneur Nexus A more fully developed explanation of the way antecedent conditions are transformed by trigger events into the genesis of a cluster is precisely what is missing in prevailing cluster formation theories. While it is important to acknowledge that the concept of chance does not lend itself to formal theorizing, closer examination of numerous cases suggests that we

250


actually can isolate specific factors which constitute the trigger for cluster emergence. Feldman, Francis, and Bercovitz (2005) present a descriptive model which provides this link by placing entrepreneurship at the center of the process of cluster formation. Entrepreneurs act as the key agents who build on the existing base of institutional assets that provide the local antecedents for cluster formation. Entrepreneurial activity stimulates the development of industrial clusters over time in a series of three phases. The first stage constitutes the latency phase in which a strong base of labor skills or human capital, or a significant research infrastructure is created in a region. The presence of these underlying assets is not sufficient on its own to trigger the process of cluster formation. What is required is some external shock to the regional economy that dramatically alters the opportunity cost for entrepreneurship and new firm formation. It may come in the form of a major downsizing in government laboratories or the unwillingness of a large research-intensive firm or laboratory to pursue new technological opportunities. These shifts ultimately lead employees, whether they are laid off in the downsizing or frustrated by the inability to pursue new commercial possibilities, to reexamine the opportunity cost of starting their own firms (Feldman, Francis, and Bercovitz 2005). The likelihood of this occurring is further enhanced when a movement along the technological frontier in key industries opens up a range of new opportunities for these entrepreneurs to exploit. Such technological shifts are frequently associated with a realignment of leadership positions among national economies, but they can also have the same effect on regional and local economies, as entrepreneurs in new localities are the first to perceive and act upon the potential created by these shifts (Zysman 1996). One of the reasons why the uptake of these opportunities occurs more rapidly in these new regions is that there is no lock-in to the existing technologies or production paradigm that prevailed previously. In this phase the cluster evolves further as entrepreneurs establish their own networks and build the deep institutional structures that constitute the industrial system or supply architecture of a region described earlier. Once a critical mass of new start-up firms has emerged, the entrepreneurial founders of the firms begin to form the support organizations needed to both sustain their own activities and encourage new entrepreneurs to take the plunge. These organizations engage in a range of activities, including peer-to-peer mentoring and the creation of angel networks that are essential to diffusing the knowledge and skills needed to grow and expand the cluster. Further, the establishment of these organizations raises the profile of the emerging cluster in both the local economy and more distant ones

251


and helps generate the kind of buzz that attracts new entrants and talent into the regional economy. In such a way, the institutional assets of the region are extended (Maskell and Malmberg 1999; Storper 1999). The final stage occurs when there is a fully functioning entrepreneurial environment where the success of the initial start-ups creates additional possibilities for new ones as well as spin-offs. This stage is also marked by the emergence of local VC to fund the activities of the second tier of start-ups (see also Chapter 9). This can come either from the success of a few first generation entrepreneurs cashing in and beginning to redeploy their assets or from venture capitalists from outside the region, drawn to it by the perceived explosion of investment opportunities. This perspective is echoed in the work of Swann and Prevezer (1998) on high-technology clusters. The seeding of clusters at particular junctures is strongly influenced by relatively minor historic events. They see positive feedback as a key factor playing a central role in the formation of clusters. Firms are drawn initially to a specific location by strong demand for their products or services in the location, a large supply of highly skilled or scientific labor, and a network of complementary strengths in neighboring firms; once the cluster has begun to develop, this process is accelerated by the presence of a critical mass for firms due to the positive feedback (or increasing returns engendered by colocating with similar firms). The further development of clusters is affected by two key dynamics: entry factors that attract new entrants to a cluster, and growth promoters that support the growth of incumbent firms in the cluster. The feedback process is important in accelerating the growth of clusters by enabling more sharing and transmission of tacit knowledge. Such knowledge spillovers primarily occur through labor mobility and/or the informal sharing of knowledge among technical staff at different firms. The case studies in this volume, as well as others, serve as examples of the way this process has evolved in a number of instances. Kenney and Patton (Chapter 3) underline the coevolution of technologies and institutions with respect to the origins of the Silicon Valley high-tech cluster. However, they also put a high priority on the role played by the underlying assets of the region, as do Braunerhjelm and Halverson in their analysis in Chapter 7 of the factors that led to the emergence of the Danish/Swedish biotechnology cluster Medicon Valley. Principally the universities (both publicly and privately funded) and corporate research laboratories provided the intellectual space for the growth of the Silicon Valley cluster— going back to the role of Frederick Terman in the prewar and early postwar period, in fostering a strong degree of entrepreneurialism among Stanford

252


engineering and computer science graduates and encouraging some of them to found their own firms. In the period from the 1970s onward, in which the contemporary contours of the cluster emerged, the computer science and electrical engineering departments at UC Berkeley contributed significantly to the process of new firm formation—a role that is often overlooked in the preeminence assigned to Stanford University. Two corporate research laboratories, IBM’s San Jose laboratory and the Xerox Palo Alto Research Centre, provided additional sources of new technologies for commercial exploitation and the entrepreneurs to bring them to market. The essential contribution made by these four key components of the research infrastructure is that they developed new technologies for startup firms to exploit and attracted both talented personnel and entrepreneurs to the region (Kenney and Patton, Chapter 3). A comparable case in the Canadian context that illustrates the longterm impact of building a strong research infrastructure is the contemporary information technology cluster in Waterloo, Ontario. The University of Waterloo, inaugurated in 1957, was established due to a confluence of local and national demand for more sophisticated and technical educational institutions. The strong postwar expansion of local industries generated a rising demand for technically trained labor that was not being met. Many local business leaders felt that the future competitiveness of the region depended on the establishment of world-class educational facilities. These concerns led to the creation of the University of Waterloo—a school that would specialize in a scientific and technical curriculum. Acutely conscious of the financial limitations that would exist for a new university; the local business advocates developed a unique solution in the form of the Waterloo Plan. This plan called for a new type of education to be offered on a cooperative basis with local industry. In sharing the burden of technical training with industry, the university would be able to support double the number of students (as one class rotated out to cooperative placements, another would take its place in the classroom), provide a greater depth of education—both theoretical and practical—and build a closer relationship with industry in order to anticipate employment needs, secure additional funding and ensure that classroom education remained on the cutting edge. Over the next three decades, the University of Waterloo came to be widely recognized for the strength of its mathematics, computer science and engineer programs, as well as the unique aspects of its cooperative system. In the late 1970s, these long-term investments by the local community and two senior levels of government bore fruit as key spin-offs from the university began to seed

253


the growth of local high-tech industry. While cluster formation was not an essential part of its mandate, the University of Waterloo has served the regional economy in two important ways: by providing a pool of local talent and by transferring cutting-edge knowledge, either in the form of entrepreneurial spin-off companies or through patenting, licensing, consulting or joint research projects (Nelles, Bramwell, and Wolfe 2005).

The Multifaceted Dynamics of Cluster While the Silicon Valley, Medicon Valley, and Waterloo cases provide strong evidence of the way in which antecedent conditions lay the groundwork for the conditions that trigger the entrepreneurial spark, another Canadian case provides a striking illustration of the way in which external shocks to the same can provide the trigger mechanism. The roots of the Ottawa telecom cluster can be traced back to the presence of federal government laboratories in the national capital region, many of which underwent substantial expansion during the research intensive period of World War II. This dense research infrastructure provided the fertile ground on which the telecom cluster took hold. However, the external shock which was delivered to the region took the form of the consent decree signed between the US Department of Justice and AT&T and its subsidiary, Western Electric in 1956, forcing them to make patent holdings available to other firms without charge and release technical information to outside suppliers. Up to that point, Western Electric had owned 44 percent of Northern Electric, the dominant equipment supplier to Bell Canada, but the consent decree forced the withdrawal of the US firm from the Canadian market. Western Electric gradually terminated its patent and information agreements with Northern Electric, out of fear that their liberal provisions would have to be extended to other North American firms. By 1962, AT&T and Western Electric had divested themselves of their holdings in Bell Canada and Northern Electric (Macdonald 2000). Cut off from its easy access to US patents and technical information, the primary sources for its product designs and development, Northern Electric realized that it needed to develop its own in-house R&D capacity to replace the designs previously licensed from Western Electric. It began the search for a location for the new research facility and, despite the fact that most of its manufacturing was in Montreal and southern Ontario; it eventually bought a substantial tract of land on the outskirts of Ottawa to be the home for Bell Northern Research. The main attraction of the

254


national capital region was the large concentration of scientists and engineers employed at the National Research Council laboratories and the Defense Research Board in technical areas of interest to Northern Electric. The Defense Research Board also offered training courses in advanced technologies, such as transistors and was authorized to transfer its technical innovations to firms in the private sector. Bell Northern Research recruited leading research scientists and engineers to its laboratories from outside the region and even the country—many of whom ultimately became leading entrepreneurs in the Ottawa telecommunications and photonics cluster (Chamberlin and de la Mothe 2003; Macdonald 2000). The central role played by inadvertent government policy in seeding the Ottawa cluster is paralleled to some extent in Maryann Feldman’s account of the emergence of the telecommunications cluster in the Washington–Baltimore corridor. Feldman’s analysis emphasizes the importance of entrepreneurship in stimulating the genesis of that cluster. She traces the roots of the entrepreneurial drive to the massive wave of downsizing and outsourcing that occurred in the US federal government in the late 1970s and 1980s. As a result of this trend, employment conditions in the federal public service became less secure and future prospects deteriorated. In the same period, public sector pay scales lagged behind those for executives in the private sector. An increased emphasis on outsourcing goods and services for the federal government provided a further inducement for prospective entrepreneurs to leave the government and start firms to supply goods and services back to their employer. Other policy initiatives launched in the early 1980s facilitated the licensing and transfer of technology from federal laboratories and provided further support for innovation in small businesses. ‘Enterprising scientists licensed technology out of their own university or government research labs to start new companies and chose to locate the new companies near their existing homes’ (Feldman 2001: 878). The strong concentration of federal research expertise in the nation’s capital established the research infrastructure for the growth of a cluster, as in the Ottawa case. Although cluster creation was not the primary goal of the federal government’s downsizing, the inadvertent role played by public policy in the formation of the cluster cannot be overlooked, together with an environment conducive to entrepreneurial initiatives. Similar dynamics can be discerned from the evolutionary paths of biotech regions. In their analysis of Boston–Cambridge, Massachusetts, and the San Francisco Bay Area, Owen-Smith and Powell (Chapter 4) argue that

255


these two leading biotech regions evolved along very different trajectories that continue to generate distinctive outputs. In the case of Boston, publicly funded research organizations (PROs—i.e. universities and hospitals) provided the initial knowledge base that led to subsequent commercial application. They suggest that academic rivalry between Harvard and MIT was an especially important characteristic of this regional network in its early years. Many of the Boston area biotech firms were founded by senior professors from these two institutions, most of whom kept their academic affiliations. With the passage of time, as the local network matured, these biotech firms developed a larger number of relations with both venture capitalists and other biotech firms. By the end of the 1990s, Genzyme and Biogen had developed a large number of linkages to other local biotech firms. Despite this late proliferation of firm-to-firm linkages, PROs, such as MIT, Harvard, and Massachusetts General Hospital, were still very important elements of the Boston network at the end of the 1990s. By contrast, biotech firms in the Bay Area exhibited strong ties to the local VC community from the earliest days of the industry’s development (late 1980s)—the original pattern being set when UCSF biochemist Howard Boyer partnered with venture capitalist Bob Swanson to establish Genentech in 1976. Owen-Smith and Powell attribute this pattern to ‘the prospecting and matchmaking efforts’ of the venture capitalist community, from which many firm founders emerged. To the extent that academic researchers were involved in firm start-ups, they tended to be at much earlier stages in their careers and were considerably more likely to leave their home institutions (either temporarily or permanently). In subsequent years, local network connections developed to include some linkages to PROs, but these were dwarfed by the rapid growth in linkages to other biotech firms. One of the densest local subnetworks developed around two lead firms established early in the region’s evolution: Genentech and Chiron. Given their less academic origins and closer links to the venture community, Bay Area firms tended to pursue more commercially focused, exploitative research along incremental trajectories. The larger implications of the work by Owen-Smith and Powell for our analysis are considerable. First it is clear that, despite the knowledgeintensive nature of biotechnology, the direct role played by universities in stimulating initial local development through spin-offs and commercialization can vary dramatically, even between two admittedly successful regions. Second, this finding also reminds us of the perils of reading off causal relations from spatial associations in ex post analysis of successful clusters: in the case of biotech at least, the local presence of Stanford,

256


UCSF, and UC Berkeley may not have been adequate, on its own, to have seeded the development of a world-leading biotech cluster in the Bay Area. The catalytic accelerator appears to have been the density of local VCs— itself a legacy of earlier rounds of venture-based industrial development focused on ICT-based firms. Restating this finding in terms of the language adopted above, the antecedent conditions are likely to vary dramatically between different biotech regions. On the basis of this analysis, OwenSmith and Powell strongly caution against the formulation of standard models of regional innovation-based success to guide policy intervention. Notwithstanding the critical role of a strong knowledge base in sciencedriven clusters, Romanelli and Feldman (Chapter 5) emphasize the strong connection between the entrepreneurial factor and cluster emergence. Entrepreneurs bring several key capabilities and assets to these processes that position them as key agents of cluster genesis and evolution. Foremost, they embody a creative spark—that is, an ability to identify viable new business opportunities amidst considerable uncertainty concerning technologies and markets. Successful long-term cluster growth also depends on local entrepreneurial firms’ ability to spin-off new secondgeneration firms at later stages in the cluster’s development. These dynamics have been especially visible in San Francisco, Boston, and San Diego, while New York, Los Angeles, and Washington, DC have been relatively less successful in generating second waves of entrepreneurial spin-offs, or in attracting entrepreneurs migrating into the region from elsewhere. Romanelli and Feldman attribute the relatively poor performance of the latter regions to their failure to generate a strong community of biotherapeutics entrepreneurs. In contrast, entrepreneurs in San Francisco, Boston, and San Diego were strongly bound together by histories of cross-institutional collaboration as well as common educational and research backgrounds.

Policy Implications Emphasizing the importance of chance events and the central role played by entrepreneurial initiative in the genesis of clusters does not eliminate the role for public policy. While it is virtually a commonplace to state that governments cannot create clusters by fiat or direct policy intervention, the preceding account of the evolutionary and path-dependent character of cluster genesis makes it clear that government policies play a critical role at many different stages of cluster formation and growth. It is important to be clear about the most valuable initiatives at the individual stages of

257


cluster development. The critical insight that the evolutionary perspective affords is that multiple locational outcomes are possible in the early stage of cluster formation, as shown by Scott (Chapter 2) and Kenney and Patton (Chapter 3). This potential makes it difficult, if not impossible for regional policymakers to target the development of specific clusters (Lambooy and Boschma 2001). Conversely, the importance of local antecedents for cluster development means that policy, across multiple levels of governance, can contribute to the accumulation of key assets in a specific location. As evident from several of the case studies in this volume, as well as previous studies, chance events that act as triggers for cluster formation or the entrepreneurial spark occur within a specific historic and geographic context. Frequently, it is public sector agencies that are critical in establishing the local antecedents that define this context. The public sector encompasses federal, state or provincial, and local governments; as well as public research institutes like Canada’s National Research Council or US government laboratories and institutions of higher education (although this would include leading private universities in the USA). In some instances, private sector research laboratories or contract research organizations can also lay the groundwork for the emergence of clusters, with strong support from public sector funding. While the ultimate impact of these policy interventions cannot be fully anticipated at the outset, over the long term, those policy interventions that strengthen the research and institutional infrastructure of a region or locality have the greatest potential to act as attractors for a cluster of firms (Wolfe and Gertler 2004). Public policies that create a strong knowledge base in the regional economy and contribute to the creation of a well-educated workforce establish the local antecedents that can support the emergence of clusters. While a strong research infrastructure and a thick labor market are distinctly local phenomena, in most industrial countries they are not exclusively the result of local, or even state and provincial government policies; the presence of the senior level of government lurks in the background. Several of the cases reviewed above underline the important roles played by different scales of political jurisdiction in the genesis of clusters. The literature on path dependency and divergent national trajectories, and the importance of culture, reinforces the point that national institutions shape the context for local development (Gertler 2002; Zysman 1994, 1996). Thus clusters can be seen as being nested within, and impacted by, other spatial scales of analysis, including regional and national innovation systems, each of which adds an important dimension

258


to the process of knowledge creation and diffusion that occurs within the cluster. The case of Silicon Valley clearly illustrates the way in which these differing scales of governance impact on the development of local clusters. The cluster exists within the distinctive features of the US system of innovation—with its unique system of laws, regulations, and conventions governing the operation of capital markets, forms of corporate governance, R&D, and other relevant factors. A number of these features are central to the story of Silicon Valley’s growth and development, including the highly decentralized nature of the postsecondary education system with complementary and interlocking roles for both the federal and state governments. The federal government played a central role as the initial customer for many of the early products of the cluster. For most of the 1960s, the US defense and space programs consumed the largest portion of the cluster’s output of integrated circuits. The US government was also the primary funder for much of the critical R&D that underpinned the growth of key segments of the computer and related industries in the cluster (National Research Council 1999a). Even in the celebrated case of Xerox’ Palo Alto Research Centre mentioned above, the initial staffing of key laboratories benefited immeasurably from the extensive research networks that had previously been developed through the Department of Defense’s Advanced Research Projects Agency (Hiltzik 1999). Once the cluster began to emerge in the 1960s and 1970s, institutional change—such as subsequent changes in capital gains tax rates, the tax treatment of stock options, and the rules governing investments in VC by pension funds— coevolved to further strengthen the cluster by facilitating the growth of a VC industry. As shown in this volume (Chapters 3 and 9), this seems to be a decisive step in the emergence of high-technology clusters. Hence, understanding the multiple factors that influence the development trajectory of a cluster and ultimately its economic performance is necessary. The other cases considered in this volume and in the previous literature provide equally clear evidence of the critical contribution made by policy interventions from all three levels to the genesis and growth of the clusters studied. In the case of the Capitol region in the USA, the dense concentration of federal laboratories constituted the breeding grounds for a whole new generation of entrepreneurs in the telecom and biotech sectors. However, a series of federal policy interventions in the early 1980s, in response to the perceived decline in the competitiveness of the US economy, reduced the barriers and increased the incentives for nascent entrepreneurs to exploit the commercial potential of intellectual property

259


generated by public research funding in these government laboratories (National Research Council 2003). The passage of the Federal Technology Transfer Act in the USA in 1986, which stimulated the creation of Cooperative Research and Development Agreements (CRADA) allowed federal agencies to partner with small firms to develop new technologies (Guston 1998). A large number of the biotech firms that emerged in the Capitol region in this period were the product of CRADAs with government laboratories. Finally, the introduction of the Small Business Innovation Research Program in 1982 that set aside a certain portion of federal R&D funding for small business provided a critical source of funding for small business start-ups in the Capitol region, with local firms receiving a significant proportion of funding under this program (Feldman and Francis 2003a, b; National Research Council 1999b). Collectively, this major shift in US policy at the federal level in the early and mid-1980s provided a powerful impetus for capitalizing on the crucial knowledge base in the research infrastructure of the Washington, DC region and stimulating the entrepreneurial impulse in the cluster. In the Canadian cases considered above, the role of the federal and provincial governments in building the local research infrastructure and building up the resources of highly skilled labor was equally critical. In the Waterloo, Ontario case, the mobilization by local business leaders to secure a charter for a new university, financed with federal and provincial funding, and their foresightedness in structuring a curriculum around math, sciences, and engineering and creating a pioneering program of cooperative education, all laid the groundwork for the future emergence of a strong information technology cluster. In this case, it was the specific pattern of interaction of dynamic, visionary leaders at the community level, with the increase in combined federal and provincial funding for postsecondary education that strengthened the local antecedents essential for the emergence of the information technology cluster. In the case of Ottawa, the Canadian capital, the dense concentration of federal government laboratories in telecommunications served as the magnet that drew Northern Electric’s primary research facility to the region. This is corroborated by the case studies in Chapters 6–9 on the emergence of biotechnology clusters in China and Denmark/Sweden, and of the ICT clusters in Ireland and Israel, which all points to the crucial role played by policymakers. In some cases, policies have intervened quite strongly—particularly in China, Ireland, and Israel—in order to build institutions and markets, whereas more general policies have been pursued in other cases (e.g. Sweden and Denmark). Orsenigo emphasizes in

260


his survey in Chapter 10 that policy matters, while simulation exercises in the following Chapter 11 stress the type of policy required to foster cluster emergence. When it comes to increasing entrepreneurial activity, the policy instruments are far less clear. As some of the cases discussed above illustrate, government policy played a critical role in stimulating the genesis of the cluster, but often in a completely inadvertent manner. In these instances, governments were pursuing policies designed to achieve other goals, but the consequences of the policy triggered the kind of chance occurrence that path dependency describes. It seems to be a key element of cluster development in the third and last stage of cluster emergence, particularly to provide an environment conducive to the entering of second and third generation start-up firms. At this stage government policies to sustain the entrepreneurial drive are perhaps the most important. These can include a broad range of government policies to support upgrading the innovative capacities of firms and promote the rapid diffusion of technologies, networks to foster greater interaction among the emerging SME’s, as well as providing much needed mentoring programs for newly minted entrepreneurs. Often local high-technology industry associations emerge with support from local and regional government agencies to play this role. A key barrier that a rapidly growing cluster often runs up against is an adequate supply of the critical skills needed to feed the growing firms. This is a policy area where local universities and colleges have played a crucial role, often with the backing of state and provincial governments, in expanding training and research programs in the areas of most crucial need. The formation of angel networks and the attraction of VC into the locality can also be supported by appropriate government policies (Feldman, Francis, and Bercovitz 2005; Porter et al. 2001). In general though, government policy at the third stage of cluster formation and development is much more varied and is often tailored to meet the needs of the specific region and locality in which the cluster is located. Furthermore, means to evaluate and strengthen the policy supports for cluster development at this stage is crucial, for example through strategic planning or innovation-based strategic planning at the regional level. The strategic planning process is valuable for helping regions develop a shared understanding of their local assets and identifying the area’s unique local characteristics that support the development of regional industry clusters. These include knowledge economy assets (such as workforce skills, knowledge and research development, creativity, advanced telecommunications infrastructure, quality of place, and financial capital), collaborative

261


institutions and organizations (such as regional development organizations, professional networks, research consortia, and entrepreneurial support networks), and the regional mindset (values and attitudes that encourage innovation, entrepreneurship, and collaboration). Strategic planning exercises have also been used to identify key gaps in the region’s mix of assets as well as common opportunities that may be exploited by its existing or emerging clusters. The common framework for understanding the region’s potential and the shared vision generated through such a planning exercise can also help mobilize support at the local level for key activities needed to boost the cluster (Gertler and Wolfe 2004; Porter et al. 2001).

Conclusions The overall lesson extracted from the case studies and the previous literature is that the path dependencies for cluster creation are highly variable and that the chance events which provide the trigger for cluster formation can come from many sources. There is a strong element of serendipity in virtually all of the cases described above and any policy analyst or cluster consultant that would try to design a formula for cluster growth on the basis of these lessons would be wildly optimistic, to say the least. However, virtually all of the cases strongly reinforce the point made concerning the intersection of historic context and chance occurrence in launching a regional or local economy along a certain trajectory of development. Public sector involvement can affect cluster trajectories in a variety of ways, though the impacts are often unpredictable and even, in some instances, unintended. Whether intentional or inadvertent, one of the most effective public policies for seeding cluster development is a sound investment in building the research infrastructure and educated labor base in a region. The establishment of a strong local talent pool of highly skilled and knowledgeable workers both feeds the growth of the local entrepreneurship in the cluster as increasing returns begin to take hold, and attracts outside firms and entrepreneurs to the cluster to gain access to the local buzz. Similarly, the ability, or inability, of the local or regional economy to develop the underlying conditions of trust and social capital that contribute to the presence of a learning economy may create a condition of lockin to a specific innovation trajectory. A related question that needs to be explored is whether the conditions that can provide a supportive culture and institutional framework for a specific regional or local economy can

262


also be influenced by direct intervention, and if so, how effectively. The presence, or absence, of key institutional elements of the local or regional innovation system may affect both their innovative capacity and their potential to serve as nodes for cluster development. Many clusters enjoy the knowledge assets and research infrastructure that are necessary for the development of an innovation-based development strategy, but they differ dramatically in their capacity to mobilize these assets in the pursuit of such a strategy. Experience demonstrates that local communities can formulate strategies to alter their economic trajectory and improve their chances of economic development. The successful initiation of this kind of process depends on the ability to collaborate across boundaries—both geographic and social.

263

13 The Role of Public Policy in Emerging Clusters Bo Carlsson

A common feature in all the preceding chapters is that cluster formation is not a single event; it is a sequential process spread out over several decades. The process involves an early stage in which a regional agglomeration of activity in a certain sector is formed, usually quite gradually and, in successful cases, a second stage in which a viable industry cluster emerges and grows rapidly. As revealed by the case studies, the first phase can happen either spontaneously, even randomly or by chance, or as a result of deliberate public policy intervention. The development in the early phase then conditions the development in the later phase, that is, path dependence sets in. As emphasized in the case studies, the second phase is characterized by entrepreneurial activities. Vigorous entrepreneurial activity is what distinguishes the successful clusters that grow rapidly from the less successful ones that stagnate or fail to develop. Thus, some triggering event coupled with an entrepreneurial spark seems necessary in order for industry clusters to emerge and enter a sustainable growth trajectory. What is the role of public policy in these processes? If serendipity is the dominant factor in determining whether clusters will fail or succeed, there seems to be limited scope for policies to impact the process. Or is it rather that serendipity is a conspicuous feature of the early phases, but that policy plays an increasingly important role in the subsequent stages of cluster emergence and dynamics? Are different policies required at different stages of cluster emergence? In particular, should the first phase policies focus on widening the opportunity space conducive to cluster

264

Role of Public Policy in Emerging Clusters

emergence, whereas the role of second stage policies is to support and reinforce the emergence of clusters? The next step is to consider some of the empirical regularities that emerge from the cases presented in this volume and to discuss the role of public policy. The challenge is to then integrate these findings into a coherent framework for policy analysis. The concluding section then discusses the functions of public policy with respect to cluster formation.

Empirical Regularities The key concepts in the early phase are path dependence and knowledge spillovers, whereas the later phase is characterized by an entrepreneurial spark and a triggering event. Each of these is discussed in turn. In most of the cases we have discerned strong influences of path dependence and knowledge spillovers. This suggests the key role of competence and technology. Path dependence can be both a positive and a negative factor. Preexisting competence in the same or related industries can help, especially in cases of incremental innovation, but can also block radical innovation (technological discontinuities). The spillover mechanisms are also different for incumbents than new entrants. The most important aspect of path dependence may be the existing entrepreneurial climate resulting from preexisting conditions. [E]ntrants and incumbents have different capacities to absorb spillovers originating in different sectors. This means that, sometimes, technological proximity does not make the absorption of spillovers easier for incumbents, giving rise to opportunities for the successful entry of new competitors. These successful entrants might come from the same sector in the industry, but are more likely to originate in others. Although most technological change is incremental and cumulative, building on the established knowledge base, . . . this process is punctuated by technological discontinuities—breakthroughs that result in the emergence of new technological paradigms . . . These technological discontinuities, resulting from radical innovation, can lead to the loss of competitiveness by users of technologies that draw upon the old science base, favoring attacks from prospective entrants . . . [E]stablished firms are likely to dominate incremental innovation, while entrants are likely to dominate radical innovation (Baptista and Swann 1999: 395).

In each of the cases analyzed in this volume, path dependence has played a critical role. In the motion-picture industry, Hollywood was one of several

265


possible locations, but once a New York–based company came to Los Angeles for a winter haven and then established a permanent studio there, other companies followed. Within a few years a self-sustaining agglomeration had been created. The story is quite similar in the automobile industry: the prior existence of related industries (carriages and wagons, bicycles, and engines) in Detroit provided the technical competence needed for the auto industry (Klepper 2003). As demonstrated by Klepper (2004), the key was the experience of the people starting new firms. He found that firms founded by people with experience either in related industries (carriages and wagons, bicycles, and engines) or in incumbent firms in the auto industry performed better than firms founded by inexperienced entrepreneurs. A particularly prominent incumbent was the Olds Motor Works whose spinoffs included Dodge, Cadillac, and Ford. Thus, Olds played much the same role as Fairchild did later in Silicon Valley: it was a repository of industrial competence and source of several spin-offs that formed the core of the new industry. Detroit had no advantage per se for hosting automobile production as compared to other cities. The experience and competence of founders in the initial phase spilled over to the second generation of entrepreneurs who spearheaded the rapid growth in the later phase. The emergence of Silicon Valley in many ways resembles that of Route 128 in Boston (Bresnahan, Gambardella, and Saxenian 2001; Dorfman 1983; Kenney 2000; Saxenian 1994). One similarity is that a key leader in a great university located in the region played a major role in creating favorable conditions for a regional agglomeration, which proved invaluable as a foundation of an electronics-based industry cluster. That key leader was Frederick Terman, dean of the Engineering School at Stanford University. Terman was actually a disciple of Vannevar Bush at MIT and had gained first-hand exposure to the close ties between MIT and industry.1 He turned down a faculty position at MIT to return to his native California, largely for health reasons. Among the many things that Terman achieved (see Roberts 1991), he promoted the development of Stanford Industrial Park, where several former students (William Hewlett and David Packard and the Varian brothers, to mention a few) started their high-technology firms. In Silicon Valley, the efforts of Fred Terman and Stanford University had created sufficiently favorable conditions for the semiconductor industry to grow out of the regional agglomeration that was forming, but the technical and scientific ideas came from the East Coast. They could have been exploited elsewhere, but once the scientists and engineers happened

266


to locate in Silicon Valley, the technological opportunities and the spillovers they generated were sufficiently large to create enormous success. Again, a parallel can be found in the Route 128 case. The policies of MIT to promote entrepreneurship and to build research labs to conduct federally sponsored research, as well as the coupling of these via huge federal defense spending, laid the foundation for the electronics industry. Instead, the proliferation of firms in Silicon Valley came from multiple spin-offs of other companies, especially Fairchild. Only eight out of 243 new technical firms studied in the Palo Alto area have their origins in Stanford University . . . , probably due in part to Stanford’s lack of major government-sponsored laboratories. Indeed, despite the distance from their alma mater, MIT alumni are surprisingly the founders of over 175 companies in northern California, accounting for 21 percent of the manufacturing employment in Silicon Valley (Roberts 1991: 35).

The environment that existed in the late 1950s in what later became known as Silicon Valley was not unique; similar conditions existed in Boston and New York, for example. And indeed, none of the scientists and engineers that led the cluster formation in Silicon Valley came from the area; they were all ‘imported’ from the East Coast of the United States. The reason they located in the Bay Area was incidental; for example, William Shockley, one of the coinventors of the transistor at Bell Labs in New Jersey, wanted to be near his mother (Moore and Davis 2001). Thus, the early evolution of Silicon Valley mirrors what happened in the motion-picture industry; there were alternative sites, random events influenced the process, and an inflow of entrepreneurs set forces in motion that created strong path dependence. In Israel, the need to scale down the defense research establishment while absorbing a large number of Russian immigrants, many of whom were scientists and engineers, led to a reorientation of R&D toward industrial and commercial development. This paved the way for the subsequent implementation of the Yozma Venture Fund program. In Ireland, a whole range of policy changes in conjunction with the Irish entry into the EU, especially the efforts to attract FDI, set the stage for the emergence of ICT clusters. In the later phase, the competence of foreign firms spilled over to indigenous firms via new firm formation. In the US Capitol region, the biotechnology competence in federal agencies and the number of entrepreneurial experiments were sufficiently large to overcome the lack of preexisting conditions favorable to entrepreneurial activities. What triggered the formation of the cluster was a change

267


in federal government policies involving downsizing of federal agencies, creating slack and surplus resources that could find new and more productive uses (Feldman 2001: 886). Thus, in the period of 1973–80, 20 percent of the start-up entrepreneurs came from the National Institutes of Health (NIH). Between 1981 and 1985, one-third of the entrepreneurs came from the NIH. But the majority of the entrepreneurs came from private laboratories or companies (65 percent in the combined period, 1973–85). Between 1986 and 1995, approximately half of the biotechnology entrepreneurs came from government agencies such as the NIH and the Walter Reed Army Institute for Research as well as the local universities. Similar to the cluster development in other regions, competence and path dependence were crucial. Many of the entrepreneurial ventures can be viewed as spin-offs from incumbents (in this case, government agencies). Incumbent firms tend to diversify into related industries; founders of new firms typically have experience in closely related industries; and entrants do not stray far geographically from their origins. One of the conditions favoring new firm formation in the Capitol region was actually the absence of established large pharmaceutical companies that could absorb the skilled labor released from federal labs.2 Once a critical mass of activity had been reached, VC was attracted from other parts of the nation. Thus, similarly to the development in Silicon Valley, VC lagged behind the cluster formation: VC firms were attracted to the new cluster once there was substantial economic activity with the expectation of future profits: In conclusion, entrepreneurship in the region was a response to exogenous factors: underemployed skilled labor brought about by changes in federal employment policy coupled with new opportunities for the private sector to contract with the federal government and commercialize new technologies. Most importantly, entrepreneurship picked up momentum in the cluster and generations of new firms spun-off from the earliest start-ups. Entrepreneurs who cashed in from one venture created other companies. Entrepreneurs also lobbied for government resources and worked to change the stance of local universities. As entrepreneurship caught hold, the cluster emerged and the familiar virtuous, self-sustaining cycles appear to be in place (Feldman and Francis 2003a. 2003b: 784).

It is interesting to compare the successful outcome of the Capitol region with the nearby Research Triangle Park in North Carolina. The idea of a research park emerged in the early postwar period when North Carolina faced the need to shift from a largely agriculturally based economy to new

268


industries. But as pointed out by Link (1995), it is hard to identify a single person who came up with the idea of building Research Triangle Park; rather, it was shifting groups of two or three people, including a land developer/construction contractor, a governor, university presidents, deans, and individual faculty members, as well as economic development people at the state and local levels. The basic concept was to exploit the location of three existing universities (Duke University, University of North Carolina, and North Carolina State University) in order to attract research companies to the area, not to participate in those companies’ research efforts (Link 1995). Similarly to the Capitol region, no state funding was involved; only verbal encouragement and support from the governor’s office while financial resources came from private individuals and corporations. However, there is little evidence as yet in Research Triangle Park of local entrepreneurial activity and few ties between entrepreneurship and the three major universities in the area (Roberts 1991). There is no prime mover or champion, and little or no knowledge spillover from one organization or firm to another. The vast majority of companies in the region are subsidiaries or spin-offs of large companies located elsewhere. Still, this may be changing: Research Triangle Park has received substantial VC funding in recent years. Thus, it is possible that we are currently witnessing the takeoff of Research Triangle Park, that is, its transformation from a regional agglomeration into an industry cluster. In China, a series of institutional and policy reforms created the conditions needed for transition to rapid development of biotechnology clusters. Prevezer and Tang (Chapter 6) distinguish two main phases in the evolution of government policy with respect to these clusters. In the initial phase, policies were ‘mainly concerned with sowing the seeds of institutional reform, creating new forms of property rights, setting out strategic programs for the development of biotechnology and other hightechnology sectors.’ This took place in the 1980s up to the mid-1990s when the second phase started which was ‘more focused on incentives to assist start-ups, on attracting potential entrepreneurs back to China from abroad, and on developing regional clusters around science parks’ (Prevezer and Tang: 127). Partly as a result of these policies, the entry of companies into the biotech clusters in each region more than doubled during the period 1995–2003 compared to the period before 1995. While most of the entries in the pre-1995 period were government initiated, many new companies formed in the latter period were founded by scientists returning from abroad. The factors influencing their locational decisions were primarily personal—such as the place where they used to

269


live or study. But also the local network of connections played a key role in the choice of location for these biotech firms, and the support they received from the government. The previous history of each region has also played a role. Beijing has been the traditional science and technology center of China and is particularly preeminent in biotechnology research. In 1999 there were sixtyfour universities in Beijing, compared with forty-one in Shanghai and only two in Shenzhen, the other two regions included in the study. The most distinguishing feature in Shanghai is the broad industrial base and in Shenzhen it is the more open and market-oriented government and the proximity to Hong Kong. Those differences were also reflected in the characteristics of the respective agglomeration. In particular, the cluster in Beijing is influenced by a strong research tradition and relatively close links to the government bureaucracy, whereas Shanghai is characterized by a strong research base, relaxation of bureaucratic procedures, and a vibrant business environment. The third cluster in Shenzhen is the least research intensive and is specialized in production of high-technology products where the relatively liberal local policy, access to capital, and strong international networks constitute the major regional advantages. A similar picture emerges for the Nordic biotechnology cluster Medicon Valley, albeit in a completely different setting. The leading universities go back 500 years (the University of Copenhagen was established in 1479 and the University of Lund in 1666) and have, over the centuries, attracted the most talented individuals in Denmark and Sweden. Thus, the knowledge was in place early even though the diffusion of knowledge outside the universities seems to have been modest until recently (see Braunerhjelm and Helgesson Chapter 7). This suggests that the emergence of clusters is associated primarily with competence and to a lesser extent with geography and certain locations. Location does play a role, not least because entrepreneurship is a local activity. However, as evident from the case studies, an inflow of entrepreneurs and skills from other regions may exert a decisive impact on the emergence of clusters. In order for an industry cluster to go from an initial formative phase to a sustainable growth phase, some triggering event coupled with an entrepreneurial spark is needed. Such triggering events may originate from a wide variety of sources: in the Hollywood motion-picture industry, it was the organizational innovations by Ince (a New York motion-picture company executive who relocated to Southern California and reorganized the whole film-making process) that led to increased division of labor and

270


economies of scale, industrializing the whole film-making process and stimulating new growth and new firm formation. In the Silicon Valley electronics cluster, it was the departure of the ‘traitorous eight’ from Shockley Semiconductor to form Fairchild and subsequently start their own individual companies that established the core of the cluster and stimulated new entrepreneurial activity. In the US Capitol region, the downsizing of the federal government led to the release and redeployment of technical talent in new biotechnology enterprises. In Israel, the formation of the Yozma program was the triggering event that launched a vigorous VC industry cluster. In Ireland it was the combination of tax policy and the development of the Single European Market that attracted major FDI, particularly in ICT. In the development of regional biotechnology clusters in China, government entrepreneurship combined with repatriating Chinese scientists and engineers from abroad provided the trigger. It is noteworthy that in every one of these cases, the entrepreneurial activity stimulated by these triggering events was led by people who had experience in similar activities in previously existing organizations. Ince came from a motion-picture company in New York; Shockley brought the microprocessor technology from Bell Labs, and the ‘traitorous eight’ gained both technical and management experience at Shockley Semiconductor and Fairchild before starting their own companies; the entrepreneurs in the US Capitol region had experience from the NIH and similar entities; the Yozma program in Israel essentially leveraged private venture capitalists (both foreign and domestic); the Irish ICT cluster was built both directly and indirectly on foreign firms; and the Chinese biotech clusters relied heavily on attracting talent from abroad. Another striking common feature in several of these cases is that once a triggering event had occurred, finance followed. In the pre–World War II cases, wealthy individuals were attracted by the new opportunities being formed. After World War II, the supply of private investments became more formalized, organized first in the Route 128 case as American Research and Development (ARD), the first VC fund (Roberts 1991: 33–4). In the formation of Silicon Valley and later clusters, venture capitalists as we now know them have played a crucial role. This further suggests the importance of an entrepreneurial culture. The analysis of the Swedish–Danish biotechnology cluster Medicon Valley attributes the comparatively stronger dynamics in the Danish part of the cluster to a more deeply rooted entrepreneurial culture in Denmark than Sweden (Braunerhjelm and Helgesson Chapter 7). Maggioni (Chapter 11)

271


makes a similar point: while infrastructure is important in the initial development of a cluster, its long-term performance is determined by whether it generates a sustained rate of formation of small indigenous firms. What seems to be required is not so much a certain ‘critical mass’ or density of activity—although that certainly helps—as an entrepreneurial spark in combination with competence in the form of entrepreneurial, organizational, and management know-how to convert technological opportunities arising from technical and scientific discoveries into successful business ventures. Knowledge entails a strong generic component that is an essential part of path dependence.

The Role of Public Policy Public policy has played an important role in all but two of the cases studied or referred to here. The two exceptions are Detroit and Hollywood. However, the role has often been inadvertent in the sense that the formation of a regional agglomeration or industry cluster was not the primary policy target. Examples are defense funding in Silicon Valley and funding for biotechnology research in the US Capitol region. The policy was intentional but not sector-specific in two cases (Ireland and Research Triangle Park) and both intentional and sector-specific in at least two cases (Israel and China). In the USA, federal government policies have been important but not targeted on industry clusters, and state and local government policies have played supportive but much more mundane roles, largely involving infrastructure, and mostly in response to rather than in anticipation of emerging industry clusters. The review of case studies in the preceding pages also suggests that there are at least some, but probably not many, historical cases of successful industry cluster formation as a result of proactive and targeted public policy intervention; the problem is simply too complex. What is at issue here is a systems cum evolutionary approach to policymaking. At a minimum, the formation of a successful industry cluster requires avoiding systemic obstacles and overcoming system failures. As shown by Avnimelech and Teubal in Chapter 10, there are many conceivable causes of system failures. Thus, policymaking in a complex, nondeterministic world is an extremely difficult art: How do you make policies when the desirable outcome lies decades down the road and cannot be specified? One implication is that too targeted policies carry obvious risks, which could lead to lock-in with long-run negative effects. Rather, a policy

272


discussion has to focus on the functional requirements that must be fulfilled if an industry cluster is to evolve and perform well.3 Each of these functions can be related to well-defined policies in terms of providing resources, the structural features in the cluster, that is, the characteristics of actors, networks, and institutions. Sometimes the functions are fulfilled by private actors, sometimes through fortuitous government policies, and sometimes through proactive and intentional public policies.

Functions of Public Policy This section identifies six functions that public policy can fulfill in order to promote cluster genesis. It also examines the industry clusters with respect to each function. First, effective public policy can ascertain the existence of a sufficient knowledge base. This refers to all the scientific, technical, and practical knowledge related to all activities within the cluster. In Silicon Valley, the transistor technology was diffused by forcing Bell Labs to disseminate the technology to the scientific community. In the US Capitol region, the know-how existed in federal agencies and labs and was commercialized via new start-up companies. In Israel, a lot of the technical knowledge came from the military and Russian immigrants and industry knowledge from foreign venture capitalists. In Ireland, most of the knowledge came via FDI and in China from both government labs and scientists returning from abroad.4 It is interesting to note the different mechanisms by which the biotechnology clusters in Boston and the San Francisco Bay Area acquired knowledge (see Chapter 4 by Owen-Smith and Powell). In Boston, the biotechnology companies were rooted in Public Research Organizations (PROs), such as Harvard and MIT, whereas the Bay Area was dominated by venture capitalists. Both clusters have been successful, but their networks and trajectories over time are quite different. Feldman and Martin (2005) propose the idea of coherent jurisdictional activity sets as a means to analyze cluster advantage and to ascertain how different knowledge bases build competitive advantage for a cluster. Second, another critical function of public policy is to create transparent incentives to reinforce positive forces or to overcome negative forces. This may take the form of reducing uncertainty, providing legitimacy, and increasing the opportunity space for entrepreneurs to engage in profitable ventures. In the case of Silicon Valley, technological opportunities

273


coincided with strong economic incentives. The biotechnology cluster in the US Capitol region shows that sometimes negative incentives (such as layoffs by Federal agencies) can spur the formation of new entities in related businesses—provided that potential entrepreneurs are embedded in an environment conducive to entrepreneurial activities. In Israel, the setting up of VC Funds with partial but substantial government support created vastly improved funding possibilities for potential entrepreneurs. Third, effective public policy may promote entrepreneurial experiments. In some cases technical opportunities when identified provide sufficient profit incentives for new firm formation. Existing firms may be stimulated to innovate and new firms may be encouraged to emerge when the entrepreneurial climate is sufficiently favorable (compare the Danish–Swedish Medicon Valley case). When such activities are lacking, public policies may help to promote entrepreneurial experiments. It is interesting to note that the promotion of entrepreneurship has been an important component of public policy in the three non-US regions studied but not generally in the USA (although Research Triangle Park and efforts in many states to support new business creation demonstrate that such policies exist also in the USA). Fourth, government may create markets or guarantee appropriate market conditions. In Silicon Valley and in the Detroit auto industry, new start-ups were able to create new markets on their own. In other cases, governmental policies created new markets for domestic firms. Such policies encompass both micro-oriented measures (e.g. IPR protection) and macropolicies and institutional changes such as joining a regional integration process. In Ireland, Israel, and China, institutional and policy reforms to improve the market conditions have been extremely important. Fifth, government policy may create resources or augment resource creation. Lack of financial resources as well as technical and legal services may impede the formation of new industry clusters. Sometimes such resources coevolve with the business opportunities in the cluster (even if they emerge after the cluster is initially formed, as happened, for example with respect to VC in Silicon Valley and the US Capitol region). But sometimes the injection of new resources can help new clusters to form, as happened in Route 128 as a result of federal defense spending during World War II and in the early postwar period. As noted by Kenney and Patton in Chapter 2, the needs for supporting services such as legal and financial services vary among industries. They suggest that mobilizing local law firms to support the entrepreneurial process is particularly important, since they are the single most localized support network actors.

274


Sixth, government can promote positive externalities. The formation of a successful industry cluster involves positive externalities within the system, such as economies in the form of common labor markets, an elaborated division of labor, and knowledge spillovers (cf. Romanelli and Feldman, Chapter 5 in this volume). Moore and Davis (2001) argue that it was the government forcing AT&T to share their findings on semiconductors freely (as part of an antitrust settlement) that represented the true birth of the semiconductor industry in Silicon Valley. And as shown by Owen-Smith and Powell, the different structures of the networks in biotechnology in Boston and the San Francisco Bay Area led to the seeming paradox that deliberate efforts by PROs in the Boston area to control information flows resulted in more open structures than the more informal approaches in the Bay Area which led to more tightly controlled and commercially oriented networks. Each of the six functions outlined above relates to one or more resource requirements that must be satisfied for a system or cluster to develop successfully. There are six resource requirements:5 A sufficient knowledge base, financial and human resources, market identification by firms, a sufficient number and variety of actors (including one or several leaders or prime movers), networks among various types of actors (industry– government, industry–university, industry–industry, professional networks, and others), and institutions (market conditions, regulations, supporting organizations, etc.). The problem for the policymaker then becomes twofold: (a) which, if any, potential industry clusters to support and (b) which of these requirements to target for the supported cluster(s). Given the uncertainty associated with the future, especially the distant future, this is an exceedingly difficult task. As pointed out already by Lindblom (1959), [p]olicy-making is a process of successive approximation to some desired objectives in which what is desired itself continues to change under reconsideration . . . Making policy is at best a very rough process. Neither social scientists, nor politicians, nor public administrators yet know enough about the social world to avoid repeated error in predicting the consequences of policy moves. A wise policy-maker consequently expects that his policies will achieve only part of what he hopes and at the same time will produce unanticipated consequences he would have preferred to avoid. If he proceeds through a succession of incremental changes, he avoids serious lasting mistakes in several ways (Lindblom 1959: 86).

Thefirstpolicyissue—thatofdeterminingwhichclustertosupport—isatypical political problem and is usually solved in the normal political process; it is not an optimization problem. Once an industry has been chosen, the second

275


issue—concerning which requirement to target—becomes relevant. It is complicated by the fact that the contours of the industry are fuzzy at best and will probably change over time. This means that both the targets and the policies will change over time. The best approach may be to look at the cluster formation process as a whole and try to identify which of the resource requirements constitutes the greatest bottleneck and then target that requirement. An analysis of the resource requirements could make use of a checklist of the following sort. With respect to the existing knowledge base, some examples are: What type of knowledge exists in the system currently (scientific knowledge and know-how relating to systems, materials, components, production, design, etc.)? Are there important gaps? Who are the carriers of knowledge, how many, how diverse, and how well connected? Is the whole value chain represented? How does this system compare to others elsewhere? Is the knowledge base expected to change in the next decade? With respect to other resources, how much financial capital is available (seed, angel, and VC, as well as other risk capital)? How many and how diverse are the sources of finance? Similar questions can be raised with respect to each functional requirement. The policymaker will then have to decide where the greatest needs exist and develop policies to address those needs. But the existence of a particular need does not necessarily require public policy involvement. Various types of organizations can fulfill each of the six functions. For example, financial resources can be provided by venture capitalists, business angels, universities, customers, government bodies, family and friends as well as financial institutions. Also the various functions outlined above are not independent of each other, but linked. There are feedbacks and interdependencies among the various functions. For instance, if a new firm enters the system, it may bring both competence and financial resources to the system. It may thus improve the probability of achieving critical mass or legitimacy of the new cluster or system and induce a greater division of labor that may reduce barriers to entry. The entry of that particular firm may then induce subsequent entries, which bring yet more resources to the system, etc. Sometimes, therefore, powerful feedback loops may occur between functions so that the evolution of the system becomes self-reinforcing. The setting up and subsequent privatization of the Yozma funds in Israel is an example of public policy creating favorable initial conditions and then withdrawing once the process had become self-sustaining. Whereas all the functions need to be filled in creating a new system or cluster, systems vary in how they fill these functions. The functional

276


pattern, that is how they are performed can be ascertained, as can the existence and nature of feedback loops, at least ex post. Hence, a functional analysis may act as a ‘filter’ or guidance to policymakers as to where the ‘bottlenecks’ or critical points may be for intervention. A ‘process goal’ for policy can be formulated in terms of a desirable functional pattern, rather than in terms of the final outcome, that is, a particular rate of diffusion, or growth. Individual policies, or instruments of intervention, can then be evaluated in terms of how they are expected to influence the functional pattern via shaping the factors that drive or block the functions. While all of these functions are necessary for the formation of a successful industry cluster or innovation system, they may not be sufficient. Public policy can support and sometimes even initiate a cluster in its early phase—although spontaneous development and serendipity seem to be more prevalent mechanisms. A more frequent role of public policy is to provide support and reinforcement in the later phase of cluster formation. As shown in the preceding analysis, the policy requirements in each phase vary a great deal both over time and between clusters. No single policy is universally applicable, and even sustained and substantial policy efforts may fail to yield the desired results. In the end, success or failure depends on the creativity and persistence of the entrepreneur—with an element of luck—as well.

Notes 1. Vannevar Bush, an electrical engineering professor at MIT, helped start the company in 1925 that later became Raytheon. In 1932 he was appointed vice president and dean at MIT. As the Second World War began he was recruited by President Roosevelt to head up the newly formed Office of Scientific Research and Development (ORSD), the first federal agency dedicated to science and research. 2. As pointed out by Owen-Smith and Powell (see Chapter 4 in this volume), large pharmaceutical companies were also absent in the Boston and San Francisco Bay Area when the biotech clusters were forming there. Also, as shown by Romanelli and Feldman (Chapter 5), the existence in New York of large pharmaceutical companies may actually have stifled the development of the biotechnology cluster there. 3. In current joint work (Carlsson, Jacobsson, and Bergek 2005), Staffan Jacobsson, Anna Bergek and I have identified these functional requirements, based on 15 years’ work on innovation systems. This work is summarized in Carlsson (1995, 1997, 2002).

277

Role of Public Policy in Emerging Clusters 4. Regarding other clusters touched upon but not really analyzed here, similar stories can be told. In the Detroit auto industry, knowledge was transferred from pre-existing related industries and in the Hollywood motion picture industry from company headquarters elsewhere. In Route 128, the results of military research spilled over into the commercial arena. In Research Triangle Park, the knowledge created does not yet seem to have been commercialized locally. 5. A ‘checklist’ can be developed for each resource requirement. See Carlsson, Jacobsson, and Bergek (2005) for details.

278

Bibliography

Abbate, J. (1999). Inventing the Internet. Cambridge, MA: Massachusetts Institute of Technology Press. Abernathy, W. J. (1978). The Productivity Dilemma. Baltimore, MD: Johns Hopkins University Press. —— and Utterback, J. M. (1978). ‘Patterns of Industrial Innovation’, Technology Review, 80(June–July): 40–7, 56. Acs, Z. J., Audretsch, D. B., and Feldman, M. P. (1992). ‘Real Effects of Academic Research: A Comment’, American Economic Review, 82: 363–7. —— —— —— (1994). ‘R&D Spillovers and Recipient Firm Size’, Review of Economic and Statistics, 76: 336–40. —— FitzRoy F. R., and Smith, I. (2002). ‘High-Technology Employment and R&D in Cities: Heterogeneity vs Specialization’, Annals of Regional Science, 36(3): 373–86. Aharonson, B. S. (2004). ‘Institutional Intervention and Localization in the Canadian Biotechnology Industry: Foundings and Exits’, Working Paper. —— Baum, J. A. C., and Feldman, M. P. (2004a). ‘Industrial Clustering and the Returns to Inventive Activity: Canadian Biotechnology Firms, 1991–2000’, DRUID Working Paper, 3. —— —— —— (2004b). ‘Industrial Clustering and the Returns to Inventive Activity: Canadian Biotechnology Firms, 1991–2000’, Danish Research Unit for Industrial Dynamics Working Paper, 04/03. Ahuja, G. (2000). ‘Collaboration Networks. Structural Holes and Innovation: A Longitudinal Study’, Administrative Science Quarterly, 45: 425–55. Aldrich, H. E. and Zimmer, C. (1986). ‘Entrepreneurship through Social Networks’, in D. L. Sexton and R. W. Smilor (eds.), The Art and Science of Entrepreneurship. Cambridge, MA: Ballinger. Allansdottir, A., Bonaccorsi, A., Gambardella, A., Mariani, M., Orsenigo, L., Pammoli, F., and Riccaboni, M. (2001). Innovation and Competitiveness in European Biotechnology. Brussels: Enterprise Directorate-General European Commission, European Communities. Allen, J. T. (1976). ‘The Decay of the Motion-Picture Patents Company’, in T. Balio (ed.), The American Film Industry. Madison, WI: University of Wisconsin Press.

279

Bibliography Almeida, P. and Kogut, B. (1999). ‘Localization of Knowledge and the Mobility of Engineers in Regional Networks’, Management Science, 45: 905–17. Almor, T. and Hashai, N. (2003). ‘Competitive Advantage and Strategic Configuration of Knowledge-Intensive, ‘Born Global’ Firms: A Modified Resource Based View’, European International Business Academy Annual Conference, Copenhagen, Denmark, December. Altshuler, R., Grubert, H., and Newlon, T. S. (2001). ‘Has US Investment Abroad Become More Sensitive to Tax Rates?’, in J. R. Hines Jr (ed.), International Taxation and Multinational Activity. Chicago, IL: University of Chicago Press. Amsden, A. and Chu, W. (2003). Beyond Late Development: Taiwan’s Upgrading Policies. Cambridge, MA: Massachusetts Institute of Technology Press. ˚ ., Andersson, M., and Wichmann, C. (1993). O ¨ resundsregionen – kreaAndersson, A tivitet, integration, va¨xt. Copenhagen: Munksgaard. Angel, D. and Engstrom, J. (1995). ‘Manufacturing Systems and Technological Change: The U.S. Personal Computer Industry’, Economic Geography, 71(1): 79–102. Anger, K. (1975). Hollywood Babylon. San Francisco, CA: Straight Arrow Books. Anselin, L., Varga, A., and Acs, Z. (1997). ‘Local Geographic Spillovers Between University Research and High Technology Innovations’, Journal of Urban Economics, 42: 422–48. Anton, J. J. and Yao, D. A. (1995). ‘Start-ups, Spin-offs, and Internal Projects’, Journal of Law, Economics and Organizations, 11: 362–78. Archibugi, D. and Lundvall, B. A. (2001). The Globalizing Learning Economy. Oxford: Oxford University Press. Arora, A. and Gambardella, A. (1990). ‘Complementarity and External Linkage: The Strategies of the Large Firms in Biotechnology’, Journal of Industrial Economics, 37(4): 361–79. —— —— (1994). ‘The Changing Technology of Technological Change: General and Abstract Knowledge and the Division of Innovative Labour’, Research Policy, 23: 523–32. —— Fosfuri, A., and Gambardella, A. (2004). Markets for Technology. Cambridge, MA: Massachusetts Institute of Technology Press. —— Gambardella, A., Pammolli, F., and Riccaboni, M. (2001). The Nature and the Extent of the Market for Technology in Biopharmaceuticals. Cambridge, MA: NBER Summer Institute. —— —— and Torrisi, S. (2001). ‘In the Footsteps of Silicon Valley? Indian and Irish Software in the International Division of Labour’, Stanford Institute for Economic Policy Research Working Paper. Arthur, W. B. (1988a). ‘Competing Technologies: An Overview’, in G. Dosi, C. Freeman, R. Nelson, G. Silverberg, and L. Soete (eds.), Technical Change and Economic Theory. London: Pinter Publishers. —— (1988b). ‘Urban System and Historical Path Dependency’, in J. H. Ausubel and R. Herman (eds.), Cities and Their Vital Systems. Washington, DC: National Academy Press.

280

Bibliography —— (1990). ‘Silicon Valley Locational Clusters: When Do Increasing Returns Imply Monopoly?’, Mathematical Social Sciences, 19: 235–51. —— (1994). Increasing Returns and Path-dependency in Economics. Ann Arbor, MI: University of Michigan Press. —— Ermoliev, Y. M., and Kaniovsky, Y. M. (1987). ‘Path-dependent Processes and the Emergence of Macrostructure’, European Journal of Operational Research, 30: 294–303. Au, C.-C. and Henderson, V. (2002). ‘How Migration Restrictions Limit Agglomeration and Productivity in China’, NBER Working Paper, 8707. Audretsch, D. B. (1989). ‘Agglomeration and the Location of Innovative Activity’, Oxford Review of Economic Policy, 14(2): 18–29. —— (2001). ‘The Role of Small Firms in U.S. Biotechnology Clusters’, Small Business Economics, 17: 3–15. —— and Feldman, M. (1996). ‘R&D Spillovers and the Geography of Innovation and Production’, American Economic Review, 86(3): 630–40. —— and Stephan P. (1996). ‘Company-Scientist Locational Linkages: The Case of Biotechnology’, American Economic Review, 86: 641–52. Avnimelech, G. (2004). An Industry Life Cycle Model of Venture Capital Industries. MA thesis, Jerusalem. —— and Teubal, M. (2004a). ‘Venture Capital—Start-Up Co-Evolution and the Emergence and Development of Israel’s New High-Tech Cluster—Part 1: Macroeconomic & Industry Analysis’, Economics of Innovation and New Technology, 13(1): 33–60. —— —— (2004b). ‘Strength of Market Forces and the Successful Emergence of Israel’s Venture Capital Industry: Insights From a Policy-led Case of Structural Change’, forthcoming in Revue Economique. —— —— (2005a). ‘Creating VC Industries Which Co-Evolve With High Tech: Insights From an Extended Industry Life Cycle (ILC) Perspective to the Israeli Experience’, in process. —— —— (2005b). ‘Evolutionary Innovation and High-Tech Policies: What Can We Learn From Israel’s Targeting of Venture Capital?’, Sanuel Neaman Institute, Technion, Israel Institute of Technology, Science, Technology and Economy Program Working Paper, 25. —— Kenney, M., and Teubal, M. (2003). ‘Building Venture Capital Industries: Understanding the U. S. and Israeli Experiences’, Working Paper. —— —— —— (2005). ‘A Life Cycle Model for the Creation of National Venture Capital Industries: Comparing the US and Israeli Experience’, in E. Guiliani, R. Rabellotti, and M. P. van Dijk (eds.), Clusters Facing Competition: The Importance of External Linkages. Hampshire, UK: Ashgate Publishing. Baccaro, L. and Simoni, M. (2002). Irish Social Partnership and the Celtic Tiger Phenomenon, (unpublished). Geneva: International Labour Office. Bahrami, H. and Evans, S. (2000). ‘Flexible Recycling and High-Technology Entrepreneurship’, in M. Kenney (ed.), Understanding Silicon Valley. Stanford, CA: Stanford University Press.

281

Bibliography Balio, T. (1976a). ‘Struggles for Control, 1908–1930’, in T. Balio (ed.), The American Film Industry. Madison, WI: University of Wisconsin Press. —— (1976b). United Artists: The Company Built by Stars. Madison, WI: University of Wisconsin Press. —— (1976c). ‘A Mature Oligopoly, 1930–1948’, in T. Balio (ed.), The American Film Industry. Madison, WI: University of Wisconsin Press. Banerjee, A. V. (1992). ‘A Simple Model of Herd Behavior’, The Quarterly Journal of Economics, 107(3): 797–817. Baptista, R. (1998). ‘Clusters, Innovation and Growth: A Survey of the Literature’, in G. Swann, M. Prevezer, and D. Stout (eds.), The Dynamics of Industrial Clusters: International Comparisons in Computing and Biotechnology. Oxford: Oxford University Press. —— and Swann, G. (1998). ‘Do Firms in Clusters Innovate More?’, Research Policy, 27: 525–40. —— —— (1999). ‘A Comparison of Clustering Dynamics in the US and UK Computer Industries’, Journal of Evolutionary Economics, 9(3): 373–99. Bar, H. (2002). ‘Is Venture Capital Special? Empirical Evidence From a Government Initiated VC Market’, Neaman Institute, Technion; Science, Technology, Working Paper STE-WP-9. Baraba´si, A.-L. (2000a). Linked: The New Science of Networks. Cambridge, MA: Perseus. —— (2002b). ‘Emergence of Scaling in Complex Networks’, in S. Bornholdt and H. G. Schuster (eds.), Handbook of Graphs and Networks: From the Genome to the Internet. Berlin: Wiley, VCH. —— and Albert, R. (1999). ‘Emergence of Scaling in Random Networks’, Science, 286: 509–12. Bardini, T. (2000). Bootstrapping: Douglas Englebart, Coevolution, and the Origins of Personal Computing. Stanford, CA: Stanford University Press. Barley, S. R., Freeman, J., and Hybels, R. C. (1992). ‘Strategic Alliances in Commercial Biotechnology’, in N. Nohria and R. G. Eccles (eds.), Networks and Organizations. Cambridge, MA: Harvard Business School Press. Barry, F., and Bradley, J. (1997). ‘‘FDI and trade: The Irish host-country experience’’, Economic Journal, 107: 1798–811. Barry, F. and Van Egeraat, C. (2005). ‘The Eastward Shift of Computer Hardware Production: How Ireland Adjusted’, National University of Ireland-Maynooth, National Institute for Regional and Spatial Analysis Working Paper, 27. ¨ rg, H., and Strobl, E. (2003). ‘Foreign Direct Investment, Agglomerations —— Go and Demonstration Effects: an Empirical Investigation’, Weltwirtschaftliches Archiv, 139(4): 583–600. Basant, R. (2002). Knowledge Flows and Industrial Clusters: An Analytical Review of Literature. Ahmedabad, India: Indian Institute of Management. Working Paper, 2002-02-01. Bass, F. (1969). ‘A New Product Growth Model for Consumer Durables’, Management Science, 15: 215–27.

282

Bibliography Bathelt, H., Malmberg, A., and Maskell, P. (2002). ‘Clusters and Knowledge: Local Buzz, Global Pipelines and the Process of Knowledge Creation’, DRUID Working paper, 12. Battat, J., Frank, I., and Shen, X. (1996). ‘Suppliers to Multinationals: Linkage Programmes to Strengthen Local Companies in Developing Countries’, FIAS Occasional Paper, 6 Washington, DC: Foreign Investment Advisory Service. Baum, J. A. C. and Haveman, H. A. (1997). ‘Love Thy Neighbor? Differentiation and Agglomeration in the Manhattan Hotel Industry, 1898–1990’, Administrative Science Quarterly, 42: 304–38. Beaudry, C. and Breschi, S. (2003). ‘Are Firms in Clusters Really More Innovative?’, Economics of Innovation and New Technology, 12(4): 325–42. Becattini, G. (1990). ‘The Marshallian Industrial District as a Socio-Economic Notion’, in F. Pyke, G. Beccattini, and W. Sengenbergers (eds.), Industrial Districts and Inter-Firm Cooperation in Italy. Geneva: International Institute for Labour Studies. Becker, R. and Hellmann, T. (2002). ‘The Genesis of Venture Capital: Lessons From the German Experience’, Research Paper Series No. 1705, Stanford, CA: Stanford University, [typescript]. Belussi, F. (1992). Nuovi modelli di impresa, gerarchie organizzative ed imprese rete. Milano: FrancoAngeli. Ben-David, J. (1977). Centers of Learning: Britain, France, Germany and the United States. New York: McGraw-Hill. Berggren, C. and Laestadius, S. (2003). ‘Co-Development and Composite Clusters— the Secular Strength of Nordic Telecommunications’, Industrial and Corporate Change, 12(1): 91–114. Berry, A., Rodriguez, E., Sandee, H. (2002). ‘Firm and Group Dynamics in the Small and Medium Enterprise Sector in Indonesia’, Small Business Economics, 18(1–3): 141–61. Bikhchandani, S., Hirshleifer, D., and Welch, I. (1992). ‘A Theory of Fads, Fashion, Custom and Cultural Change as Informational Cascades’, Journal of Political Economy, 5: 992–1026. BioScan (1987–2003). Atlanta, GA: Thomson Group. Biosea Biotech (2004). ‘Personal Interview’, March 26, 2004. Bioson Bioengineering (2004). ‘Personal Interview’, March 22, 2004. Bischia, G.-I., Dawid, H., and Michael, K. (2003). ‘Spillover Effects and the Evolution of Firm Clusters’, Journal of Economic Behavior and Organization, 50(1): 47–75. ¨ rkbacka, R. (2002). Klusterdynamik och industrins lokalisering—en ja¨mfo¨rande Bjo studie av skillnader i industrins lokalisering i Life Science-klustret Medicon Valley, mimeo. Stockholm: University of Stockholm. Black, B. and Gilson, R. (1998). ‘Venture Capital and the Structure of Capital Markets: Banks versus Stock Markets’, Journal of Financial Economics, 47: 243–77. —— —— (1999). Does Venture Capital Require an Active Stock Market. Blakely, E. J. (1989). Planning Local Economic Development. Theory and Practice. London: Sage.

283

Bibliography Blue Star Biotech (2004). ‘Personal Interview’, March 16, 2004. Bolaote Biotech (2004). ‘Personal Interview’, March 16, 2004. Bordwell, D, Staiger, J., and Thompson, K. (1985). The Classical Hollywood Cinema: Film Style and Mode of Production to 1960. New York: Columbia University Press. Borgatti, S. P. and Foster, P. C. (2003). ‘The Network Paradigm in Organizational Research: A Review and Typology’, Journal of Management, 29: 991–1013. Borrell, J. (2001). ‘The Bankers’, Upside, October, 96–108. Boston Consulting Group (2002). Commercial Attractiveness of Biomedical R&D in Medicon Valley-The Role of R&D in Attracting Regional Investments. Copenhagen: The Boston Consulting Group. Bottazzi, G., Fagiolo, G., and Dosi, G. (2002). ‘Mapping Sectoral Patterns of Technological Accumulation Into the Geography of Corporate Locations. A Simple Model and Some Promising Evidence’, LEM Working Paper Series. Bowser, E. (1990). The Transformation of Cinema, 1907–1915. New York: Charles Scribner’s Sons. Bramanti, A. and Maggioni, M. A. (1997). La dinamica dei sistemi produttivi territoriali: teorie, tecniche, politiche. Milano: Angeli. Braun, D. (1994). Structure and Dynamics of Health research and Public Funding: An International Institutional Comparison. Amsterdam: Kluwer Academic Publishers. Braunerhjelm, P. and Carlsson, B. (2003). ‘Introduction: Regional Growth, Clusters and Institutions’, Industry and Innovation, 10(1): vii–ix. —— Faini, R., Norman, V., Ruane, F., Seabright, P. (2000). Integration and the Regions of Europe. How Right Polcies Can Prevent Polarization, Monitoring European Integration 10. London: CEPR. Breschi, S. and Lissoni, F. (2001). ‘Knowledge Spillovers and Local Innovation Systems: A Critical Survey’, Industrial and Corporate Change, 10(4): 975–1006. —— and Malerba, F. (2001). ‘The Geography of Innovation and Economic Clustering: Some Introductory Notes’, Industrial and Corporate Change, 10(4): 817–33. —— Malerba, F., and Orsenigo, L. (2000). ‘Technological Regimes and Schumpeterian Patterns of Innovation’, The Economic Journal, 110(463): 388–410. —— Lissoni, F., and Orsenigo, L. (2001 and 2003). ‘Success and Failure in the Development of Biotechnology Clusters: The Case of Lombardy’, in G. Fuchs (ed.), Biotechnology in Comparative Perspective. London and New York: Routledge. Bresnahan, T. F., Gambardella, A., and Saxenian, A. (2001). ‘Old Economy’ Inputs for ‘New Economy’ Outcomes: Cluster Formation in the New Silicon Valleys’, Industrial and Corporate Change, 10(4): 835–60. Brezis, E. S. and Krugman, P. (1997). ‘Technology and the Life Cycle of Cities’, Journal of Economic Geography, 2(4): 369–83. —— Krugman, P., and Tsiddon, D. (1993). ‘Leapfrogging in International Competition: A Theory of Cycles in National Technological Leadership’, American Economic Review, 83(5): 1211–9. Brock, W. A. (1999). ‘Scaling in Economics: A Reader’s Guide’, Industrial and Corporate Change, 8(3): 409–46.

284

Bibliography Brownlow, K. (1979). Hollywood: The Pioneers. London: Collins. Bruno, G. and Orsenigo, L. (2003). ‘Variables Influencing Industrial Funding of Academic Research in Italy. An Empirical Analysis’, International Journal of Technology Management, 26: 277–302. Burt, R. S. (1992). Structural Holes: The Social Structure of Competition. Cambridge, MA: Harvard University Press. Business Alert—China (2003). ‘General Survey of Salaries in Major Cities’, Hong Kong Development Council. Bygrave, W. D. and Timmons, J. A. (1992). Venture Capital at the Crossroads. Boston: Harvard Business School Press. Caniels, M. C. J. and Romijn, H. A. (2003). ‘Firm-level Accumulation and Regional Dynamics’, Industrial and Corporate Change, 12: 1253–78. Cantwell, J. A. (1989). Technological Innovations and Multinational Corporations. Oxford: Basic Books. Cao (2004). ‘Personal Inverview with Cao, Shanghai Biotech and Pharmaceutical Association, February 24, 2004. Campbell, J. (1972). Myths to Live By. New York: Penguin. Carlsson, B. (1995). Technological Systems and Economic Performance. The Case of Factory Automation. Boston, MA: Kluwer Academic Publishers. —— (1997). Technological Systems and Industrial Dynamics. Boston, MA: Kluwer Academic Publishers. —— (2002). Technological Systems in the Bio Industries: An International Study. Boston, MA: Kluwer Academic Publishers. —— and Jacobsson, S. (2004). ‘Dynamics of Innovation Systems—Policy-Making in a Complex and Non-Deterministic World’, Case Western Reserve University and Chalmers University of Technology, Working Paper. —— and Mudambi, R. (2003). ‘Globalization, Entrepreneurship, and Public Policy: A Systems View’, Industry and Innovation, 10(1): 103–16. —— Jacobsson, S., and Bergek, A. (2005). ‘Dynamics of Innovation Systems— Policy-Making in a Complex and Non-deterministic World’, Paper presented at the DRUID Conference, Copenhagen, June 28–30. —— —— Holmeń, M. and Rickne, A. (2002). ‘Innovation Systems: Analytical and Methodological Issues’, Research Policy, 31(2): 233–45. Carrol, G. C. (1988). Ecological Models of Organizations. Cambridge, MA: Ballinger. Carson, R. L. (1999). ‘Personal Interview by Martin Kenney’, September 29, 1999. Casper, S. (1999). National Institutional Framework and High-Technology Innovation in Germany: The Case of Biotechnology, mimeo. Berlin. —— and Kettler, H. (2001). ‘National Institutional Frameworks and the Hybridization of Entrepreneurial Business Models: The German and UK Biotechnology Sectors’, Industry and Innovation, 8: 5–13. Cassady, R. (1982). ‘Monopoly in Motion-picture Production and Distribution, 1908–1915’, in G. Kindem (ed.), The American Movie Industry: The Business of Motion Pictures. Carbondale, IL: Southern Illinois University Press.

285

Bibliography Chamberlin, T. and de la Mothe, J. (2003). ‘Northern Light: Ottawa’s Technology Cluster’, in D. A. Wolfe (ed.), Clusters Old and New: The Transition to a Knowledge Economy in Canada’s Regions. Kingston: School of Policy Studies, Queen’s University. Chesbrough, H. W. (2003). Open Innovation: The New Imperative for Creating and Profiting from Technology. Boston, MA: Harvard Business School Press. Chinese Academy of Social Science (2003a). The Developing China’s Biotech Industry, www.cass.net.cn, Journal of the Chinese Academy of Social Science, September 18, 2003. China Science and Technology Statistics (2003b). Center of Science and Technology Statistics, Huazhong University of Science and Technology, Wuhan. http://www.sts.org.cn/tjbg/other/documents/2003/0301.htm China Venture Capital Yearbook 2002 (2003). Edited by Venture Capital Development Centre of Renmin University and Beijing Venture Capital China Co. Lts, Beijing: Democracy and Development Press. Choi, J. P. (1997). ‘Herd Behaviour, the ‘‘Penguin Effect’’, and the Suppression of Informational Diffusion: An Analysis of Informational Externalities and Payoff Interdependency’, Rand Journal of Economics, 28(3): 407–25. Chompalov, I., Genuth, J. and Shrum, W. (2002). ‘The Organization of Scientific Collaborations’, Research Policy, 31(5): 749–67. Christensen, C. M. (1992). The Innovator’s Challenge: Understanding the Influence of Market Environment on Processes of Technology Development in the Rigid Disk Drive Industry, DBA dissertation. Boston, MA: Harvard Business School. Chuangshiji Modified Genetic Technology (2004). ‘Personal Interview’, March 9, 2004. Ciconne, A. (2002). ‘Agglomeration-Effects in Europe’, European Economic Review, 46: 213–27. Clancy, P., O’Malley, E., O’Connell, L., and Van Egeraat, C. (2001). ‘Industry Clusters in Ireland: An Application of Porter’s Model of National Competitive Advantage to Three Sectors’, European Planning Studies, 9(1): 7–28 [available online]. Clark, B. R. (1995). Places of Inquiry. Research and Advanced Education in Modern Universities. Berkeley, CA: University of California Press. Clarke, C. G. (1976). Early Film Making in Los Angeles. Los Angeles, CA: Dawson’s Book Shop. Cockburn, I. and Henderson, R. (1996). ‘Public-Private Interaction in Pharmaceutical Research’, Proceedings of the National Academy of Sciences, 93(23): 12725–30. —— —— and Stern, S. (1997). Fixed Effects and the Diffusion of Organizational Practice in Pharmaceutical Research, mimeo. Cambridge, MA: Massachusetts Institute of Technology Press. Coleman, J. S. (1988). ‘Social Capital I the Creation of Human Capital’, American Journal of Sociology, 94: 95–120. Cooke, P. (2001). ‘Regional Innovation Systems, Clusters, and the Knowledge Economy’, Industrial and Corporate Change, 10(4): 945–74.

286

Bibliography —— (2001). ‘Biotechnology Clusters in the UK: Lessons from Localisation in the Commercialisation of Science’, Small Business Economics, 17: 43–59. —— (2002). Knowledge Economies: Clusters, Learning & Cooperative Advantages. London: Routledge. Coriat, B. and Dosi, G. (1998). ‘The Institutional Embeddedness of Economic Change: an Appraisal of the ‘‘Evolutionary’’ and ‘‘Regulationist’’ Research Programmes’, in K. Nielsen and B. Johnson (eds.), Institutions, and Economic Change: New Perspectives on Markets, Firms and Technology. Cheltenham, UK: Edgar Elgar. Cowan, R. and Cowan, W. (1998). ‘On Clustering in the Location of R&D: Statics and Dynamics’, Economics of Innovation and New Technology, 6: 201–29. —— and Jonard, N. (2004). ‘Network Structure and the Diffusion of Knowledge’, Journal of Economic Dynamics and Control, 28: 1557–71. Cy-tech Bio Science (2004). ‘Personal Interview’, March 17, 2004. Dahl, M. S. and Pedersen, C. Ø. R. (2003). ‘Knowledge Flows through Informal Contacts in Industrial Clusters: Myths or Realities?’, DRUID Working paper, 1. —— —— and Dalum, B. (2003). ‘Entry by Spinoff in a High-tech Cluster’, DRUID Working Paper, 2003-11. Dahlman, C. J., Ross-Larsen, B., and Westphal, L. (1987). ‘Managing Technological Development: Lessons from Newly Industrialized Countries’, World Development, 15(6): 759–77. David, P. A. (1969). ‘A Contribution to the Theory of Diffusion’, Centre for Research in Economic Growth Research, Memorandum 71. Stanford, CA: Stanford University Press. —— (1985). ‘Clio and the Economics of QWERTY’, American Economic Review, 75: 332–7. —— (1997). ‘Path Dependence and the Quest for Historical Economics: One More Chorus of the Ballad of QWERTY’, Discussion Papers in Economic and Social History, No. 20. University of Oxford. —— Foray, D., and Dalle, J. M. (1998). ‘Marshallian Externalities and the Emergence of Spatial Stability and Technological Enclaves’, Economics of Innovation and New Technology, 6: 147–82. —— and Rosenbloom, J. L. (1990). ‘Marshallian Factor, Market Externalities and the Dynamics of Industrial Localisation’, Journal of Urban Economics, 28: 349–70. Davies, S. and Lyons, B. (1996). Industrial Organisation in the EU. Oxford: Clarendon Press. Davis, R. L. (1993). The Glamour Factory: Inside Hollywood’s Big Studio System. Dallas, TX: Southern Methodist University Press. Dedrick, J. and Kraemer, K. (2002). ‘Globalization of the Personal Computer Industry: Trends and Implications’, Paper 254, Center for Research on Information Technology and Organizations. Irvine, CA: University of California. de Geus, A. J. (co-founder, Synopsys). (1988). ‘Personal Interview by Richard Florida and Martin Kenney’, March 30, 1988.

287

Bibliography de Geus, A. J. (1998). ‘Personal Interview by Richard Florida and Martin Kenney’, March 30, 1998. Dendrinos, D. S. and Mullally, H. (1985). Urban Evolution. Studies in the Mathematical Ecology of Cities. Oxford: Oxford University Press. Desai, M. A., Foley, C. F., and Hines, J. R., Jr. (2002). ‘Chains of Ownership, Regional Tax Competition and Foreign Direct Investment’, NBER Working Paper, No. 9224. Development of Commercialization of China’s Biotechnology, www.medboo. com Dixit, A. K. and Stiglitz, J. E. (1977). ‘Monopolistic Competition and Optimum Product Diversity’, American Economic Review, 67: 297–308. Dorfman, N. (1983). ‘Route 128: The Development of a Regional High Technology Economy’, Research Policy, 13: 299–316. Dosi, G. (1982). ‘Technological Paradigms and Technological Trajectories’, Research Policy, 11: 147–62. —— (1984). Technical Change and Economic Transformation. London: Macmillan. —— Freeman, C., and Fabiani, S. (1994). ‘The Process of Economic Development: Introducing Some Stilized facts and Theories on Technologies, Firms, and Institutions’, Industrial and Corporate Change, 3(1): 1–46. —— Freeman, C., Nelson, R. R., Silverberg, G., and Soete, L. (eds.) (1988). Technical Change and Economic Theory. London: Pinter Publishers. —— Marsili, O., Orsenigo, L., and Salvatore, R. (1995). ‘Technological Regimes, Selection and Market Structures’, Small Business Economics, 7(6): 411–36. Dossani, R. and Kenney, M. (2002). ‘Creating an Environment for Venture Capital in India’, World Development, 30(2): 227–53. Durkan, J. (1998). Intel: The Concept and Measurement of Embeddedness, Unpublished manuscript. Belfield, Dublin: University College Dublin. Echeverri-Carroll, E. L. and Brennan, W. (1999). ‘Are Innovation Networks Bounded by Proximity?’, in M. Fischer, L. Suarez-Villa, and M. Steiner (eds.), Innovation, Networks and Localities. Berlin: Springer-Verlag, pp. 28–49. Economist Intelligence Unit (2003). 2003 Country Profile, China. London: The Economist Intelligence Unit. Edquist, C. (1997). Systems of Innovation: Technologies, Institutions and Organisations. London: Pinter Publishers. Eisenhardt, K. M. and Schoonhoven, C. B. (1990). ‘Organizational Growth: Linking Founding Team, Strategy, Environment, and Growth Among U.S. Semiconductor Ventures, 1978–1988’, Administrative Science Quarterly, 35(3): 504–29. EIU (2000). 2000 China-Country Report, 1st Quarter. Beijing, Economist Intelligence Unit. Eliasson, G. (2003). ‘Global Economic Integration and Regional Attractors of Competence’, Industry and Innovation, 10(1): 75–102. Ellison, G. and Glaeser, E. L. (1999). ‘The Geographic Concentration of Industry: Does Natural Advantage Explain Agglomeration?’, American Economic Review, 89(2): 311–6. Erlich, Y. (1998 and 2000), ‘Interview by Morris Teubal and Gil Avnimelech’.

288

Bibliography European Commission (2003). Third European Report on Science and Technology Indicators: Towards a Knowledge-Based Economy. Brussels: CEC. Farrell, J. and Saloner, G. (1985). ‘Standardization, Compatibility, and Innovation’, Rand Journal of Economics, 16(1): 407–25, 701–83. —— —— (1987). ‘Competition, Compatibility and Standards: The Economics of Horses, Penguins, and Lemmings’, in H. L. Gabel (ed.), Product Standardization and Competitive Strategy. Amsterdam: Elsevier. Feldman, M. P. (1994). The Geography of Innovation. Dordrecht: Kluwer Academic Publishers. —— (2000). ‘Location and Innovation: The New Economic Geography of Innovation, Spillovers, and Agglomeration’, in G. Clark, M. Feldman, and M. S. Gertler (eds.), Oxford Handbook of Economic Geography. Oxford: Oxford University Press. —— (2001). ‘The Entrepreneurial Event Revisited: Firm Formation in a Regional Context’, Industrial and Corporate Change, 10(4): 861–91. —— (2003). The Locational Dynamics of the U.S. Biotech Industry: Knowledge Externalities and The Anchor Hypothesis, mimeo. —— (2004). ‘Homegrown Solutions: Fostering Cluster Formation’, Economic Development Quarterly, 18(2): 127–37. —— and Audretsch, D. B. (1999). ‘Innovation in Cities: Science-Based Diversity, Specialization, and Localized Competition’, American Economic Review, 43: 409–29. —— and Desrochers, P. (2003). ‘The Evolving Role of Research Universities in Technology Transfer: Lessons From the History of Johns Hopkins University’, Industry and Innovation, 10: 5–24. —— and Florida, R. (1994). ‘The Geographic Sources of Innovation: Technological Infrastructures and Product Innovation in the United States’, Annals of the Association of American Geographers, 84: 210–29. —— and Francis, J. L. (2003a). Home Grown Solutions: Fostering Business Formation Through Science and Technology Policy, mimeo. —— —— (2003b). ‘Fortune Favors the Prepared Region: The Case of Entrepreneurship and the Capitol Region Biotechnology Cluster’, European Planning Studies, 11(7): 765–88. —— and Schreuder, Y. (1996). ‘Initial Advantage: The Origins of the Geographic Concentration of the Pharmaceutical Industry in the Mid-Atlantic Region’, Industrial and Corporate Change, 5(3): 839–96. —— Francis, J. L., and Bercovitz, J. (2005). ‘Creating a Cluster While Building a Firm: Entrepreneurs and the Formation of Industrial Clusters’, Regional Studies, 39(1): 129–41. —— and Martin, R. (2005). ‘Jurisdictional Advantage’, in A. B. Jaffe, J. Lerner and S. Stern (eds.), Innovation Policy and the Economy, Vol. 5. Cambridge, MA: Massachusetts Institute of Technology Press. Ferguson, C. H. (1999). High Stakes, No Prisoners: A Winner’s Tale of Greed and Glory in the Internet Wars. New York, NY: Times Books.

289

Bibliography Fernett, G. (1973). Poverty Row. Satellite Beach, FL: Reef Publications. —— (1988). American Film Studios: An Historical Encyclopedia. Jefferson, NC: McFarland. Ferreira, L. and Vanhoudt, P. (2002) ‘Catching the Celtic Tiger by its Tail’, EIB Economic and Financial Report No. 2002/01 (2002). Forthcoming in European Journal of Education. Fiedler, M. and Hellmann, T. (2001). ‘Against All Odds: The Late but Rapid Development of the German Venture Capital Industry’, Journal of Private Equity. Fields, G. (2004). Territories of Profit: Communications, Capitalist Development and the Innovative Enterprises of G. F. Swift and Dell Computer. Stanford, CA: Stanford University Press. FitzGerald, J. (1998). ‘An Irish Perspective on the Structural Funds’, European Planning Studies, 6(6): 677–94. Fleming, L. and Sorenson, O. (2001). ‘Technology as a Complex Adaptive System: Evidence from Patent Data’, Research Policy, 30: 1019–39. Florey, R. (1923). Filmland: Los Angeles et Hollywood, les Capitales du Cine´ma. Paris: Editions de Cine´magazine. —— (1948). Hollywood d’Hier et d’Aujourd’hui. Paris: Editions Prisma. Florida, R. and Kenney, M. (1988). ‘Venture Capital-financing Innovation and Technological Change in the US’, Research Policy, 17: 119–37. —— —— (1988). ‘VC, High Technology and Regional Development’, Regional Studies, 22: 33–48. —— —— (1990). The Breakthrough Illusion. New York, NY: Basic Books. Forfa´s (2003). Business Expenditure on Research and Development, 2001. Dublin: Forfa´s. Fosfuri, A. and Rønde, T. (2004). ‘High-Tech Clusters, Technology Spillovers, and Trade Secret Laws’, International Journal of Industrial Organization, 22(1): 45–65. Frank, L. (2002). ‘Biotechnology in Medicon Valley’, Nature, 20: 433–435. Freiberger, P. and Swaine, M. (1984). Fire in the Valley. Berkeley, CA: Osborne/ McGraw-Hill. Fruchterman, T. and Reingold, E. (1991). ‘Graph Drawing by Force-Directed Replacement’, Software—Practice and Experience, 21: 1129–64. Furman, J. L., Porter, M. E., and Stern, S. (2002). ‘The Determinants of National Innovative Capacity’, Research Policy, 31(6): 899–933. Fujita, M. (1988). ‘A Monopolistic Competition Model of Spatial Agglomeration’, Regional Science and Urban Economics, 18: 87–124. —— (1989). Urban Economic Theory. Land Use and City Size. Cambridge, MA: Cambridge University Press. —— and Thisse, J.-F. (1996). ‘Economics of Agglomeration’, Journal of the Japanese and International Economies, 10: 339–78. —— Krugman, P., and Venables, A. I. (1999). The Spatial Economy: Cities, Regions, and International Trade. Cambridge, MA: Massachusetts Institute of Technology Press.

290

Bibliography Gabe, T. (2003). ‘Local Industry Agglomeration and New Business Activity’, Growth and Change, 34(1): 17–39. Gabriele, A. (2001). ‘Science and Technology Policies, Industrial Reform and Technical Progress in China: Can Socialist Property Rights be Compatible with Technological Catching Up?’ UNCTAD Discussion Paper, 155. Galambos, L. and Sturchio, J. (1996). ‘The Pharmaceutical Industry in the Twentieth Century’, History and Technology, 13: 83–100. Gambardella, A. (1995). Science and Innovation in the US Pharmaceutical Industry. Cambridge, MA: Cambridge University Press. —— Orsenigo, L. and Pammolli, F. (2001). ‘Industrial Competitiveness in Pharmaceuticals. A European Perspective’, DG Enterprise, European Commission, Working Paper, 1. Gambarotto, F. and Maggioni, M. A. (1998). ‘Regional Development Strategies in Changing Environments: an Ecological Approach’, Regional Studies, 32: 49–61. Gargiulo, M. and Benassi, M. (2000). ‘Trapped in Your Own Net? Network Cohesion, Structural Holes and the Adaptation of Social Capital’, Organization Science, 11: 183–96. Garofoli, G. (1991). Modelli locali di sviluppo. Milano: FrancoAngeli. Garud, R. and Karnøe, P. (2001). ‘Path Creation as a Process of Mindful Deviation’, in R. Garud and P. Karnoe (eds.), Path Dependence and Creation. Mahwah, NJ: Lawrence Erlbaum, 1–28. —— —— (2003). ‘Bricolage versus breakthrough: distributed and embedded agency in technology entrepreneurship’, Research Policy, 32(2): 277–300. Gertler, M. S. (2002). ‘Technology, Culture and Social Learning: Regional and National Institutions of Governance’, in M. S. Gertler and D. A. Wolfe (eds.), Innovation and Social Learning: Institutional Adaptation in an Era of Technological Change. Basingstoke, UK: Palgrave, 111–34. —— (2004). Manufacturing Culture: The Institutional Geography of Industrial Practice. Oxford: Oxford University Press. —— and Wolfe, D. A. (2004). Local Social Knowledge Management: Community Actors, Institutions and Multilevel Governance in Regional Foresight Exercises’, Futures, 36(1): 45–65. Gilson, R. (2003). ‘Engineering a Venture Capital Market: Lessons from the American Experience’, Stanford Law Review, 55: 1067–84. Gilson, R. J. and Black, B. S. (1998). ‘Venture Capital and the Structure of Capital Markets’, Journal of Financial Economics, 47: 243–77. Gittelman, M. (2000). Mapping National Knowledge Networks: Scientists, Firms and Institutions in Biotechnology in the United States and France, Ph.D. dissertation. New York: New York University. Giuliani, E. (2004). ‘Laggard Clusters as Slow Learners, Emerging Clusters as Locus of Knowledge Cohesion (and Exclusion). A Comparative Study in the Wine Industry’, LEM Working Paper, 2004/09.

291

Bibliography Glaeser, E. L. (2003). ‘Reinventing Boston: 1640–2003’, Harvard Institute of Economic Research Discussion Paper, 2017. —— Kallal, H. D., Scheinkman, J. A., and Shleifer, A. (1992). ‘Growth of Cities’, Journal of Political Economy, 100: 1126–52. Gomery, J. (1976). ‘The Coming of the Talkies: Invention, Innovation, and Diffusion’, in T. Balio (ed.), The American Film Industry. Madison, WI: University of Wisconsin Press. Gompers, P. A. and Lerner, J. (1999). The Venture Capital Cycle. Cambridge, MA: Massachusetts Institute of Technology Press. —— —— (2001). ‘The Venture Capital Revolution’, Journal of Economic Perspectives, 15(2): 145–68. —— —— and Scharfstein, D. (2003). ‘Entrepreneurial Spawning: Public Corporations and the Genesis of New Ventures, 1986–1999’, NBER Working Paper, 9816. ¨ rg, H. and Ruane F. (1998). ‘Linkages Between Multinationals and Indigenous Go Firms: Evidence for the Electronics Sector in Ireland’, Trinity Economic Papers, Technical Paper 13, Trinity College, Dublin. —— and Strobl, E. (2003). ‘Multinational Companies, Technology Spillovers and Plant Survival: Evidence for Irish Manufacturing’, Scandinavian Journal of Economics, forthcoming. Gort, M. and Klepper, S. (1982). ‘Time Paths in the Diffusion of Product Innovations’, Economic Journal, 92: 630–53. Gowrisankaran, G. and Stavins, J. (2002). ‘Network Externalities and Technology Adoption: Lessons from Electronic Payments’, NBER Working Paper, 8943. Green, R., J. Cunningham, J., Giblin, I., Moroney, M., and Smyth, L. (2001). ‘The Boundaryless Cluster: Information and Communications Technology in Ireland’, in Innovative Clusters: Drivers of National Innovation Systems. Paris: OECD. Green Gingko Biotech (2004). ‘Personal Interview’, March 25, 2004. Grilliches, Z. (1957). ‘Hybrid Corn: an Exploration in the Economics of Technological Change’, Econometrica, 25(4): 501–22. —— (1992). ‘The Search for R&D Spillovers’, Scandinavian Journal of Economics, 94: 29–47. Gropp, R. and Kostial, K. (2000). ‘The Disappearing Tax Base: is Foreign Direct Investment Eroding Corporate Income Taxes?’, European Central Bank Working Paper, 31. Guston, D. (1998). ‘Technology Transfer and the Use of CRADAs at the National Institutes of Health’, in L. M. Branscomb and J. H. Keller (eds.), Investing in Innovation. Cambridge, MA: Massachusetts Institute of Technology Press, 221–49. Hall, P. and Soskice, D. (2001). Varieties of Capitalisms. The Institutional Foundations of Comparative Advantage. Oxford: Oxford University Press. Hagedoorn, J. and Roijakkers, N. (2002). ‘Small Entrepreneurial Firms and Large Companies in Inter-Firm R&D Networks—the International Biotechnology Industry’, in M. A. Hitt (ed.), Strategic Entrepreneurship. Cambridge, MA: Blackwell Publishing. Hampton, B. B. (1931). A History of the Movies. New York, NY: Covici, Friede.

292

Bibliography Hannan, M. T. and Carrol, G. (1992). Dynamics of Organizational Populations: Density, Legitimation, and Competition. Oxford: Oxford University Press. —— and Freeman, J. (1989). Organizational Ecology. Cambridge, MA: Harvard University Press. Hanson, G. H. (1996). ‘Agglomeration, Dispersion and the Pioneer Firm’, Journal of Urban Economics, 39: 255–81. Hanson, P. K. (1988). The American Film Insitute Catalog of Motion Pictures Produced in the United States. Berkeley, CA: University of California Press. Hargadon, A. B. (2003). How Breakthroughs Happen: The Surprising Truth About How Companies Innovate. Cambridge, MA: Harvard Business School Press. —— and Douglas, J. Y. (2001). ‘When Innovations Meet Institutions: Edison and the Design of the Electric Light’, Administrative Science Quarterly, 46: 476–501. Harrison, B., Kelley, M. R., and Gant, J. (1996). ‘Innovative Firm Behavior and Local Milieu: Exploring the Intersection of Agglomeration, Firm Effects, and Technological Change’, Economic Geography, 72: 233–58. Henderson, J. V. (1974). ‘The Size and Types of Cities’, American Economic Review, 64: 640–56. —— (1977). Economic Theory and the Cities. New York, NY: National Academic Press. —— (1986). ‘The Efficiency of Resource Usage and City Size’, Journal of Urban Economics, 19: 47–70. —— (1999). ‘Marshall’s Scale Economies’, Working Paper, 7358. Henderson, R. (1994). ‘The Evolution of Integrative Competence: Innovation in Cardiovascular Drug Discovery’, Industrial and Corporate Change, 3(3): 607–30. —— and Cockburn, I. (1994). ‘Measuring Competence? Exploring Firm Effects in Pharmaceutical Research’, Strategic Management Journal, 15: 63–84. —— —— (1996). ‘Scale, Scope and Spillovers: The Determinants of Research Productivity in Drug Discovery’, Rand Journal of Economics, 27(1): 32–59. —— Orsenigo, L., and Pisano, G. (1999). ‘The Pharmaceutical Industry and the Revolution in Molecular Biology: Exploring the Interactions Between Scientific, Institutional and Organizational Change’, in D. Mowery and R. Nelson (eds.), The Sources of Industrial Advantages. Cambridge, MA: Cambridge University Press. Hiltzik, M. A. (1999). Dealers of Lightning: Xerox PARC and the Dawn of the Computer Age. New York: HarperBusiness. Hirschman, A. O. (1958). The Strategy of Economic Development. New Haven, CT: Yale University Press. Hirsh, S. (1967). Location of Industry and International Competetiveness. Oxford: Clarebdon Press. Hirshleifer, D. (1993). ‘The Blind Leading the Blind: Social Influences, Fads, and Informational Cascades’, Working Paper of the Anderson Graduate School of Management, 24. Hochtberger, K., White, M., and Grimes, S. (2004). ‘The Evolution of Multinational Computer Services Affiliates in Ireland’, Centre for Innovation and Structural Change, Working Paper, 13.

293

Bibliography Huashengyuan Genetic Engineering (2004). ‘Personal Interview’, March 9, 2004. Hsu, D. and Kenney, M. (2005). ‘Organizing Venture Capital: The rise and demise of American Research & Development Corporation, 1946–1973’, Industrial and Corporate Change, 14: 579–616. Hui Lin (2004). Pudong Biotech and Pharmaceutical Association/NSBPIB Development Co Ltd, ‘Personal Interview’, February 24, 2004. Huettig, M. (1944). Economic Control of the Motion Picture Industry. Philadelphia, PA: University of Pennsylvania Press. Ireland, N. and Stoneman, P. (1985). ‘Order Effects, Perfect Foresight and Intertemporal Price Discrimination’, Recherche´ Economique de Louvain, 51(1): 7–20. —— —— (1986). ‘Technological Diffusion, Expectations and Welfare’, Oxford Economic Papers, 38: 283–304. ISA (2002). Biotechnology—Excellence in All Steps of Drug Discovery and Development. Stockholm: A.T.T. Grafiska. Isard, W. (1951). ‘Distance Inputs and the Space Economy: The Locational Equilibrium of the Firm’, Quarterly Journal of Economics, 65: 373–97. —— (1956). Location and Space-Economy. Cambridge, MA: MIT Press. IVA (1997–2003). 1997–2003, Yearbooks, ‘A Survey of Venture Capital and Private Equity in Israel’, edited by IVC. Izod, J. (1988). Hollywood and the Box Office, 1895–1986. London: Macmillan. Jacobs, L. (1939). The Rise of the American Film: A Critical History. New York: Harcourt, Brace. —— (1969). The Economy of Cities. New York: Random House. Jaffe, A. B. (1989). ‘Real Effects of Academic Research’, American Economic Review, 79: 957–70. —— and Trajtenberg, M. (2002). Patents, Citations & Innovations: A Window on the Knowledge Economy. Cambridge, MA: MIT Press. —— Trajtenberg, M. and Henderson, R. (1993). ‘Geographic Localization of Knowledge Spillovers as Evidenced by Patent Citations’, Quarterly Journal of Economics, 108(3): 578–98. Jarvie, I. (1992). Hollywood’s Overseas Campaign: The North Atlantic Movie Trade, 1920–1950. Cambridge, MA: Cambridge University Press. Jerusalem Institute of Management (JIM) (1986). Report to the Office of the Chief Scientist (1986). Jessen, C. (1915). ‘California Is World’s Photoplay Stage’, Motion Picture News, April 3, West Coast Section, 127–8. Jiena Bio-products (2004). ‘Personal Interview’, March 4, 2004. Jin, H., Qian, Y. and Weingast, B. R. (2001). Regional Decentralization and Fiscal Incentives: Federalism, Chinese Style, mimeo. Stanford, CA: Stanford University. Jones, C. (2001). ‘Co-evolution of Entrepreneurial Careers, Institutional Rules and Competitive Dynamics in American Film, 1895–1920’, Organization Studies, 22: 911–44.

294

Bibliography Kaldor, N. J. (1970). ‘The Case for Regional Policies’, Scottish Journal of Political Economy, 17: 337–48. Kamada, T. and Kawai, S. (1989). ‘An Algorithm for Drawing General Undirected Graphs’, Information Processing Letters, 31: 7–15. ¨ rg, H. (2002). ‘Linkages, Agglomerations and Knowledge SpillKearns, A. and Go overs in the Irish Electronics Industry: The Regional Dimension’, International Journal of Technology Management, 24(7): 743–63. Kenney, M. (1986). Biotechnology: The University-industrial Complex. New Haven, CT: Yale University Press. —— (2000). Understanding Silicon Valley. The Anatomy of an Entrepreneurial Region. Stanford, CA: Stanford University Press. —— (2003). ‘The Growth and Development of the Internet in the United States’, in B. Kogut (ed.), The Global Internet Economy. Cambridge, MA: MIT Press. —— and von Burg, U. (1999). ‘Technology and Path Dependence: The Divergence Between Silicon Valley and Route 128’, Industrial and Corporate Change, 8(1): 67–103. —— —— (2001). ‘Technology, Entrepreneurship and Path-Dependence: Industrial Clustering in Silicon Valley and Route 128’, Industrial and Corporate Change, 10: 67–104. —— and Rao, H. (2004). ‘New Forms as Settlements: A Study of the Construction and Development of the Small Business Investment Corporations; 1930–1973’, Working Paper. —— and von Burg, U. (1999). ‘Technology and Path Dependence: The Divergence between Silicon Valley and Route 128’, Industrial and Corporate Change, 8(1): 76–103. Kimberly, J. R. and Miles, H. R. (1980). The Organizational Life Cycle Issues in the Creation, Transformation, and Decline of Organizations. San Francisco, CA: JosseyBass. Kim, L. (1997). From Imitation to Innovation: The Dynamics of Korea’s Technological Learning. Boston, MA: Harvard Business School Press. —— (2000). ‘Korea’s National System of Innovation in Transition’, in L. Kim and R. Nelson (eds.), Technology, Learning and Innovation: Experiences of Newly Industrializing Economies. Cambridge, MA: Cambridge University Press. Kim, L. and Nelson, R. (2000). Technology, Learning and Innovation: Experiences of Newly Industrializing Economies. Cambridge, MA: Cambridge University Press. Klapper, L., Laeven, L. and Rajan, R. (2004). ‘Business Environment and Firm Entry: Evidence from International Data’, World Bank Policy Research Working Paper, 3232. Klepper, S. (1996). ‘Entry, Exit, Growth and Innovation Over the Product Life Cycle’, American Economic Review, 86(3): 562–83. —— (2002a). ‘The Capabilities of New Firms and the Evolution of the U.S. Automobile Industry’, Industrial and Corporate Change, 11(4): 645–666.

295

Bibliography Klepper, S. (2002b). ‘The Evolution of the Automobile Industry and Detroit as its Capital’, Department of Social and Decision Sciences, Carnegie Mellon University, Working Paper. —— (2003). ‘The Geography of Organizational Knowledge’, Carnegie Mellon University, Working Paper. —— (2004). ‘Agglomeration Through Spinoffs: How Detroit Became the Capital of the U.S. Automobile Industry’, Carnegie Mellon University, Working Paper. —— and Sleeper, S. (2002). Entry by Spinoffs, mimeo. Pittsburgh, PA: Carnegie Mellon University. Klimenko, M. M. (2004). ‘Competition, Matching, and Geographical Clustering at Early Stages of the Industry Life Cycle’, Journal of Economics and Business, 56(3): 177–95. Knight, F. H. (1921). G. J. Stigler (ed.), Risk, Uncertainty and Profit . Chicago, IL: University of Chicago Press. Kolko, J. (2001). ‘Silicon Mountains, Silicon Molehills: Geographic Concentration and Convergence of Internet Industries in the US’, WIDER DP 2001/02, UN University, 14(2): 211–32. Kortum, S. and Lerner, J. (1998). ‘Stronger Protection or Technological Revolution: What is Behind the Recent Surge in Patenting?’, Carnegie-Rochester Conference Series on Public Policy, 48: 247–304. Koszarski, R. (1990). An Evening’s Entertainment: The Age of the Silent Feature Picture, 1915–1928. New York: Charles Scribners’ Sons. Kraemer, K. and Dedrick, J. (2002). ‘Dell Computer: Organisation of a Global Production Network’, Center for Research on Information Technology and Organizations, University of California, Irvine. Krugman, P. (1983). ‘Targeted Industrial Policies: Theory and Evidence’, typescript. —— (1991a). Geography and Trade. Cambridge, MA: MIT Press. —— (1991b). ‘History and Industry Location: The Case of the Manufacturing Belt’, The American Economic Review, 81: 80–83. —— (1991c). ‘History versus Expectations’, Quarterly Journal of Economics, 106: 651–67. —— (1991d ). ‘Increasing Returns and Economic Geography’, The Journal of Political Economy, 99: 483–99. —— (1993). ‘First Nature, Second Nature, and Metropolitan Location’, Journal of Regional Science, 33: 129–44. —— (1994). ‘Complex Landscapes in Economic Geography’, The American Economic Review, 84: 412–16. —— (1996). The Self-Organizing Economy. Cambridge, MA: Blackwell. —— and Venables, A. J. (1995a). ‘The Seamless World: A Spatial Model of International Specialization’, CEPR, Discussion Paper, No. 1230, London. —— —— (1995b). ‘Globalization and the Inequality of Nations’, Quarterly Journal of Economics, 110: 857–80.

296

Bibliography —— —— (1996). ‘Integration, Specialization and Adjustment’, European Economic Review, 40: 959–68. Lach, S. (2000). ‘Do R&D Subsidies Stimulate or Displace Private R&D? Evidance From Israel’, Neaman Institute, Working Paper. Lai, H.-C. and Shyu J. Z. (2005). ‘A Comparison of Innovation Capacity at Science Parks Across the Taiwan Strait: The Case of Zhangjiang High-Tech Park and Hsinchu Science-Based Industrial Park’, Technovation, 25(7): 805–13. Lall, S. and Albadalejo, M. (2003). ‘China’s Competitive Performance: a threat to East Asian Manufactured Exports?’, typescript. Oxford: Queen Elizabeth House. Lambooy, J. G. and Boschma, R. A. (2001). ‘Evolutionary Economics and Regional Policy’, The Annals of Regional Science, 35: 113–31. Lamoreaux, N. R. and Sokoloff, K. L. (1997). ‘Location and Technological Change in the American Glass Industry During the Late Nineteenth and Early Twentieth Centuries’, Working Paper, 5938. Lampel, J. (2001). ‘Show-and-Tell: Product Demonstrations and Path Creation of Technological Change’, in R. Garud and P. Karnoe (eds.), Path Dependence and Creation. New York: Lawrence Erlbaum. Langlois, R. N. (1990). ‘Creating External Capabilities: Innovation and Vertical Disintegration in the Microcomputer Industry’, Business and Economic History, 19: 93–102. —— (2003). ‘The Vanishing Hand: The Changing Dynamics of Industrial Capitalism’, Industrial and Corporate Change, 12(2): 351–85. Lau, L. T., Qian, Y., and Roland, G. (2000). ‘Reform without losers: An interpretation of China’s Dual Track Approach to Transition’, Journal of Political Economy, 108(1): 120–43. Leachman, R. C. and Leachman, C. H. (2004). ‘Globalization of Semiconductors: Do Real Men Have Fabs, or Virtual Fabs?’, in M. Kenney and R. Florida (eds.), Locating Global Advantage: Industry Dynamics in a Globalizing Economy. Stanford, CA: Stanford University Press. Leavitt, B. and March, J. G. (1988). ‘Organizational Learning’, Annual Review of Sociology, 14: 319–40. Lee, C. M., Miller, W. F., Hancock, M. G., and Rowen, H. S. (2000). The Silicon Valley Edge: A Habitat for Innovation and Entrepreneurship. Stanford, CA: Stanford University Press. Lemarie`, S., Mangematin, V., and Torre, A. (2001). ‘Is the Creation and Development of Biotech SMEs Localised? Conclusions Drawn From the French Case’, Small Business Economics, 17: 61–76. Lerner, J. (1998). ‘Angel Financing and Public Policy: An Overview’, Journal of Banking and Finance, 22: 773–783. —— (1999). ‘The Government as Venture Capitalist: The Long-Run Impact of the SBIR Program’, Journal of Business, 72(3): 285–318. —— (2002). ‘When Bureaucrats Meet Entrepreneurs: The Design of Effective ‘Public Venture Capital’ Program’, The Economic Journal, 112: 73–84.

297

Bibliography Leslie, S. (2000). ‘The Biggest ‘Angel’ of Them All’, in M. Kenney (ed.), Understanding Silicon Valley: The Anatomy of an Innovative Region. Stanford, CA: Stanford University Press, pp. 48–70. Lewis, M. (2000). The New New Thing: A Silicon Valley Story. New York: W. W. Norton. Lindblom, C. (1959). ‘The Science of ‘Muddling Through’ ’, Public Administration Review, 19: 79–88. Loewendahl, H. (2001). ‘A Framework for FDI Promotion’, Transnational Corporations, 10(1): 1–42. Lorenz, E. and Valeyre, A. (2004). ‘Organisational Change in Europe: National Models or the Diffusion of a New ‘One Best Way’?’, DRUID Working Paper, 4. Lotka, A. J. (1925). Elements of Physical Biology. Baltimore, MD: Williams & Wilkins. Link, A. (1995). Generosity of Spirit. The Early History of the Research Triangle Park. Research Triangle Park, NC: Research Triangle Foundation of North Carolina. Liu, X. L. and White, S. (2001). ‘Comparing Innovation Systems: A Framework and Application to China’s Transitional Context’, Research Policy, 30: 1091–114. Lu, Q. and Lazonick, W. (2001). ‘The Organization of Innovation in a Transitional Economy: Business and Government in Chinese Electronic Publishing’, Research Policy, 30: 55–77. Lundvall, B. (1992). National Systems of Innovation: Towards a Theory of Innovation and Interactive Learning. London: Pinter Publishers. LvHan Biotech (2004). ‘Personal Interview’, March 4, 2004. Lyons, D. (1995). ‘Agglomeration Economies Among High Technology Firms in Advanced Production Areas: the Case of Denver/Boulder’, Regional Studies, 29(3): 265–78. Macdonald, L. (2000). Nortel Networks: How Innovation and Vision Created a Network Giant. Toronto: John Wiley & Sons. Macneil, I. R. (1978). ‘Contracts—Adjustment of Long-Term Economic Relations under Classical, Neoclassical, and Relational Contract Law’, Northwestern University Law Review, 72: 854–905. MacSharry, R. and White, P. (2000). The Making of the Celtic Tiger: The Inside Story of Ireland’s Booming Economy. Dublin: Mercier Press. Maggioni, M. A. (1993). ‘Ecological Dynamics and Critical Mass Processes in the Location of High-Tech Firms’, Paper presented at the 40th RSAI (Regional Sciences Association International) Conference, North American Section, Houston, November 11–14. —— (2002a). Clustering Dynamics and the Location of High-Tech Firms. Heidelberg and New York: Springer-Verlag. —— (2002b). ‘The Development of High-Tech Clusters: Theoretical Insights and Policy Implications’, in M. Feldman and N. Massard (eds.), Institutions and Systems in the Geography of Innovation. Dordrecht: Kluwer. —— (2004a). ‘High-tech Firms’ Location and The Development of Innovative Industrial Cluster: A Survey of the Literature’, Economia Politica, n.1, a. XXI, n.1: 127–65.

298

Bibliography —— (2004b). ‘The Rise and Fall of Industrial Clusters Technology and the Life Cycle of Regions’, IEB Document de Treball 2004/6, Universidad de Barcelona. —— (2005a). ‘Of Flies and Elephants: Some Observations on Enterprise Policy Design and Implementation’, in M. Baussola and F. Timpano (eds.), Economic Development and Local Policies. Milano: Giuffre`, pp. 102–4. —— (2005b). ‘High-Tech Firms and Innovative Industrial Clusters’, in C. Karlsson (ed.), Handbook of Research on Clusters: Theories, Policies and Case Studies. Edward Elgar, forthcoming. Mahoney, J. (2000). ‘Path Dependence in Historical Sociology’, Theory and Society, 29: 507–48. Malecki, E. (1980). ‘Dimensions of R&D Location in the United States’, Research Policy, 9: 2–22. Malerba, F. (1993). ‘The National System of Innovation: Italy’, in R. Nelson (ed.), National Innovation Systems: A Comparative Analsys. New York–Oxford: Oxford University Press. —— and Orsenigo, L. (1997). ‘Technological Regimes and Sectoral Patterns of Innovative Activities’, in G. Dosi, F. Malerba, and L. Orsenigo (eds.), Industrial and Corporate Change, Special Issue on Technological Regimes and the Evolution of Industrial Structures, pp. 83–118. —— —— (2000). ‘Knowledge, Innovative Activities and Industry Evolution’ (con F. Malerba), in P. Cohendet and W. E. Steinmueller (eds.), Industrial and Corporate Change, Special Issue on The Codification of Knowledge, 9(2): 289–314. —— Nelson, R., Orsenigo, L., and Winter, S. (1999). ‘A History-Friendly Model of the Evolution of the Computer Industry’, Industrial and Corporate Change, 8(1): 3–40. Mangematin, V., Lemarie`, S., Boisson, J.-P., Catherine, D., Corolleur, C., Coronini, R., and Trommettier, M. (2003). ‘Development of SMEs and Heterogeneity of Trajectories: the case of Biotechnology in France’, Research Policy, 42: 621–38. Mariani, M. (2001). ‘The Location of R&D and the Network of Inventors in the Chemical and Bio-Pharmaceutical Sectors’, EPRIS Project Working Paper. Markusen, A. (1996). ‘Sticky Places in Slippery Space. A Typology of Industrial Districts’, Economic Geography, 72: 293–313. Marshall, A. (1890). Principles of Economics. London: Macmillan. —— (1923). Industry and Trade. London: Macmillan. Martin, R. and Sunley, P. (2003). ‘Deconstructing Clusters: Chaotic Concept or Policy Panacea?’, Journal of Economic Geography, 3(1): 5–35. Maskell, P. (2001). ‘Towards a Knowledge-Based Theory of the Geographical Cluster’, Industrial and Corporate Change, 10(4): 921–43. —— and Lorenzen, M. (2003). ‘The Cluster as Market Organization’, DRUID Working Paper, 14. —— and Malmberg, A. (1999). ‘Localized Learning and Industrial Competitiveness’, Cambridge Journal of Economics, 23: 167–86.

299

Bibliography Maude, A. (2004). ‘Regional Development Processes and Policies in Australia: A Review of Research 1990–2002’, European Planning Studies, 12(1): 3–26. Mayer, D. and Kenney, M. (2004). ‘Ecosystems and Acquisition Management: Understanding Cisco’s Strategy’, Innovation and Industry, forthcoming. Maynard Smith, J. (1974). Models in Ecology. Cambridge, MA: Cambridge University Press. Mazzoleni, R. and Nelson, R. (1998). ‘The Benefits and Costs of Strong Patent Protection: A Contribution to the Current Debate’, Research Policy, 27: 273–84. McKelvey, M. D. (1996). Evolutionary Innovations: The Business of Biotechnology. Oxford: Oxford University Press. McKendrick, D. G. (1998). ‘Dispersed Concentration: Industry Location and Globalization in Hard Disk Drives’, Report 98–03, The Information Storage Industry Center, University of California at San Diego. —— (2001). ‘Global Strategy and Population Level Learning: The Case of Hard Disk Drives’, Strategic Management Journal, 22: 307–34. —— Doner, R., and Haggard, S. (2000). From Silicon Valley to Singapore: Location and Competitive Advantage in the Hard Disk Drive Industry. Stanford, CA: Stanford University Press. —— Jaffee, J., Carroll, G. R., and Khessina, O. M. (2000). ‘In the Bud? Disk Array Producers as a (Possibly) Emergent Organizational Form’, Information Storage Industry Center Report 2000–04. San Diego, CA: University of California. Merges, R. P. and Nelson, R. (1994). ‘On Limiting or Encouraging Rivalry in Technical Progress: The Effects of Patent Scope Decisions’, Journal of Economic Behavior and Organization, 25: 1–24. Merton, D. (1973). The Sociology of Science: Theoretical and Empirical Investigation. Chicago, IL: University of Chicago Press. Metcalfe, S. (1995). ‘The Economics Foundations of Technology Policy: Equilibrium and Evolutionary Perspective’, in P. Stoneman (ed.), Handbook of Economics of Innovation and Technological Change. Oxford, UK and Cambridge, USA: Blackwell. Mezias, J. M. and Mezias, S. J. (2000). ‘Resource Partitioning, the Founding of Specialist Firms, and Innovation: the American Feature Film Industry, 1912– 1929’, Organization Science, 11: 306–22. —— and Kuperman, J. C. (2000). ‘The Community Dynamics of Entrepreneurship: the Birth of the American Film Industry, 1895–1929’, Journal of Businbess Venturing, 16: 209–33. Midelfart, K.-H., Overman, H. G., Redding, S. J., and Venables, A. J. (2000). ‘The Location of European Industry’, European Commission Economic Papers, 142. Milgram S. (1967). ‘The Small World Problem’, Psychology Today, 2: 60–67. Ministry of Science and Technology (2001). ‘Joint Office of ‘‘863’’ Plan: The 15-Years’ Achievement of the National ‘‘863’’ Plan’, Biotech Volume, http:// www.863.org.cn/863_95/biology/bly04_001.html

300

Bibliography —— (2003a). ‘The Development of mingying keji qiye in the 1990s’, http:// www.mykj.gov.cn —— (2003b). The China Science and Technology Statistics Center, www.sts.org.cn/ tjbg/qtzx/documents/2003/0321.htm and http://gov.finance.sina.com.cn/zsyz/ 2004-09-09/25820.html Ministry of Science and Technology China Biotech Development Center (2003). Database of Chinese Biotech Research Organizations and Researchers: www.biotech.og.cn; www.bio-engine.com; www.genetide.com; www.cnm21.com; and www.yahoo.com.cn Miyao, T. and Kanemoto, Y. (1987). Urban Dynamics and Urban Externalities. London: Harwood Academic Publishers. Moore, G. and Davis, K. (2001). ‘Learning the Silicon Valley Way’, Stanford Institute for Economic Policy Research Discussion Paper, 00/45. Mowery, D. C., Nelson, R. R., Sampat, B., and Ziedonis, A. (2001). ‘The Growth of Patenting and Licensing by US Universities: An Assessment of the Effects of the Bayh-Dole Act of 1980’, Research Policy, 30: 99–119. Murtha, T. P., Lenway, S. A., and Hart, J. A. (2001). Managing New Industry Creation: Global Knowledge Formation and Entrepreneurship in High Technology. Stanford, CA: Stanford University Press. Musser, C. (1994). The Emergence of Cinema: the American Screen to 1907. Berkeley, CA: University of California Press. Myrdal, G. (1957). Economic Theory and Under-Developed Regions. New York: Harper & Row. Nadvi, K. (1995). ‘Industrial Clusters and Networks: Case Studies of SME Growth and Innovation’, UNIDO, Vienna. —— and Schmitz, H. (1997). ‘SME Responses to Global Challenges: Case Studies of Private and Public Initiatives’, Paper presented at Seminar on New Trends and Challenges in Industrial Policy, UNIDO, Vienna, October. National Research Council (1999a). Funding a Revolution: Government Support and Computing Research. Washington, DC: National Academy Press. —— (1999b). The Small Business Innovation Research Program: Challenges and Opportunities. Washington, DC: National Academy Press. —— (2003). Government–Industry Partnerships for the development of new technologies. Washington, DC: National Academy Press. Nelles, J., Bramwell, A., and Wolfe, D. A. (2005). ‘History, Culture and Path Dependency: Origins of the Waterloo ICT Cluster’, in D. A. Wolfe and M. Lucas (eds.), Global Networks and Local Linkages: The Paradox of Cluster Development in an Open Economy. Montreal and Kingston: McGill-Queen’s University Press for the School of Policy Studies and Queen’s University, pp. 227–52. Nelson, R. (1981). ‘Assessing Private Enterprise—an Exegesis of Tangled Doctrine’, Bell Journal of Economics, 12: 93–111. Nelson, R. (1986). ‘Institutions Supporting Technical Advance in Industry’, American Economic Review, 76: 186–9.

301

Bibliography Nelson, R. (1991). National Innovation Systems: A Comparative Analysis. Oxford: Oxford University Press. —— (1994). ‘The Co-Evolution of Technology, Industrial Structure and Supporting Institutions’, Industrial and Corporate Change, 3(1): 47–63. —— (1995). ‘Recent Evolutionary Theorizing about Economic Change’, Journal of Economic Literature, 33: 48–90. —— and Winter, S. G. (1982). An Evolutionary Theory of EconomicChange. Cambridge, MA: Harvard University Press. Newgenco Bio Science (2004). ‘Personal Interview’, March 11, 2004. Nielson, M. and Mailes, G. (1995). Hollywood’s Other Blacklist: Union Struggles in the Studio System. London: British Film Institute. Nijkamp, P. and Reggiani, A. (1992). Interaction, Evolution and Chaos in Space. Heidelberg and New York: Springer-Verlag. Nilsson, M., Svensson-Henning, M., and Wilkenson, O. (2002). Ska˚nska kluster och ¨ : Na¨ringsutveckling Region Ska˚ne. profilomra˚den—en kritisk granskning. Malmo Niosi, J. and Bas, T. G. (2001). ‘The Competencies of Regions. Canada’s Clusters in Biotechnology’, Small Business Economics, 17: 31–42. —— —— (2003). ‘The Competencies of Regions. Canada’s Clusters in Biotechnology’, in G. Fuchs (ed.), Biotechnology in Comparative Perspective. London and New York: Routledge. Nocke (2003). ‘A Gap for Me: Entrepreneurs and Entry’, Penn Institute for Economic Research, Working Paper, mimeo. North, C. J. (1926). ‘Our Foreign Trade in Motion Pictures’, Annals of the American Academy of Political and Social Science, 128: 100–108. North, D. N. (1990). Institutions, Institutional Change and Economic Performance Cambridge, MA: Cambridge University Press. NVCA (National Venture Capital Association) (2000a). ‘Venture Capital Funds Raise A Record $46.55 Billion in 1999’, March 27. —— (2000b). 2000 National Venture Capital Association Yearbook. Washington, DC: National Venture Capital Association. O’Dell, P. (1970). Griffith and the Rise of Hollywood. New York: A. Zwemmer Limited—A. S. Barnes. OECD (2002). Information Technology Outlook. Paris: OECD. Office of Management and Budget, Bulletin 03/04 (2003). O’Gorman, C., E. O’Malley and J. Mooney (1997). ‘The Irish Indigenous Software Industry: An Application of Porter’s Cluster Analysis’, Research Series Paper No. 3, Dublin: National Economic and Social Council. OhUallachain, B. (1997). ‘Restructuring the American Semiconductor Industry: Vertical Integration of Design Houses and Wafer Fabricators’, Annals of the Association of American Geographers, 87(2): 217–37. O’Malley, E. and Van Egeraat, C. (2000). ‘Industry Clusters and Irish Indigenous Manufacturing: Limits of the Porter View’, Economic and Social Review, 31(1): 55–79.

302

Bibliography ´ Riain, S. (1997). ‘An Offshore Silicon Valley? The Emerging Irish Software IndusO try’, Competition and Change, 2: 175–212. Orsenigo, L. (1989). The Emergence of Biotechnology. Institutions and Markets in Industrial Innovation. London: Pinter Publishers. —— (2001). ‘The (failed?) Development of a Biotechnology Cluster. The Case of Lombardy’, Small Business Economics. —— Pammolli, F., and Riccaboni, M. (2000). ‘Technological Change and Network Dynamics. The Case of the Bio-pharmaceutical Industry’, Research Policy. —— —— —— (2001). ‘Technological Change and Network Dynamics—Lessons from the Pharmaceutical Industry’, Research Policy, 30: 485–508. Ottaviano, G. I. P. and Puga, D. (1998). ‘Agglomeration in the Global Economy: A Survey of the ‘New Economic Geography’ ’, World Economy, 21: 707–31. —— and Thisse, J.-F. (1998). ‘Agglomeration and Trade Revisited’, CEPR and CORE Discussion Paper, 1903, Louvain. —— and Powell, W. W. (2004). ‘Knowledge Networks in the Boston Biotechnology Community’, Organization Science, 15: 5–21. —— Riccaboni, M., Pammolli, F., and Powell, W. W. (2002). ‘A Comparison of U.S. and European University-Industry Relations in the Life Sciences’, Management Science, 48(1): 24–43. PaiteBoen Biotech Development (2004). ‘Personal Interview’, March 25, 2004. ¨ resundsregionen—Bakgrund, verklighet, mo¨jPalludan, U. and Persson, H. (2003). O ligheter. Copenhagen: The Copenhagen Institute for Future Studies. Palmer, E. O. (1937). History of Hollywood. Hollywood, FL: Edwin O. Palmer Publisher. Pammolli, F. and Riccaboni, M. (2001). ‘Geographical Clusters in the Biotechnology Industry’, University of Siena, EPRIS Working Paper. Pan, A. (2004). ‘China’s Capital Markets and Biotech Industry’ in High-tech Industry Department of National Development and Reform Commission, and China Biotechnology Engineering Association (ed.), 2003 Report of China’s Biotechnology Development. Beijing: Chemical Industry Press. —— and Danqing, L. (2004). ‘The Current Situation of China’s Bio-pharmaceutical Industry’, in High-tech Industry Department of National Development and Reform Commission and China Biotechnology Engineering Association (ed.), 2003 Report of China’s Biotechnology Development. Beijing: Chemical Industry Press. Pan (2004). ‘Personal Interview with Pan, SHDNA Biotech’, March 9, 2004. Paniccia, I. (1998). ‘One, a Hundred, Thousand of Industrial Districts. Organizational Variety in Local Network of Small and Medium-sized Enterprises’, Organization Studies, 19(4): 667–99. Papageorgiou, Y. Y. (1979). ‘Agglomerations’, Regional Science and Urban Economics, 9: 41–9. —— and Smith, T. (1983). ‘Agglomeration and Local Instability of Spatially Uniform Steady-States’, Econometrica, 51: 1109–19.

303

Bibliography Parhankangas, A. and Arenius, P., ‘From a Corporate Venture to an Independent Company: A Base for a Taxonomy for Corporate Spin-Off Firms’, Research Policy, 32(3): 463–81. Pascal, A. H. and McCall, J. J. (1980). ‘Agglomeration Economies, Search Costs and Industrial Location’, Journal of Urban Economics, 8: 383–8. Pearl, R. and Reed, L. J. (1920). ‘On the Rate of Growth of the Population of the United States since 1790 and its Mathematical Representation’, Proceedings of the National Academic Sciences, 6: 275–88. —— —— (1925). ‘Skew-Growth Curves’, Proceeding of the National Academy of Natural Sciences of the USA, 11: 16–22. Perez, C. (2003). Technological Revolutions and Financial Capital: The Dynamics of Bubbles and Golden Ages. Cheltenham, UK: Edgar Elgar. Perkins, A. (2000). ‘Investors: Brace Yourselves for the Next Bubble Bath’, Red Herring, November 13: 21–2. ˆ le de croissance’, Economie Appliqueé, 8: Perroux, F. (1955). ‘Note sur la notion de po 307–20. Piore, M. and Sabel, C. (1984). The Second Industrial Divide: Possibilities for Prosperity. New York: Basic Books. Pisano, G. P. (1991). ‘The Governance of Innovation. Vertical Integration and Collaborative Arrangements in the Biotechnology Industry’, Research Policy, 20: 237–49. Pitts, M. R. (1997). Poverty Row Studios, 1929–1940. Jefferson, NC: McFarland. Polanyi, M. (1966). The Tacit Dimension. New York: Doubleday. Polli, R. and Cook, V. (1969). ‘Validity of the Product Life Cycle’, The Journal of Business, 42(4): 385–400. Ponti, J. (1992). Hollywood East. Orlando, FL: Tribune Publishing. Porter, K. (2004). You Can’t Leave Your Past Behind: The Influence of Founder’s Career Histories on their Firms. Ph.D. dissertation. Stanford, CA: Stanford University. —— Whittington, B., and Powell, W. W. (2005). ‘The Institutional Embeddedness of High-Tech Regions: Relational Foundations of the Boston Biotechnology Community’, in S. Bresnehan and F. Malerba (eds.), Clusters, Networks, and Innovation. London: Oxford University Press. Forthcoming. Porter, M. (1990). The Competitive Advantage of Nations. New York: Free Press. —— (1998a). ‘Clusters and the New Economics of Competition’, Harvard Business Review, 77–90. —— (1998). ‘Clusters and the New Economics of Competition’, Harvard Business Review, Nov.–Dec.: 77–90. —— (1998b). ‘Clusters and Competition: New Agendas for Companies, Governments, and Institutions’, in M. Porter (ed.), On Competition. Cambridge, MA: Harvard Business Review Books, pp. 197–287. Porter, M., Monitor Group (2001). Clusters of Innovation: Regional Foundations of US Competitiveness. Washington, DC: Council on Competitiveness. Powell, W. W. (1990). ‘Neither Market nor Hierarchy—Network Forms of Organization,’ Research in Organizational Behavior, 12: 295–336.

304

Bibliography —— (1996). ‘Interorganizational Collaboration in the Biotechnology Industry’, Journal of Institutional and Theoretical Economics, 152(1): 197–215. —— and Brantley, P. (1992). ‘Competitive Cooperation in Biotechnology: Learning Through Networks?’, in R. Eccles and N. Nohria (eds.), Networks and Organizations. Boston, MA: Harvard University Press. —— —— (1996). ‘Magic Bullets and Patent Wars: New Product Development in the Biotechnology Industry’, in T. Nishiguchi (ed.), Managing Product Development. New York: Oxford University Press. —— Koput, K., and Smith-Doerr, L. (1996). ‘Interorganizational Collaboration and the Locus of Innovation: Networks of Learning in Biotechnology’, Administrative Science Quarterly, 41(1): 116–45. —— —— —— and Owen-Smith, J. (1999). ‘Network Position and Firm Performance: Organizational Returns to Collaboration’, Research in the Sociology of Organizations, 16: 129–59. —— —— Bowie, J. I., and Smith-Doerr, L. (2002). ‘The Spatial Clustering of Science and Capital: Accounting for Biotech Firm-Venture Capital Relationships’, Regional Studies, 36(3): 291–306. —— White, D. R., Koput, K., and Owen-Smith, J. (2005). ‘Network Dynamics and Field Evolution: The Growth of Inter-Organizational Collaboration in the Life Sciences’, American Journal of Sociology, forthcoming. Prevezer, M. (1997). ‘The Dynamics of Industrial Clustering in Biotechnology’, Small Business Economics, 9. —— (2001). ‘Ingredients in the Early Development of the U.S. Biotechnology Industry’, Small Business Economics, 17: 17–29. —— (2003). ‘The Development of Biotechnology Clusters in the USA from the late 1970s to the Early 1990s’, in G. Fuchs (ed.), Biotechnology in Comparative Perspective. Routledge. —— Owen-Smith, J. and Colyvas, J. (2007). Innovation and Emulation: Lessons from the Experiences of U. S. Universities in Selling Private Rights to Public Knowledge. Minerva. Puga, D. and Venables, A. J. (1996). ‘The Spread of Industry: Spatial Agglomeration in Economic Development’, Journal of the Japanese and International Economies, 10: 440–64. Qi, C. Y. (2003). ‘Driving the Development of China’s Biotechnology by Clustering’, in High-tech Industry Department of National Development & Reform Commission and China Biotechnology Engineering Association (ed.), Report of China’s Biotechnology Development. Beijing: Chemical Industry Press. Qian, Y. (2003). ‘How Reform Worked in China’, in D. Rodrik (ed.), In Search of Prosperity: Analytic Narratives on Economic Growth. Princeton, NJ: Princeton University Press, 297–333. Qiu, H. (2003). ‘Who Abandons the United Gene’, 21st Century Economic Report, March.

305

Bibliography Quah, D. (2001). ‘ICT Clusters in Development: Theory and Evidence’, EIB Papers, 6(1): 85–100. Quirmbach, H. C. (1986). ‘The Diffusion of New Technology and the Market for an Innovation’, Rand Journal of Economics, 17: 33–47. Rabellotti, R. (1998). ‘Helping Small Firms to Network—The Experience of UNIDO’, Small Enterprise Development, March. Rallet, A. and Torre, A. (1999). ‘Is Geographical Proximity Necessary in the Innovation Networks in the Era of Global Economy?’, GeoJournal, 49: 373–80. Ramsaye, T. (1926). A Million and One Nights; a History of the Motion Picture. New York, NY: Simon & Schuster. Rauch, J. E. (1993). ‘Does History Matter only when it Matters Little? The Case of City-Industry Location’, Quarterly Journal of Economics, 108: 843–67. Reinganum, J. F. (1981). ‘On the Diffusion of New Technology: A Game Theoretic Approach’, Review of Economic Studies, XLVIII: 395–405. Richardson, H. W. (1978). Regional and Urban Economics. Harmondsworth, UK: Penguin Books. Robbins-Roth, C. (2001). From Alchemy to IPO: The Business of Biotechnology. New York: HaperCollins. Roberts, B. H. and Enright, M. J. (2004). ‘Industry Clusters in Australia: Recent Trends and Prospects’, European Planning Studies, 12(1): 99–121. Roberts, E. (1991). Entrepreneurs in High Technology. Lessons from MIT and Beyond. Oxford: Oxford University Press. Robinson, D. (1968). Hollywood in the Twenties. New York: A. S. Barnes. —— (1977). ‘The Twenties’, in P. Cowie (ed.), Hollywood, 1920–1970. New York: A. S. Barnes. Rocco, L. (2003). Externalities in the Economics of Science, mimeo. Milan: University of Milan Bicocca. Romanelli, E. (1989). ‘Organization Birth and Population Variety: A Community Perspective on Origins’, in B. Staw and L. L. Cummings (eds.), Research in Organizational Behavior, vol. 11. Greenwich, CT: JAI Press, pp. 211–46. —— and Schoonhoven, C. B. (2001). ‘The Local Origins of New Firms’, in C. B. Schoonhoven and E. Romanelli (eds.), The Entrepreneurship Dynamic: Origins of Entrepreneurship and the Evolution of Industries. Stanford, CA: Stanford University Press. Rosenberg, N. and Nelson, R. (1994). ‘American Universities and Technical Advance Industry’, Research Policy, 23: 323–48. Rosenkopf, L. and Almeida, P. (2003). ‘Overcoming Local Search Through Alliances and Mobility’, Management Science, 49: 751–66. Ross, M. (1941). Stars and Strikes: Unionization of Hollywood. New York: Columbia University Press. —— (1947). ‘Labor relations in Hollywood’, Annals of the American Academy of Political and Social Science, 254: 58–64. Roughgarden, J. (1979). Theory of Population Genetics and Evolutionary Ecology: An Introduction. New York and London: Macmillan.

306

Bibliography Saxenian, A. (1994). Regional Advantage. Culture and Competition in Silicon Valley and Route 128. Cambridge, MA: Harvard University Press. —— (1998). Regional Development: Silicon Valley and Route 128. Cambridge, MA: Harvard University Press. Scarpetta, S., Hemmings, P., Tressel, T., and Woo, J. (2002). ‘The Role of Policy and Institutions for Productivity and Firm Dynamics: Evidence from Micro and Industry Data’, OECD Working Paper, 329 ECO/WKP. Schatz, T. (1988). The Genius of the System: Hollywood Film-Making in the Studio Era. New York: Pantheon. Schumpeter, J. (1934). The Theory of Economic Development. Cambridge, MA: Harvard University Press. Scott, A. J. (1988). New Industrial Spaces: Flexible Production Organization and Regional Development in North America and Western Europe. London: Pion. —— (1993). Technopolis: High-Technology Industry and Regional Development in Southern California. Berkeley, CA: University of California Press. —— (2002). ‘A New Map of Hollywood: the Production and Distribution of American Motion Pictures’, Regional Studies, 36(9): 957–75. —— (2005). On Hollywood: The Place, the Industry. Princeton, NJ: Princeton University Press. —— and Storper, M. (1987). ‘High Technology Industry and Regional Development: A Theoretical Critique and Reconstruction’, International Social Science Journal, 39: 215–32. Sea Forrest Biotech (2004). ‘Personal Interview’, March 4. SEMI (1986). Semiconductor Industry Genealogy. San Jose: SEMI. Sharpe, M. (1991). ‘Pharmaceuticals and Biotechnology: Perspectives for the European Industry’, in C. Freeman (ed.), Technology and the Future of Europe. London: Pinter. Shindler, C. (1996). Hollywood in Crisis: Cinema and American Society, 1929–1939. London: Routledge. Shisheng Cell Technology (2004). ‘Personal Interview’, February 26. Sklar, R. (1975). Movie-Made America: A Social History of American Movies. New York: Random House. Slide, A. (1994). Early American Cinema. Metuchen, NJ: Scarecrow Press. Solomon, J. (1988). ‘Personal Interview by Richard Florida and Martin Kenney’, March 31. ¨ ., Lindqvist, G., and Ketels, C. (2003). The Cluster Initiative Greenbook. ¨ lvell, O So Stockholm: IvorytowerAB. Sorenson, O. (2003). ‘Social Networks and Industrial Geography’, Journal of Evolutionary Economics, 13(5): 513–27. —— and Audia, P. G. (2000). ‘The Social Structure of Entrepreneurial Activity: Footwear Production in the United States, 1940–1989’, American Journal of Sociology, 106: 424–62. —— and Stuart, T. E. (2001). ‘Syndication Networks and the Spatial Distribution of Venture Capital Investments’, American Sociological Review, 106(6): 1546–88.

307

Bibliography Soskice, D. (1997). ‘German Technology Policy, Innovation and National Institutional Frameworks’, Industry and Innovation, 4: 75–96. Spencer, R. V. (1911). ‘Los Angeles as a Producing Center’, Moving Picture World, 8(14): 768. Spender, J. C. (1989). Industry Recipes. Oxford: Blackwell. Staber, U. (1997). ‘An Ecological Perspective on Entrepreneurship in Industrial Districts’, Entrepreneurship and Regional Development, 9: 45–64. Staiger, J. (1982). ‘Dividing Labor for Production Control: Thomas Ince and the Rise of the Studio System’, in G. Kindem (ed.), The American Movie Industry: The Business of Motion Pictures. Carbondale: University of Southern Illinois Press. Steinle, C. and Schiele, H. (2002). ‘When Do Industries Cluster?: A Proposal on How to Assess an Industry’s Propensity to Concentrate at a Single Region or Nation’, Research Policy, 31(6): 849–58. Stern, S. (1945). Birth of a Nation (London: British Film Institute (Special Supplement to Sight and Sound, Index Series, No 4).). Stinchcombe, A. L. (1965). ‘Social Structures and Organizations’, in J. G. March (ed.), Handbook of Organizations. Chicago, IL: Rand McNally. —— (1986). Stratification and Organization. Cambridge, MA: Cambridge University Press. Stoneman, P. (1987). The Economic Analysis of Technology Policy. Oxford: Clarendon Press. Storper, M. (1995). ‘The Resurgence of Regional Economies, Ten Years Later: The Region as a Nexus of Untraded Interdependencies’, European Urban and Regional Studies, 2(3): 191–221. —— (1999). ‘The Resurgence of Regional Economies: Ten Years Later’, in T. J. Barnes and M. S. Gertler (eds.), The New Industrial Geography: Regions, Regulation and Institutions. London: Routledge, pp. 23–53. —— and Christopherson, S. (1987). ‘Flexible Specialization and Regional Agglomerations: The Case of the U.S. Motion Picture Industry’, Annals of the Association of American Geographers, 77: 104–17. —— and Venables, A. J., (2002). ‘Buzz: The Economic Force of the City’, Paper presented at the DRUID Summer Conference on Industrial Dynamics of the New and Old Economy—Who is Embracing Whom? Copenhagen. —— and Walker, R. (1984). ‘The Spatial Division of Labour: Labour and the Location of Industries’, in L. Sawers and W. K. Tabb (eds.), Sunbelt/Snowbelt: Urban Development and Regional Restructuring. Oxford: Oxford University Press. —— —— (1989). The Capitalist Imperative. Territory, Technology and Industrial Growth. New York: Basil Blackwell. Streb, J. (2003). ‘Shaping the National System of Inter-Industry Knowledge Exchange: Vertical Integration, Licensing and Repeated Knowledge Transfer in the German Plastics Industry’, Research Policy, 32(6): 1125–40. Stuart, T. and Sorenson, O. (2003a). ‘The Geography of Opportunity: Spatial Heterogeneity in Founding Rates and the Performance of Biotechnology Firms’, Research Policy, 32(2): 229–53.

308

Bibliography —— —— (2003b). ‘Liquidity Events and the Geographic Distribution of Entrepreneurial Activity’, Administrative Science Quarterly, 48: 175–201. Sturgeon, T. (2000). ‘How Silicon Valley Came to Be’, in M. Kenney (ed.), Understanding Silicon Valley. Stanford, CA: Stanford University Press, pp. 15–47. Sturgeon, T. J. (2003). ‘What Really Goes on in Silicon Valley? Spatial Clustering and Dispersal in Modular Production Networks’, Journal of Economic Geography, 3(2): 199–225. Suarez-Villa, L. and Walrod, W. (1997). ‘Operational Strategy, R&D and IntraMetropolitan Clustering in a Polycentric Structure: The Advanced Electronics Industries of the Los-Angeles Basin’, Urban Studies, 34(9): 1343–80. Suchman, M. C. (1994). On Advice of Counsel: Law Firms and Venture Capitalist Funds as Information Intermediaries in the Structuration of Silicon Valley, Unpublished Ph.D. dissertation. Department of Sociology, Stanford, CA: Stanford University. —— (2000). ‘Dealmakers and Counselors: Law Firms as Intermediaries in the Development of Silicon Valley’, in M. Kenney (ed.), Understanding Silicon Valley. Stanford, CA: Stanford University Press. Sutton, J. (1998). Technology and Market Structure: Theory and History. Cambridge, MA: Massachusetts Institute of Technology Press. Swann, P. (1993). ‘Can High Technology Services Prosper if High Technology Manufacturing Doesn’t’, Centre for Business Strategy Working Paper Series, 143. —— and Prevezer, M. (1996). ‘A Comparison of the Dynamics of Industrial Clustering in Computing and Biotechnology’, Research Policy, 25: 1139–57. —— —— (1998). ‘Introduction’, in P. Swann, M. Prevezer, and D. Stout (eds.), The Dynamics of Industrial Clustering: International Comparisons in Computing and Biotechnology. Oxford: Oxford University Press. —— —— and Stout, D. (1998). The Dynamics of Industrial Clustering. International Comparisons in Computing and Biotechnology. Oxford: Oxford University Press. Tauchen, H. and Witte, A. D. (1983). ‘An Equilibrium Model of Office Location and Contact Patterns’, Environment and Planning A, 15: 1311–26. Teece, D. J. (1986). ‘Profiting from Technological Innovation: Implications for Integration, Collaboration, Licensing and Public Policy’, Research Policy, 15(6): 285–305. Teubal, M. (1993). ‘The Innovation System of Israel: Description, Performance and Outstanding Issues’, in R. Nelson (ed.), National Systems of Innovation: A Comparative Analysis. Oxford and New York : Oxford University Press. —— (1996). ‘R&D and Technology Policy at NIC’s as Learning Processes’, World Development, 24(3): 449–60. —— (1998). ‘Policies for Promoting Enterprise Restructuring in National System of Innovation: Triggering Cumulative Learning and Generating System Effects’, in STI Review, 22. Paris: OECD. —— (1999). ‘Towards an R&D Strategy for Israel’, The Economic Quarterly, 46(2): 359–83.

309

Bibliography Teubal, M. (2000). ‘What Is the Systems of Innovation (SI) Perspective to Innovation and Technology Policy (ITP) and How Can We Apply It to Developing and Industrialized Economies?’, Journal of Evolutionary Economics, 12: 233–57. —— (2002). ‘What Is the Systems of Innovation (SI) Perspective to Innovation and Technology Policy (ITP) and How Can We Apply it to Developing and Industrialized Economies?’, Journal of Evolutionary Economics, 12: 233–57. —— and Andersen, E. (2000). ‘Enterprise Restructuring and Embeddedness: A Policy and Systems Perspective’, Industrial and Corporate Change, 9(1): 87–111. —— and Avnimelech, G. (2003). ‘Foreign Acquisitions and R&D Leverage in HighTech Industries of Peripheral Economies: Lessons and Policy Issues from the Israeli Experience’, International Journal of Technology Management, 26(2–4): 362–85. The Economist (2004), ‘China’s Development String of Pearls’, November 20, 2004: 71. —— (2004), ‘Southern Comfort, Eastern Promise’, December 9, 2004. Thompson, K. (1985). Exporting Entertainment: America in the World Film Market, 1907–34. London: British Film Institute. Thirtle, C. G. and Ruttan, V. W. (1987). The Role of Demand and Supply in the Generation and Diffusion of Technical Change. Chur: Harwood Academic. Tian, L., and the DFID-LBS Chinese Team (2004). ‘Some Comments on the Template of Institutional Studies’, Private Correspondence, May 10. Todtling, F. (1994). ‘Regional Networks of High-Technology Firms—The Case of the Greater Boston Region’, Technovation, 14(5): 323–43. Torrence, B. T. (1982). Hollywood: the First Hundred Years. New York: Zoetrope. Tracey, P., Clark, G., and Smith, H. L. (2004). ‘Cognition, Learning and European Regional Growth: An Agent-Centred Perspective on the ‘‘New’’ Economy’, Economics of Innovation & New Technology, 13(1): 1–18. Trajtenberg, M. (1990). ‘A Penny For Your Quotes: Patent Citations and the Value of Information’, Bell Journal of Economics, 21: 172–87. —— (2001). ‘R&D Policy in Israel: An Overview and Reassessment’, Neaman Institute Working Papers. UNCTAD (1994). ‘Technological Dynamism in Industrial Districts: An Alternative Approach to Industrialisation in Developing Countries?’ UNCTAD/ITD/TEC/11. —— (1998). ‘Promoting and Sustaining SMEs, Clusters and Networks for Development’, Trade and Development Board TD/B/COM.3/EM.5/2. —— (2001). World Investment Report: Promoting Linkages. Geneva: UN. Ungermann, R. (1995). ‘Personal Communication with Urs von Burg and Martin Kenney’, April 26. Usher, A. P. (1954). A History of Mechanical Inventions. Boston, MA: Beacon Hill. Uzzi, B. (1997). ‘Social Structure and Competition in Inter-Firm Networks: The Paradox of Embeddedness’, Administrative Science Quarterly, 42: 35–67. Van Egeraat, C. and Jacobson, D. (2004). ‘The Rise and Demise of the Irish and Scottish Computer Hardware Industry’, European Planning Studies, forthcoming.

310

Bibliography Varmus, H. and Weinberg, R. A. (1992). Genes and the Biology of Cancer. New York, NY: Scientific American Library. Venables, A. J. (1996). ‘Equilibrium Locations of Vertically Linked Industries’, International Economic Review, 37: 341–59. Verhulst, P. F. (1845). ‘Recherches Mathe´matique sur la loi d’accroissement de la Population’, Nouveaux Me´moires de l’Acade´mie Royale des Sciences et Belles-Lettres de Bruxelles, 18(2): 3–38. Vernon, R. (1966). ‘International Investment and International Trade in the Product Life Cycle’, Quarterly Journal of Economics, 80: 190–207. Vettas, N. (1999). ‘Location as a Signal of Quality’, CEPR Working Paper, 2165. Vinnova, 2001. The Swedish Biotechnology Innovation System, Rapport 2001:2. Stockholm: Vinnova. Von Burg, U. (2001). The Triumph of Ethernet. Stanford, CA: Stanford University Press. —— and Kenney, M. (2003). ‘Sponsors, Communities, and Standards: Ethernet vs. Token Ring in the Local Area Networking Business’, Industry and Innovation, 10(4): 351–75. —— and —— (2000). ‘There at the Beginning: Venture Capital and the Creation of the Local Area Networking Industry’, Research Policy, 29(9): 1135–55. Von Hippel, E. (1988). The Sources of Innovation. Oxford: Oxford University Press. Walker, G., Kogut, B., and Shan, W. (1997). ‘Social Capital, Structural Holes and the Formation of an Industry Network’, Organization Science, 8(2): 109–25. Walker, R. (1985). ‘Technological Determination and Determinism: Industrial Growth and Location’, in M. Castells (ed.), High Technology, Space and Society. Beverly Hills, CA: Sage Publications. —— (1988). ‘The Geographical Organization of Production-Systems’, Environment and Planning D: Society and Space, 6: 377–408. —— (1998). ‘Personal Interview with Rob Walker’, July 9. http://silicongenesis. stanford.edu/transcripts/walker.htm Wanzing Biotech (2004). ‘Personal Interview’, March 1. Wasserman, S. and Faust, K. (1994). Social Network Analysis: Methods and Applications. Cambridge, UK: Cambridge University Press. Watts, D. and Strogatz, S. (1998). ‘Collective Dynamics of Small-world Networks’, Letters to Nature, 393: 440–42. Weber, A. (1929). Theory of the Location of Industry. Chicago, IL: Chicago University Press. Westphal, L. (1990). ‘Industrial Policy in an Export Propelled Economy’, Journal of Economic Perspectives, 4(3): 41–59. Whittington, B. K., Owen-Smith, J., and Powell, W. W. (2006). ‘Networks, Propinquity, and Innovation in Technological Communities’, Stanford University, Working Paper. Wilson, J. (1988). ‘Personal Interview by Richard Florida and Martin Kenney’, March 29.

311

Bibliography Wolfe, D. A. and Gertler, M. S. (2004). ‘Clusters from the Inside and Out: Local Dynamics and Global Linkages’, Urban Studies, 41(5–6): 1071–93. World Bank Group (2004). ‘Economy Comparative Data Doing Business’, http:// rru.worldbank.org/DoingBusiness/SnapshotReports/Country.aspx?regionid¼42 Xiao, L. (2002). ‘The Development of Mingying Economy’, China Economic Times, November 2. Xinhua News Agency (2004). ‘80% Scientists in Chinese Academy of Sciences Are Returnees’, www.xinhuanet.com, March 1. Xu, Pacific Biotech (2004). ‘Personal Interview’, February 26. Yamazaki, H. (2002). ‘The Evolution and Structure of Industrial Clusters in Japan’, Small Business Economics, 18(1–3): 121–40. Yang, P. (2001). ‘Industrial Policies in China: A Local Official’s Experience’, Strategic Management. Yeats, A. J. (2001). ‘Just How Big is Global Production Sharing?’, in S. W. Arndt and H. Kierzkowski (eds.), Fragmentation: New Production Patterns in the World Economy. Oxford: Oxford University Press. Yifei, S. (2002). ‘Sources of Innovation in China’s Manufacturing Sector: Imported or Developed in-house’, Environment and Planning, 34: 1059–72. Yla-Antilla, P. and Lemolla, T. (2003). ‘Transformation of Innovation System in a Small Country-the Case of Finland’, typescript. Young, A. A. (1928). ‘Increasing Returns and Economic Progress’, The Economic Journal, 38: 527–42. Young, J. (1988). Steve Jobs: The Journey is the Reward. Glenview, IL: Scott, Foreman and Company. Yu, N. (2004). ‘Materials Supplied by the NSBPIB, Presentation script for the introduction of NSBPIB’. Zhang, J. (2003). ‘Growing Silicon Valley on a Landscape: An Agent-Based Approach to High-Tech Industrial Clusters’, Journal of Evolutionary Economics, 13(5): 529–48. Zhang, J., Wu, Q. P. et al. (2004). ‘The Analysis of the Development of Biotech Industry in Guangdong’, in High-tech Industry Department of National Development & Reform Commission and China Biotechnology Engineering Association (eds.), 2003 Report of China’s Biotechnology Development. Beijing: Chemical Industry Press. Zhang, S. (2004). ‘Building Capital Channel, Accelerating the Development of China’s Biotech Industry’, in High-tech Industry Department of National Development and Reform Commission and China Biotechnology Engineering Association (eds.), 2003 Report of China’s Biotechnology Development. Beijing: Chemical Industry Press. Zhong Liu (2004). ‘Personal Interview, Proexact Biotech’, February 27. Ziaoyan (2004). AGTC Genetic Technology, ‘Personal Interview’, April 5. Zierer, C. M. (1947). Hollywood—World Center of Motion-picture Production’, Annals of the American Academy of Political and Social Science, 254: 12–17.

312

Bibliography Zollo, M. and Winter, S. (2002). ‘Deliberate Learning and the Evolution of Dynamic Capabilities’, Organization Science, 13(3): 339–51. Zook, M. A. (2002). ‘Grounded Capital: Venture Financing and the Geography of the Internet Industry, 1994–2000’, Journal of Economic Geography, 2(2): 151–77. Zucker, L. and Darby, M. (1995). ‘Costly Information in Firm Transformation, Exit, or Persistent Failure’, NBER Working Paper, 5577. —— —— (1996). ‘Star Scientists and Institutional Transformation: Patterns of Invention and Innovation in the Formation of the U.S. Biotechnology Industry’, Proceedings of the National Academy of Science, 93: 706–17. —— —— (1997). ‘Present at the Biotechnological Revolution: Transformation of Technological Identity for a Large Incumbent Pharmaceutical Firm’, Research Policy, 26(4): 429–46. —— Darby, M., and Armstrong, J. (1998). ‘Geographically Localised Knowledge: Spillovers or Markets?’, Economic Inquiry, 36: 65–86. —— Darby, M., and Brewer, M. (1998). ‘Intellectual Human Capital and the Birth of US Biotechnology Enterprises’, American Economic Review, 88(1): 290–306. Zysman, J. (1994). ‘Rooted Trajectories of Growth’, Industrial and Corporate Change, 3(1): 243–83. —— (1996). ‘Institutions and Economic Development in the Advanced Countries’, in G. Dosi and F. Malerba (eds.), Organization and Strategy in the Evolution of the Enterprise. London: Macmillan.

313


Index

Appendices, figures, notes, and tables are indexed in small, bold letters, e.g., 218n. Abbate, J. 53 Abernathy, W. J. 3, 60n academic excellence 213 academic research 90, 195–6, 206, 207–10, 256 see also research commercialization 211 academic systems (Medicon Valley) 140–1 accounting (Medicon Valley) 136 acquisitions 92, 213 Israel 176 Acs, Z. J. 218n actors 275 administration (Republic of Ireland) 154 aerospace sector 42 agglomerations 3, 9, 17, 29, 113, 156, 169, 195, 202–5, 206, 213–14, 216, 224, 237, 266, 270 see also Marshallian agglomerations activities 5–6, 178, 185, 222, 252, 264, 272 benefits 221, 222, 222f, 224 economies 4, 36, 206–7, 219–20, 221, 223, 231, 233, 236, 247, 248 forces 9, 28, 196, 198, 204, 216, 217, 219–20 industrial 20–1 mechanisms 201, 202

regional 266, 269, 272 Aharonson, B. S. 206, 212–13 Ahuja, G. 200 Allansdottir, A. 133, 212, 213 Allen, J. T. 21, 26 Almeida, P. 197, 218n alternative mechanisms 195 Altshuler, R. 170n Amdahl 49, 50 American Mutoscope and Biograph Company 18, 21 Americas 161 see also USA Amgen 105 Amsden, A. 149, 150, 169–70 ˚ . 135, 136, 137 Andersson, A Andersson, M. 135, 136, 137 Angel, D. 50 Anger, K. 34 Apple 49, 50, 148, 156 Arora, A. 214, 215 Arthur, W. B. 36, 205, 240n, 245, 246, 248 Asia 12, 156, 157, 161, 173 AstraZeneca 141 AT&T 254, 275 Audia, P. G. 87 Audretsch, D. B. 196, 201–2, 206, 211, 212, 218n automobile industry 214, 266, 274

315

Index Avnimelech, G. 58, 174, 176, 185, 187, 189, 190n, 191n Baccaro, L. 152 Bahrami, H. 40 Balio, T. 21, 29, 35 Baptista, R. 248, 265 Barry, F. 157, 164, 171n Bas, T. G. 211, 212 Battat, J. 167 Baum, J. A. C. 206, 212–13 Bay areas see San Francisco Bay Area Bayh-Dole Act 1980 (USA) 210, 211 Beaudry, C. 203 Becattini, G. 207 Becker, R. 188 Beijing (China): biotechnology clusters 118–27, 270 ownership types 124f(a) Bell Labs 6, 42, 168, 267, 271, 273 Ben-David, J. 208, 209 Bercovitz, J. 251, 261 binational structure (Medicon Valley) 140, 145 Benassi, M. 200 Bergek, A. 277n, 278n bioengineering 90, 117 biomedical technology 38, 88 biotechnology clusters 6, 7–8, 9, 56, 61–82, 88–90, 110, 205–6, 210, 213, 214, 255–7, 260, 269, 273 see also Boston/Cambridge; Canada; high-technology clusters; human biotherapeutics; industry clusters; San Francisco Bay Area China 113–31, 131t, 132, 260, 270, 271, 272 failures 126; geographic distribution 120f; laboratories 123; policies 113–15t, 116–18; research organizations 119f

316

entrepreneurship 207, 259, 268 Israel 272 Medicon Valley (Europe) 8, 133–47, 270, 271 start-ups 142f organizations 64–5 US Capitol 274 USA 61, 211, 212, 267–8, 271, 274 biotherapeutics see human biotherapeutics birth rates 223, 231 ¨ rkbacka, R. 141, 144, 145 Bjo Black, B. 188 Blakely, E. J. 236 blind-booking 32 block-booking 32 Boisson, J.-P. 213, 215 Bonaccorsi, A. 133, 212, 213 Bordwell, D. 18, 26 Borgatti, S. P. 204 Boschma, R. A. 258 Boston/Cambridge (USA) 6, 7, 8, 40, 57, 61, 74, 102, 257, 267 see also dedicated biotech firms; public research organizations; San Francisco Bay Area; USA biotechnology clusters 63–81, 211, 255–6, 273, 275 dyads 66f(a), 67 human biotherapeutics 99, 105, 106, 108, 110 networks 69f, 70, 81 semiconductors 108–9 Bottazzi, G. 205, 218n Bowie, J. L. 65 Bowser, E. 21, 22, 23 Bradley, J. 171n Bramwell, A. 254 brand management (motion-picture industry) 25 Brantley, P. 62 Brazil 117 Brennan, W. 198

Index Breschi, S. 197, 201, 202, 203 Bresnahan, T. F. 200, 202, 204, 266 Brewer, M. 218n Brezis, E. S. 220, 226, 227 Brock, W. A. 205 Brownlow, K. 21 Bruno, G. 209 budgetary revenues (China) 121–2 Bunker-Whittington, Kjersten 82n bureaucracy: China 124, 126 Republic of Ireland 167, 170 Burt, R. S. 200 business 39, 46, 274 academic collaboration 90 climate 146 consultancies 1 environment 168 models 5–6, 36, 39, 47, 53, 54, 58, 59–60 plans 55 productivity software 55 sector (Israel) 176, 177–8, 183, 188, 189 strategies 28 Bygrave, William D. 51 call centers (Republic of Ireland) 148 Campbell, Joseph 2 Canada 258 biotechnology clusters 212, 250, 253, 260 Ottawa telecom cluster 254–5 photonics cluster 255 Caniels, M. C. J. 203 capital 41, 108, 221 China 124 Europe 212 gains 49, 56, 57, 58, 59 taxes 259 investments 58 fixed 34

Israel 179, 182t, 185 markets 173, 174, 176, 184 social 262 Carnegie Foundation 97 carrying capacities 10, 12, 231, 232, 233, 237 Carlsson, B. 277n, 278n Casper, S. 202 Cassady, R. 21 Catherine, D. 213, 215 causation 203–4, 215–16, 219, 237, 247, 256 Celtic Tiger economy (Republic of Ireland) 150–3 Central Europe 156–7, 167 Cetus Corporation 88–9 Chamberlin, T. 255 chance events see serendipity China 7, 11, 12, 156, 173, 260 see also under Beijing; bureaucracy; entrepreneurship; government; high-technology clusters; infrastructures; location theory; markets; networks; property rights; regulations; research; research and development; science; Shanghai; Shenzhen (Guangdong); start-ups; technology; universities; venture capital Chinese Academies of Science 117, 125 Chiron 256 Christensen, C. M. 51 Christopherson, S. 32 Chu, W. 149, 150, 169–70 cinematic entertainment see motionpicture industry Cisco 52–3, 250 citations 73, 74, 77–9, 118t biotechnology 141, 141t(b) China 116, 117 Medicon Valley (Europe) 136, 141t(a)

317

Index Clark, B. R. 208, 209 Clarke, C. G. 22, 23, 27 clusters 260–1, 270 see also path dependence development 87, 216–17, 228f, 257–8, 261, 263, 268, 272, 275 demographics 90–1; geographic convergence 95–9 formation 11, 244, 249–55, 258, 261, 264–5, 267–8, 272, 275–6 growth 257, 259, 262, 264 clusterization 202–5 Coleman, J. S. 200 collaborations 1, 198, 199, 215, 257, 263 collective learning (Israel) 174, 184, 190 colocalization 199 colocation 221–2, 248, 252 commercialization strategy 65, 67, 68, 70, 71–2, 76, 79, 116, 122–3, 256, 267 academic research 136 biotechnology 208 communication 154, 197 equipment 44 infrastructures 232 communications (Israel) 174 competencies 5, 203, 204, 215, 249, 265, 266, 268, 270, 272, 276 Medicon Valley (Europe) 136, 139–40 competitive dynamics 249–50 competitiveness 21, 79, 221, 223, 233, 236, 243, 248 advantages 3–4, 18, 22, 28, 195, 249, 273 Canada 253 China 113, 128 dynamics 29 industrial 2, 4 Israel 174, 177 Medicon Valley (Europe) 136, 137, 140, 146 motion-picture industry 36–7

318

product-market 164 Republic of Ireland 151, 152 USA 259 competitors 74–5, 249 components suppliers 51, 157, 166 computers 1, 38, 42, 44, 48–50, 56 see also hardware; software employment (EU) 155t(b) networking 51–3, 58 Republic of Ireland 166 assembly 156–7; employment 148–9, 149f; exports 154 science 253 concentrations see spatial concentrations congestion 221, 223, 233 consultancies 249, 254 Continental Europe 150, 161, 206, 208, 209 Cooke, P. 198, 202 coordination 34 multidimensional process (Israel) 189–90 Corolleur, C. 213, 215 Coronini, R. 213, 215 corporations 213 research laboratories 39, 59, 252–3 taxes 156, 167; Israel 184–5; Republic of Ireland 153, 158, 161, 163–4, 168 costs 38–9 industrial clusters 220–23 Cowan, R. 200, 205 Cowan, W. 205 creation myths 2, 5–6 critical mass see agglomeration, activities cultures 39, 258 entrepreneurial 271 Israel 188 Medicon Valley (Europe) 134, 144

Index regional 108, 246 Republic of Ireland 160 Silicon Valley (USA) 56, 58 supportive 262–3 cumulativeness 203, 226–7 Dalle, J. M. 205 Darby, M. 206, 218n data communications 52 David, P. A. 36, 205, 245 Davies, S. 164 Davis, K. 6, 39, 47, 267, 275 de Geus, A. J. 47 de la Mothe, J. 255 debt (Republic of Ireland) 152 decentralization 207 China 121–2 decision-making 114 Israel 185 dedicated biotech firms (DBFs) 89 Boston/Cambridge (USA) 62, 65, 67, 68, 71–7 San Francisco Bay Area (USA) 62, 65, 67, 68, 71–3, 75–7 Dedrick, J. 156, 157 deep clusters 169 Dell 148, 156, 157, 168–9 demonstration effects (Republic of Ireland) 164–5 Dendrinos, D. S. 242n Denmark 8, 12, 133–5, 138–46, 161, 252, 260, 270, 271, 274 dense networks 200 see also networks Desai, M. A. 153 design software 48 Detroit (USA) 272, 274 development agencies (Republic of Ireland) 167, 169–70 development clusters 6–9 development dynamic 21 development policies see policies diagnostic equipment 61, 64, 123

diagnostics (China) 119 digitization 40 direct entries: human biotherapeutics 96, 106 diseases 89–90 diversification 202, 215 China 129 scientific research 206 DNA molecules 88 DNA process 210–1 Doner, R. 51, 60n Dorfman, N. 266 Dosi, G. 67, 203, 205, 218n dot-com bubble 42, 44, 45 drugs 64, 90 see also medicines biotechnology 77–8, 78t, 79, 91, 140, 144 orphan 76 Dublin (Republic of Ireland) 8, 168 see also hardware (computers), employment; Republic of Ireland Durkan, J. 157–8 dyads 65, 68 Boston/Cambridge (USA) 66f(a), 68 San Francisco Bay Area (USA) 66f(b), 67–8 dynamics 11–12 growth 36 East Europe 156–7, 167 Echeverri-Carroll, E. L. 198 economic development 13, 38, 236, 237, 243, 247, 263 economic environment 232, 232f, 233 economic exploitation 211 economic growth 1, 3, 11–12, 13, 110–11, 215, 216, 236, 252, 253–4 phase (Israel) 174, 182 Route 128 (USA) 201

319

Index economic policy 9–11, 190, 196–7 initiatives 10 instruments 10–11 interventions 11, 233 Israel 8, 172–3, 175–6, 183–5, 187–9 Medicon Valley (Europe) 134, 137, 140, 145–6 role of 195 USA 175 economics 245, 247, 248, 259 economies 206–7, 227, 250, 251–2, 274 Israel 175 Republic of Ireland 150, 152–3 economies of scale 166–7, 248, 270–1 ecosystem see innovation system Edison Manufacturing Company 18, 21 education 257 academic 209, 252–3 Canada 260 Continental Europe 208–9 higher 258 Medicon Valley (Europe) 145 Republic of Ireland 150, 153, 163, 167 science-based 154, 169 technical (Canada) 253 USA 258 worforce 258 efficiency gains 200, 232 Eisenhardt, K. M. 40 electrical engineering 253 Electronic Data Systems (EDS) 162 electronics 38, 41, 42, 169, 266, 267, 271 China 116 Israel 172 Republic of Ireland 148, 154, 156, 157, 162, 165, 166, 167, 169, 170 Ellison, G. 197 embedded software (Republic of Ireland) 162

320

EMEA market 156–7, 161 emigration (Republic of Ireland) 150 empirical regularities 265–72 employees see also labor Medicon Valley (Europe) 142, 142t, 143–44 employment 42 Canada 253 Republic of Ireland 168 foreign firms 148, 161, 164 Israel 177 USA 255 endowments 9, 196, 197, 201, 202, 204, 232, 246 institutional 249 Engstrom, J. 50 enterprises: foreign-owned 120–1 entertainment software 55 entrepreneurship 2, 3, 4, 6, 7, 9, 11–12, 38, 41, 48, 52, 58, 87, 90, 113, 139, 205–6, 207, 212, 214–7, 250, 251–4, 257, 258, 261, 262, 264, 265, 267, 271, 272, 274, 277 China 114–16, 126, 128, 269, 271, 274 human biotherapeutics 91, 92, 98–9, 102, 105, 106, 109–11 relocations 100t Israel 187 local 108, 270 Medicon Valley (Europe) 144 motion-picture industry 29 Republic of Ireland 165 Silicon Valley (USA) 39–42, 53, 54, 56–7 support infrastructure 39, 56, 58, 59–60 technological 188 USA 255, 267–8, 274 Washington/Baltimore (USA) 269

Index entry factors 252, 276 equity 90 see also stock options private companies (Israel) 179, 182t, 188 Ethernet 52 Europe 8, 10, 11, 12, 18, 35, 64, 117, 129, 146, 150, 154, 156, 159, 160, 161, 166, 173, 206, 207, 208, 209, 211, 212, 213, 237 see also Medicon Valley; Republic of Ireland European Union (EU) 8, 12, 138, 144, 148, 150, 151, 153–4, 158–9, 169, 267 Evans, S. 40 evolutionary policy 243–5, 257–8, 272 exit markets 92, 224 Israel 188 exports 161, 164 China 130 Israel 177 Republic of Ireland 161, 165, 166 services 157; specializations 168 externalities 196–7, 199, 202–3, 205, 214, 216, 231, 247, 248, 275 fabrication: costs 46 semiconductors 58 Fagiolo, G. 205, 218n Fairchild Semiconductor 40–2, 46–7, 271 Faust, K. 82n feedback 250, 276, 277 Feldman, M. P. 3, 196, 201–2, 204, 206, 212–3, 218n, 251, 255, 261, 268, 273 Ferguson, C. H. 53 Fernett, G. 31, 34 Ferreira, L. 153

Fiedler, M. 191n Fields, G. 157 film industry see motion-picture industry finance 11, 65, 68, 261, 271, 275 capital 276 Israel 173 bank 183; institutions 188; joint 184 Republic of Ireland 161 Finland 161 firms 11 characteristics 203 Medicon Valley (Europe) 142t, 143–44 fiscal policy 8, 232 strategy (Republic of Ireland) 150, 151 FitzGerald, J. 152–3 flagship projects (Republic of Ireland) 164, 167, 168 Fleming, L. 74 Florey, R. 23, 27 Florida, R. 40, 218n Foley, C. F. 153 Foray, D. 205 foreign direct investment (FDI) (Republic of Ireland) 148–71, 267 foreign expertise (Israel) 8 Forseń, Sture 136 Fosfuri, A. 214 Foster, P. C. 204 France 28, 133, 146, 161, 208–9 Francis, J. L. 251, 261, 268 Frank, L. 134, 167 free trade (Republic of Ireland) 150 Freiberger, P. 49 Fruchterman, T. 68 Fujita, M. 217n, 218n functional patterns 276–7 functionality 38–9 funding, phase-released (China) 126

321

Index Galambos, L. 62 Gambardella, A. 62, 133, 200, 202, 204, 212, 213, 214, 215, 266 Gambarotto, F. 240n, 241n game boxes 55 game theory 215 Gant, J. 203 Gargiulo, M. 200 Garud, R. 38 Genentech 89, 90, 256 geography 63, 88, 195, 211, 221, 227, 237, 248, 263, 268, 270 benefits 221, 224 clusters 9, 197, 220–1, 258 demographics 92–4 innovation 9, 196–7 locations 63, 110, 154, 189, 196, 207, 214, 217 Germany 133, 161, 188–9, 209 Gertler, M. S. 249, 258, 262 Gilson, R. 188 Gittelman, M. 208 Giuliani, E. 198, 200 Glaeser, E. L. 197 globalization 37, 137, 156, 249 GNP (gross national product): Republic of Ireland) 151f tax-to 152 Goe, W. Richard 40 Gompers, P. A. 39, 87, 109 Gomery, J. 35 ¨ rg, H. 164, 169, 171n Go government: agencies 64–5, 268 China: business 121–3; clusters 269; role of 126, 128–9, 270 clusters 257 Israel 189–90 policies 176, 183; venture contributions 179, 181–2 incentives 156

322

laboratories 88 Canada 254, 258 policies 257–8, 261, 274 USA 271–2 governance: policy 258 private 62 USA 259 grants 97–8 Israel 183, 183t horizontal 176 regional 156 Great Britain see UK Great Depression 35 Greece 156 Griffith, D. W. 26, 27, 29 Grimes, S. 162, 163 Gropp, R. 170n growth see economic growth Grubert, H. 170n guided missiles 42, 44 Guston, D. 260 Hagedoorn, J. 62 Haggard, S. 51, 60n Hall, P. 202 Hampton, B. B. 19, 26 Hancock, M. G. 40 hard disk drives (HDDs) 51 hardware (computers) 44 country exports 155t(a) employment (Dublin) 148, 148f, 149, 168 Republic of Ireland 150, 154–7, 165, 169, 170 Hargadon, A. B. 56 harmonization 144 Harrison, B. 203 Hart, J. A. 59, 108 health (China) 119 Hellmann, T. 188, 191n Henderson, J. V. 196, 217n, 231, 240n Henderson, R. 218n

Index Hewlett-Packard 41, 50, 148, 162, 168 high-risk investment 60 high-technology clusters 6, 7, 9, 11, 39, 42, 57, 59, 198, 237, 252, 259 see also Silicon Valley Canada 253–4 China 114, 116, 127–8, 270 Israel 172–5t, 176–90 Medicon Valley (Europe) 139, 252 Silicon Valley (USA) 266–7 Republic of Ireland 148, 153, 160, 163, 168 Hiltzik, M. A. 259 Hines, J. R. 153 Hirschman, A. O. 247 Hochtberger, K. 162, 163 Hollywood (USA) see motion-picture industry Hong Kong 127, 130, 270 Horizontal Grants to Business Sector (Israel) 177–8, 187 horizontal policies 183 hospitals 72, 80–1, 206, 212 Medicon Valley (Europe) 134, 138, 139 Hsu, D. 60n Huettig, M. 32 human biotherapeutics 7 1976–2003 (USA) 87–111, 214, 257 firms 96t human capital 157, 251, 275 human disease 89, 91 human diagnostics 90 Human Genome Project (China) 117, 125 human resources 108 Israel 177 human therapeutics 61, 64, 90 IBM 41, 44, 48, 49, 50, 55, 148, 158, 162–3, 168, 253 ICE technology 38–9 illnesses 79–80

immigration 177 Inbal funds (Israel) 177, 184 incentives 4–5, 12, 156, 216 design 12 economic 58, 274 individual 11 micro-level 10, 237 risk-sharing (Israel) 179 state and private business 128 structures 9 transparent 273–4 income, real national (Republic of Ireland) 151 increasing returns 245, 246, 247–8, 249, 250, 252 incubator regions 40, 116, 165, 214 Israel 176, 179, 187 India 56, 117, 173, 188 industrialization (Republic of Ireland) 149–50 industries 267–9, 275–6 associations 165, 161 representatives (Medicon Valley) 138 specializations 197, 201–2, 206 China 129; Republic of Ireland 149 structure (Israel) 183 supporting services 249–50, 274 industry clusters 94–5, 106–11, 220–4, 236, 237, 243, 248, 251, 264–5, 269, 272, 273, 275, 277 see also biotechnology clusters; high-technology clusters; life cycle growth 223–4, 227, 230f(a), 230f(b), 230–3, 234f(a), 234f(b), 235–6, 238–9ap, 247, 264–5, 277 new 226, 226f, 274 old 225, 225f information 38, 65, 80–1, 109, 139, 197, 232, 254 local sources 198 market failures 215

323

Index information and communication technology (ICT) 8, 211, 257, 260 Israel 172, 174–5, 185, 186t, 260 Republic of Ireland 148, 153, 165–70, 260, 267, 271 USA 212 infrastructures 156, 206, 221, 224, 232, 237, 247, 250, 258, 272 China 111, 122 Israel 173 Medicon Valley (Europe) 136 Republic of Ireland 151, 154, 163–4 innovations 3, 6, 27, 40, 52, 59, 61, 63, 73, 74–5, 79–81, 195, 198, 201, 211, 212, 215, 217, 219, 224, 247, 261 clusters 195–6, 198, 199–201, 203, 205–12, 220, 236, 237 content 28–9 Israel 9, 176–82, 189 motion-picture industry 35, 36 processes 196, 203 production 236 Republic of Ireland 170 systems 2, 3, 38, 244, 258–9, 263, 277 institutional learning (Republic of Ireland) 150, 151, 152, 163–4, 170 institutions 4, 29, 39, 82, 111, 113, 199, 204, 206, 213, 215, 244, 246, 251–2, 260, 275 architecture 248–9 China 114 reforms 127–8, 269, 274 design 11, 12 Europe 8, 208 framework 146, 262–3 Hollywood (USA) 34 Israel 174, 177, 179, 183, 188, 274 Medicon Valley (Europe) 134 reforms 7, 8 Republic of Ireland 274

324

San Francisco Bay Area (USA) 59 Silicon Valley (USA) 57 supporting 40, 195, 204 integrated circuits (ICs) 47–8, 53 Intel 39, 49, 50, 148, 157–8, 168–9 intellectual property (IP) 39, 65, 74, 78–9, 206 rights (IPR) 211–2 USA 259–60 international standards (China) 129 Internet see World Wide Web inventions 211 investments 64, 74, 252, 271 government (Israel) 179, 189–90 local 108, 109 Republic of Ireland 167–8 Silicon Valley (USA) 54 subsidizing (China) 126 Ireland see Republic of Ireland Isard, W. 240n Israel 8, 11, 12, 57, 158, 172–90, 260, 267, 271 see also business, sector; capital; collective learning; economic policy; equity; finance; government; grants; hightechnology clusters; incubator regions; information and communication technology; innovations; institutions; investments; markets; networks; partnerships; products; research and development; science; Silicon Valley; software (computers); start-ups; system failures; technology; venture capital; Yozma Program IT services: India 188 Republic of Ireland 158, 167 Italy 207 Izod, J. 21, 27

Index Jacobs, Jane 111 Jacobs, L. 18, 19 Jacobson, D. 170n Jacobsson, S. 277n, 278n Jaffe, A. B. 83n, 218n Japan 47 Jarvie, I. 28 Jessen, C. 23 Jin, H. 122 joint ventures: biotechnology 121 human biotherapeutics 90, 92, 95 Jonard, N. 200 Jones, C. 24, 25, 29

spillovers 197, 201, 221, 223, 231, 236, 247, 252, 265, 269, 275 USA 259–60 Kogut, P. 197, 216, 218n Kolko, J. 169 Koput, K. 63, 82n, 215 Korea 118 Kortum, S. 211 Kostial, K. 170n Koszarski, R. 32 Kraemer, K. 156, 157 Krugman, P. 196, 218n, 220, 226, 227, 248 Kuperman, J. 29

Kaldor, N. J. 247, 248 Kallal, H. D. 197 Kamada, T. 68 Kanemoto, Y. 240n Karnoe, P. 38 Kawai, S. 68 Kearns, A. 169 Kelley, M. R. 203 Kenney, M. 39, 40, 52, 53, 56, 58, 60n, 62, 88, 89, 90, 174, 190n, 201, 204, 210, 211, 266 Ketels, C. 249 Kettler, H. 202 Klepper, S. 60n, 87, 202–3, 214, 266 Knight, F. H. 146n knowledge 7, 8, 11, 12, 53, 62, 75, 79, 105, 111, 114, 161, 195–6, 199, 201, 202, 203, 204, 206–7, 212, 214, 216, 227, 248, 251, 257, 258–9, 261, 263, 272, 273, 276 biotechnology 256, 275 China 129 diffusion 199 Israel 273 local 198–200, 215, 217, 254 Medicon Valley (Europe) 136, 137, 138, 139–40, 143–5, 270

labor 35, 41, 215, 221, 247, 248, 253, 262 China 119, 127, 129, 130 costs 144, 156 division of 209, 270–1, 275, 276 markets 164, 197, 258, 274 Medicon Valley (Europe) 139, 142, 142t, 143, 143t, 145 mobility 199, 252 motion-picture industry (USA) 32–6 scientific 63, 252 Republic of Ireland 150, 154, 163, 164, 168 Silicon Valley (USA) 57 skills 117, 251, 260, 268 laboratories 251, 268 national 206, 209 research 267 USA 258, 259–60 Lai, H.-C. 17, 118 Lambooy, J. G. 258 large multinational companies (MNCs) 121 latency phase 251 lateral thinking 237 Latin America 118 Lau, Lawrence J. 128 law (China) 126

325

Index Lazonick, X. 114 Leachman, C. H. 46 Leachman, R. C. 46 leadership 10, 62, 108, 111, 227, 251 Israel 183 learning 247 see also institutional learning economies 227–9 effects 226, 227, 228f interactive 198 Israel 180 Lee, C. M. 40 legal consultancies 1 legalities (Medicon Valley) 136 Lemarie`, S. 213, 215 Lenway, S. A. 59, 108 Lerner, J. 39, 87, 109, 190n, 211 Leslie, S. 42 Lewis, Michael 57 liberalization (Republic of Ireland) 153 licensing 52, 65, 68, 70, 75, 90, 208, 210, 254 life cycle: industrial clusters 223–36, 237 modeling 229–31 product theory 3, 46 life formats (USA) 210–1 life sciences 80, 88 Medicon Valley (Europe) 134, 137, 143–44 life demographics 92 linkages 167, 169, 175, 198, 200, 206, 256, 276 Lindblom, C. 275 Lindqvist, G. 249 Link, A. 269 Lissoni, F. 197 Liu, X. 114 living standards 167 local antecedents 258, 260 local development 258 local networks 256 see also networks; regional networks

326

area (LANs) 51 China 270 local economy 2, 13, 262–3 localization 160–1, 198 location theory 1, 4, 7, 20–1, 23, 33, 154, 157, 169, 197, 219, 221–2, 237, 244, 250, 252, 258, 261, 266, 270 see also colocation; relocation benefits 220–4, 227, 228, 229 Canada 254–5 China 114–6, 122, 128–9, 269–70 human biotherapeutics 98, 108, 109–10 Medicon Valley (Europe) 134 Republic of Ireland 163–4, 166, 168 lock-ins 36, 81, 243, 246, 251, 262, 272–3 motion-picture industry (USA) 5, 18–19, 22, 35, 36 negative 200 Loewendahl, H. 171n Los Angeles (USA) 257 see also motion-picture industry human biotherapeutics 102–3, 104t, 105, 106, 110 Lotka, A. J. 242n Lu, X. 114 Lucent Technologies (Dublin) 168 Lyons, D. 164, 198 Macau 130 Macdonald, L. 254, 255 MacNeil, I. R. 63 macroeconomics 168, 232, 233 MacSharry, R. 153, 166, 171n Maggioni, M. A. 221, 223, 232, 239n, 240n, 241n, 242n Magnet Program (Israel) 176, 177–8, 187 magnetic data see peripherals Mahoney, J. 246 Mailes, G. 35

Index Malaysia 156 Malerba, F. 201, 202, 203 Malmberg, A. 248, 251 management compensation (Israel) 185 Mangematin, V. 213, 215 manufacturing: China 130 costs 46 European Union (EU) 159 Republic of Ireland 148, 160, 165 zero taxes 150, 158 Mariani, M. 133, 212, 213 markets 8, 32, 260, 274, 275 China 126–7 mechanisms 122, 128; orientation 113–16, 127 European Union (EU) 159, 159t Israel: failures 183, 184 Medicon Valley (Europe) 145 niches 39, 49 opportunities 204 outcomes 76–7 strategy 138 turbulence 236 marketing/management skills (Israel) 184 Marshall, A. 248 Marshallian agglomerations 187, 213–14, 247 see also agglomerations Republic of Ireland 149–50, 164, 169 Marsili, O. 203 Martin, R. 273 Maskell, P. 198, 248, 251 Mayer, D. 53 Maynard Smith, J. 242n Mazzoleni, R. 210 McKelvey, M. D. 62 McKendrick, D. G. 51, 60n, 157 medical devices (Republic of Ireland) 148

medical technology (Medicon Valley) 140 medicines 62, 75, 76, 79–82 see also drugs Medicon Valley (Europe) 8, 254, 260, 274 see also biotechnology clusters; competencies; competitiveness; economic policy; employees; hospitals; knowledge; labor; networks; pharmaceutical corporations; research; service providers; start-ups; universities mergers 213 human biotherapeutics 92, 95, 96 Israel 176 Merges, R. P. 210, 211 Merton, D. 210 metropolitan statistical areas (MSAs): human biotherapeutics 91–7t, 98–9, 108 relocations 100t, 101t Mexico 19, 156 Mezias, J. M. 29, 31 Mezias, S. J. 31 microcomputers 44, 49, 50, 156 microelectronics 209, 211 microprocessors 48, 50, 157–8, 271 Microsoft 55, 60, 148, 160 Midelfart, K.-H. 167 migration 10, 34, 214, 231, 257 Milgram, S. 200 Miller, W. F. 40 minicomputers 48–9, 51, 52 Minying Keji Enterprises (MKEs) 114–5 Miyao, T. 240n molecular biology 62, 63, 209 Mooney, J. 161, 165 Moore, G. 6, 39, 47, 267, 275 mortality rates 223, 231 motion-picture industry 6, 21, 267 see also labor; lock-ins;

327

Index motion-picture industry (cont.) vertical integration; production companies 31 employment 30, 30t France 28 Hollywood (USA) 5, 17, 18–20, 23–9, 30, 32, 34, 35–7, 265–6, 270–2 Los Angeles (USA) 19, 23–4, 25f, 28, 30, 31f, 33, 34, 266 New York-New Jersey 18, 21, 23–4, 27, 28–9, 30, 35, 36, 271 Mowery, D. C. 208, 211 Mullally, H. 242n Murtha, T. P. 59 Musser, C. 18 Myrdal, G. 247 nanotechnology 56, 59 natural sciences: Medicon Valley (Europe) 135 Republic of Ireland 154 Nelles, J. 254 Nelson, R. 2, 38, 67, 183, 203, 208, 210, 211 Netherlands 161 Netscape 53–4, 57 networks 1, 36, 51, 52, 63, 65, 68, 70, 74–6, 80–1, 198–200, 204, 215, 217, 246, 251, 252, 261, 275 see also computers, networking China 122, 126 interorganizational 6, 62, 79 Israel 178, 180–1, 187 Medicon Valley (Europe) 138, 139, 139f New New Thing, The (Lewis) 57 New York (USA) 257, 267 human biotherapeutics 105, 106, 107t, 109, 110 Newlon, T. S. 170n Nielson, M. 35 Nijkamp, P. 242n Nilsson, M. 134

328

Niosi, J. 211, 212 Nordic states 133, 270 Nortel 250 North, C. J. 29 North, D. N. 245 North America 18, 156 Northern Electric (Canada) 254–5, 260 O’Dell, P. 27n O’Gorman, C. 161, 165 OhUallachain, B. 46 O’Malley, E. 161, 165 operations: clusters 157 hubs 160 Organization for Economic Cooperation and Development (OECD) 152, 154, 158, 159, 188 organizations 39, 87, 88, 215, 251, 262 automobiles 214 classification 93–4 human biotherapeutics 90, 95–7, 102, 110 innovation 207, 270 Medicon Valley (Europe) 137 structures 201, 204 O’ Riain, S. 161, 165 Orsenigo, J. 62, 133, 199, 201, 202, 203, 206, 209, 212, 213 Ottaviano, G. I. P. 218n Ottawa telecom cluster (Canada) 254–5, 260 Overman, H. G. 167 Owen-Smith, J. 63, 65, 67, 75, 82n, 204, 206 ownership (companies): China 130app influence of 164 Republic of Ireland 152, 154, 156, 160, 162, 164–5

Index Palludan, U. 134 Palmer, E. O. 19 Pammolli, F. 62, 199, 206, 212, 213 Papageorgiou, Y. Y. 217n, 240n partnerships 81 joint (Republic of Ireland) 168 limited (Israel) 174, 176, 179, 184–5 foreign 178, 180–1, 181t, 185 patents 19, 22, 64, 72, 73–9, 89, 116, 136, 140, 208, 210–1, 254 path dependence 3–4, 10, 11, 36, 169, 261, 264, 265–7, 268, 272 cluster formation 243–63 Israel 188 pay increases (Republic of Ireland) 150 PC assembly firms 156–8 PCB boards 157 Pearl, R. 242n pension funds 175, 259 peripherals 50, 52 see also computers; personal computers magnetic storage 50–1, 58 Republic of Ireland (exports) 154 Perkins, A. 54 personal computers 49, 50, 51, 59, 201 see also computers; IBM software 55 Persson, H. 134 pharmaceutical corporations 62–3, 64–5, 75, 76–7, 89, 90, 92, 109, 117, 268 China 124, 126 Republic of Ireland 148, 161 Medicon Valley (Europe) 135, 139, 140 Sweden 144 Philippines 157 photonics (Canada) 255 Pitts, M. R. 31 policies 206, 213, 220, 231–1f, 232–5, 237, 244, 249–50, 275 China 260, 269

instruments 261 interventions 259, 262–3 Israel 260 process goals 277 regional 258 Republic of Ireland 260, 267 role of 243–3 simulations 233–5t, 236 targeted 272–3 USA 260, 267–8 policy-induced clusters see China, biotechnology clusters politics 275–6 EU 154 Ponti, J. 22 population ecology theory 220, 237 Porter, K. 67, 70 Porter, M. 47, 171m, 249, 250, 261, 262 Portugal 156 positive feedback 245, 252 postgraduate studies (USA) 208 Powell, W. W. 62, 63, 65, 67, 70, 75, 82n, 198, 204, 206, 215 Prevezer, M. 206, 211, 213, 215, 218, 237, 241n, 252 price mechanism (China) 128 privatization (Israel) 181 production 32, 195–6, 219–20, 226, 246, 247, 251 capacity 30, 30t, 46 motion-picture industry (USA) 17–18, 20, 22, 23, 26, 28, 32, 35, 36 Republic of Ireland 149, 150, 152 units 20–1 products: 198, 245 development 79, 80, 89, 91, 226 Israel 177 development 184; global 173, 176 productivity spillovers 164–5 products 198, 249 profitability 220, 224–5, 232, 268, 274 property rights 7, 215

329

Index property rights (cont.) China 114, 121, 127, 269 USA 210 proto-venture capitalism 41–2 public health 81 public policies 10, 255, 257, 258, 262, 272–7 see also policies emerging clusters 264–7 interventions 272 public relations 34 Medicon Valley (Europe) 136 public research organizations (PROs) 6, 64–5, 67, 68, 71, 72, 74–5, 79, 81–2, 199 Boston/Cambridge (USA) 70, 80, 256, 273, 275 research 195, 212 public sector: agencies 258 funding 258 interventions 247 research organizations 72, 80 Puerto Rica 157 Puga, D. 218n Qi, C. Y. 117 Qian, Y. 122 Quah, D. 169 Raleigh-Durham (USA): human biotherapeutics 105–6 Rallet, A. 199 Ramsaye, T. 19 Redding, H. J. 167 Reed, L. J. 242n Reggiani, A. 242n regional conditions 87 China 129 regional clusters 246, 261 regional development 247, 248 regional economy 93, 244, 251, 258, 262–3

330

regional innovation policies 72–3, 236–7, 257 regional networks 67, 70, 74, 77, 108, 232, 256 see also networks regionalization 137, 145 regulations: China 123, 124, 126–8 Denmark 144 Reingold, E. 68 relational contracting 63 relational databases 55 relocation 88, 227 see also collocation; location theory human biotherapeutics 94, 95, 96, 98–9, 101t, 102, 105, 109–10 Republic of Ireland 166–7 Republic of Ireland 8, 11, 12, 148–71, 260, 267, 272 see also bureaucracy; Dublin; competitiveness; computers; corporation, taxes; economics; education; electronics; employment; exports; flagship projects; foreign direct investment; hardware, computers; high-technology clusters; information and communication technology; infrastructures; labor; location theory; manufacturing; Marshallian agglomerations; ownership (of companies); production; skills; venture capital; wages research 1, 72, 89, 197, 200, 206, 209, 250, 251, 254, 257, 261, 267 Canada 260 centers 9, 216 China 123–7, 129, 270 centres 125–6; institutes 128; organizations 118; projects 124t(b)

Index excellence 206 infrastructure 251, 253, 258, 262, 263 Medicon Valley (Europe) 134, 139, 141 clinical 140; funding 133; parks 136, 138 institutions 81, 92, 105, 114, 206, 258 publicly funded 210 USA 208 research and development (R&D) 12, 198, 211, 216 Canada 254 China 115, 116, 118, 121t, 123, 140, 141, 146, 158, 164, 168 Israel 173, 175–8, 183, 184, 187, 188, 189, 267 USA 63, 65, 67, 68, 71, 72, 73, 73t, 74, 77–80, 258 Research Triangle Park (NC, USA) 268–9, 272 resources 106, 108, 274, 275, 276 returning scientists (China) 117, 117t, 122, 126–8, 271 Riccaboni, M. 62, 199, 206, 212, 213 Richardson, H. W. 240n Robbins-Roth, C. 75, 82n, 88, 89 Roberts, E. 266, 267, 269, 271 Robinson, D. 34 Roijakkers, N. 62 Romanelli, E. 87 Romijn, H. A. 203 Rosenberg, N. 208 Ross, M. 34, 35 Route 128 (USA) 200–1, 266, 267, 271, 274 Rowen, H. S. 40 royalty payments 75 Ruane, F. 171n rule of law 121 Russia 121, 128 salaries (China) 117–18 sales revenue (China) 124

Salvatore, R. 203 Sampat, B. 208, 211 San Diego (USA) 7, 257 biotechnology 123 high-technology 102 human biotherapeutics 103t, 105, 106, 108, 110 San Francisco Bay Area (USA) 6, 8, 19, 56, 61, 70, 102, 257, 267 see also biotechnology clusters; dedicated biotech firms; dyads biotechnology 63, 64–6f(b), 67–1, 211, 255–7, 273, 275 corporate research institutes 39–40 electronics 42 employment 44f establishments 45f human biotherapeutics 99, 105, 106, 108, 110 innovations 74 networks 69f Saxenian, A. 57, 87, 108, 109, 200–1, 202, 204, 207, 266 Scandinavia 135 Scharfstein, D. 39, 87, 109 Scheinkman, J. A. 197 Schoonhoven, C. B. 40 Schumpeter, Joseph 105 science 62, 77, 80, 90, 205–6, 215, 216 analysis 197 China: industry parks 115–16; institutions 118; policies 113–15, 115t; returnees 116–17, 117t clusters 257 industries 144 Israel 177 research 208–11, 214 Republic of Ireland 170 scientific instruments 44

331

Index Scotland: employment (computers) 154, 156, 159 Scott, A. J. 20, 32, 35, 248 second-generation growth 7, 261 human biotherapeutics 105, 106, 108–10 sector specifics 272 Medicon Valley 145 sector specialization 120t, 248, 249 seed firms 233, 250, 252, 262 seed funding 123, 138–9 self-organization 11 semiconductors 39, 40–2, 45–8, 50–1, 58, 108–9, 158, 266, 275 design software 54–5 sequential processes 3 sequential strategy 188 serendipity 7, 10, 12, 40, 41, 243–6, 249, 250–1, 257, 261, 262, 264–5, 277 service providers (Medicon Valley) 134, 138–9 Shan, W. 216 Shanghai (China): biotechnology clusters 118–27, 270 ownership firms 124t Sharpe, M. 63 Shen, X. 167 Shenzhen (Guangdong) (China): biotechnology clusters 118–27, 270 ownership firms 124t Shindler, C. 34 Shleifer, A. 197 Shockley Semiconductor 40–1 Shockley, William 40, 41, 271 short term survival see survival strategy Shyu, J. Z. 115, 116 silicon semiconductor 40 Silicon Valley (USA) 1, 3, 6, 7, 11, 38–60, 178, 185, 200–1, 252, 254, 259, 266, 271, 272, 273, 274, 275

332

see also under entrepreneurship; Israel; start-ups clones 236 dedicated biotech firms (DBFs) 78t, 79 semiconductors 108–9, 266–7 Simoni, M. 152 Singapore 156 Single European Market 153, 163, 271 skills 156, 169, 197, 198, 204, 250, 251, 261 China 117, 129 Israel 184 labor force 214, 252 Medicon Valley (Europe) 146 Republic of Ireland 152, 153, 154, 162–3, 166, 168 workers 224, 262 Sklar, R. 19 Slide, A. 23, 27 small businesses (USA) 260, 261 small worlds 200 see also networks Smith, T. 217n Smith-Doerr, L. 65, 215 social dynamics 4 social interaction 108 social network analysis 199 social partnership model 8 Republic of Ireland 151, 152 social processes 106, 108 social relations 3, 199 social welfare spending (Republic of Ireland) 152 soft institutions 214 software (computers) 42, 44–5, 47–8, 50, 54–6 custom 161 employment (Dublin) 148, 149, 149t, 168 European Union (EU) 159, 159t India 188 Israel 174, 177, 186t

Index packed 159–60t, 161 Republic of Ireland 150, 154–6, 158–3, 166, 170 research and development (R&D) 162–3 Solomon, J. 47 ¨ . 249 ¨ lvell, O So Sorenson, O. 74, 81, 87 Soskice, D. 202 South Korea 117 space 195 Spain 156 spatial collocations (Republic of Ireland) 163–5, 168 spatial concentrations 9, 202, 203, 216, 219–20 geography 227 spatial dimensions 6, 7, 196 research 195 spatial interactions 220 spatial localizations 205 spatial patterns 30 spatiotemporal patterns 204–5 specialization see industry, specialization Spencer, R. V. 19 Spender, J. C. 38 spillovers 214, 216, 217, 248, 267 spin-offs 199, 202–3, 213, 214–16, 219, 223, 252, 256, 257, 266, 267, 268, 269 academic 208, 211 Canada 253–4 China 114, 128 human biotherapeutics 92, 95, 96, 106 Republic of Ireland 165 spontaneous development 277 Staiger, J. 18, 26 stars 25–6 start-ups 57, 58, 102, 113, 122, 206, 216, 232, 251, 252, 253, 256, 260, 268, 273, 274

see also venture capital, start-ups China 62, 114–15, 124, 127, 269 economy 38 human biotherapeutics 90, 92, 105, 109 Israel 174–5, 176, 179, 183t, 184–5, 188 high-tech 183 Medicon Valley (Europe) 136 biotech firms 142f, 143 networking 53 science 106 Silicon Valley (USA) 39, 41–2, 46, 47, 50 USA 175 Stephan, P. 218n Stern, S. 27 Stinchcombe, A. L. 87 stock markets (Israel) 184 stock options 57, 259 Storper, M. 20, 32, 108, 237, 248, 252 Stout, D. 218n, 237, 241n strategic alliances 61, 63–4, 91 strategic planning 261, 262 Strobl, E. 164, 169 Strogatz, S. 200 structural holes 200 Stuart, T. 81 Sturchio, J. 62 Sturgeon, T. 41, 42 subsidiaries 92, 269 subsidies 12 Medicon Valley (Europe) 146 Sun 49 superconductors 56, 59 supplier industry 48 supply chains 169 survival strategy: entrepreneurship 12 short-term 2 Svensson-Henning, M. 134 Swaine, W. 49

333

Index Swann, P. 206, 213, 215, 218n, 237, 241n, 252, 265 Sweden 8, 11, 12, 133–5f, 136, 138, 140–6, 159, 161, 252, 260, 270, 271, 274 system failures 272 Israel 183, 184, 185 Taiwan 46, 118, 169–70 talent 11 Tauchen, A. 240n taxes: exemptions (China) 126 incentives (Republic of Ireland) 166, 271 increases (Republic of Ireland) 152 issues (Sweden) 146 reductions 8, 152 zero, on profits 150 technical development 82 technical formats 160 technical training 81, 261 Canada 253 technology 2, 3, 38, 40, 56, 57–8, 74, 201, 205, 216, 236, 246, 248–9, 251–3, 261, 265 China 123, 126 government-owned 114; industry parks 115–16; institutions 118; policies 113–15, 115t choice 245 clusters 157, 200–2, 226–9, 243, 245 dynamics 220, 237 innovation 9–10, 207, 219, 227–8, 237 Israel 173, 177, 179–80, 189 intensive cluster 8, 172; policy 173, 176 opportunities 204, 251, 267, 272, 273, 274 Republic of Ireland 166, 170 Silicon Valley (USA) 43f, 58

334

research 214 regimes 201, 202 spillovers 169 transfers 81, 197, 209, 212 turbulence 236 Teece, D. J. 63, 210 telecommunications 250, 255, 261 entrepreneurship 259 Ottawa telecom cluster (Canada) 254, 260 Republic of Ireland 161, 165, 167, 170 temporal dimensions 7 Terman, F. 41 Teubal, M. 58, 174, 176, 177, 185, 187, 189, 190n, 191n Texas Instruments 40 Thailand 156 therapeutics 76, 78 see also human therapeutics therapies 77 third generation growth 261 Thisse, J.-F. 218n Thompson, K. 18, 26, 28 Timmons, Jeffry A. 51 Todtling, F. 40 Torch program (China) 116 Torre, A. 199, 213, 215 Torrence, B. T. 19 Trajtenberg, M. 83n, 218n transaction: costs 215, 221 market 36 order 34 transistor industry 40, 273 transport 221, 232 Republic of Ireland 166–7 trickle-down effects 247 Trommettier, M. 213, 215 trust 262 UK 133, 146, 152, 156, 159, 161, 164, 208 unconditional objects 205

Index unemployment: mass 10 Republic of Ireland 150, 151 unionization 35 universities 9, 39–40, 58, 65, 72, 80–1, 92, 97, 105, 109, 114, 123, 198, 199, 206, 208, 214, 216, 232, 252, 256, 261, 266, 269 Canada 253–4 China 114, 124, 125t, 128 entrepreneurial 208, 211 Europe 208–9 interaction with business 41, 62, 197 laboratories 62, 88 Medicon Valley (Europe) 134, 135, 137–1, 145 research 195, 210 university–industry interactions 207–11 urbanization economies 216 Canada 212 US Capitol (USA) 260, 267–9, 271, 272, 273, 274 US Securities and Exchange Commission 91 USA 7, 12, 18, 21, 22, 28, 29, 34, 36, 53, 61, 64, 75, 117, 126, 129, 136, 141, 145, 153, 154, 156, 158, 173, 175, 178, 207–9, 211, 212, 213, 250, 254, 255, 258, 259–60 see also under biotechnology; human biotherapeutics; entrepreneurship; motion-picture industry; research; research and development; venture capital user–producer relationships 199 Utterback, J. M. 3, 60n Uzzi, B. 200 value chain 157, 160 Van Egeraat, C. 157, 170n

Vanhoudt, P. 153 Varmus, H. 80 Venables, A. I. 218n Venables, A. J. 108, 167 venture capital (VC) 1, 6, 7, 8–9, 41, 42, 48, 52, 56, 57, 59–60, 65, 70, 71, 76, 77, 79, 80, 87, 108, 205–6, 207, 214, 252, 259, 261, 273 Canada 212 China 62, 63, 64–5, 67, 68, 114–6, 126, 127 human biotherapeutics 90, 92, 98, 109 Israel 172–80, 180t(b), 181–3t, 184, 189–90, 271, 274 policy failure 187–9 Medicon Valley (Europe) 133, 134, 136, 138–9 R&D strategy 53 Republic of Ireland 163–4, 167 Research Triangle Park (NC, USA) 269 San Francisco Bay Area (USA) 68, 72, 80, 256, 257, 273 Silicon Valley (USA) 53, 268 start-ups 49, 51 USA 173, 175, 211, 212 Washington/Baltimore (USA) 268 World Wide Web (WWW) 54 Verhulst, P. F. 242n vertical integration: motion-picture industry (USA) 32, 34, 35 disintegration 32 Route 128 (USA) 201 Von Burg, U. 39, 51, 52, 60, 201, 204 von Hippel, E. 47 wages: China 130 determination 8 Republic of Ireland 151–2, 164

335

Index Walker, G. 216 Walker, R. 47, 237 Washington/Baltimore (USA) 257 Wasserman, S. 82n Watts, D. 200 Weber, A. 240n Weinberg, R. A. 80 Weingast, B. R. 122 welfare effects 11 West Europe 150, 167 White, D. R. 63, 82n White, M. 162, 163 White, P. 153, 166, 171n White, S. 114 Whittington, B. 67, 70 Wichman, C. 135, 136, 137 Wilkenson, O. 134 willingness-to-take- a-chance 58–9 Winter, S. 2, 38, 203 Witte, H. D. 240

336

workstations 48, 49 Wolfe, D. A. 254, 258, 262 World Wide Web (WWW) 53–4, 57 Xerox 41, 49, 52, 253, 259 Xiao, L. 131 Yeats, A. J. 169 Young, A. A. 247 Young, R. N. 49 Yozma Program (Israel) 176, 178–80t(a), 181, 181t, 182, 184–5, 187, 188, 189, 190, 267, 271, 275 Ziedonis, A. 208, 211 Zierer, C. M. 19 Zook, M. A. 54 Zucker, L. 197, 206, 218n Zysman, J. 251, 258

Cluster Genesis: Technology-Based Industrial Development

Cluster Genesis: Technology-Based Industrial Development

Cluster Genesis: Technology-Based Industrial Development