Complexity and Industrial Clusters

Contributions to Economics
http://www.springer.delcgi-bin/seurch_book.pl ?series= 1262

Michael Carlberg Gustav A. HornlWolfgang Scheremetl

Intertemporal Macroeconomics Rudolf Zwiener
1998. ISBN 3-7908-1096-7 Wages and the Euro
1999. ISBN 3-7908-1199-8
Sabine Spangenberg
The Institutionalised Trans- Dirk Willer
formation of the East German The Development of Equity Capital
Economy Markets in Transition Economies
1998. ISBN 3-7908-1103-3 1999. ISBN 3-7908-1198-X

Hagen Bobzin Karl Matthias Weber

Indivisibilities Innovation DifTusion and Political
1998. ISBN 3-7908-1123-8 Control of Energy Technologies
1999. ISBN 3-7908-1205-6
Helmut Wagner (Ed.)
Current Issues in Monetary Heike Link et al.
Economics The Costs of Road Infrastructure
1998. ISBN 3-7908-1127-0 and Congestion in Europe
1999. ISBN 3-7908-1201-3
Peter Michaelis/Frank SWhlcr (Eds.)
Recent Policy Issues in Environ- Simon Duindam
mental and Resource Economics Military Conscription
1998. ISBN 3-7901l-1137-8 1999. ISBN 3-7901l-1203-X
Jessica de Wolff Bruno Jcitziner
The Political Economy Political Economy of the
of Fiscal Decisions Swiss National Bank
1998. ISBN 3-7908-1130-0 1999. ISBN 3-7908-1209-9
Georg Bol/Gholamreza Nakhacizadeh/
Karl-Hcinz Vollmcr (Eds.) Irene Ring et al. (Eds.)
Risk Measurements, Econometrics Regional Sustainability
and Neural Networks 1999. ISBN 3-7908-1233-1
1998. ISBN 3-7908-1152-1
Katharina Muller/Andreas Rylll
Joachim Winter Hans-llirgen Wagener (Eds.)
Investment and Exit Decisions Transformation of Social Security:
at the Plant Level Pensions in Central-Eastern Europe
1998. ISBN 3-7908-1154-8 1999. ISBN 3-7908-1210-2

Bernd Meyer Stefan Traub

Intertemporal Asset Pricing Framing EfTects in Taxation
1999. ISBN 3-7908-1159-9 1999. ISBN 3-7908-1240-4

Uwe Walz Pahlo Coto-Milhin

Dynamics of Regional Integration Utility and Production
1999. ISBN 3-7908-1185-8 . 1999. ISBN 3-7908-1153-X

Michael Carlberg Frank Riedel

European Monetary Union Imperfect Information
1999. ISBN 3-7908-1191-2 and Investor Heterogeneity
in the Bond Market
Giovanni Galizzi/ 2000. ISBN 3-7908-1247-1
Luciano Venturini (Eds.)
Vertical Relationships and Kirsten Ralf
Coordination in the Food System Business Cycles
1999. ISBN 3-7908-1192-0 2000. ISBN 3-7908-1245-5
continued on page 308
Alberto Quadrio Curzio . Marco Fortis
(Editors)

COll1plexity
and Industrial Clusters
Dynamics and Models
in Theory and Practice

With 23 Figures and 8 Tables

Physica-Verlag
A Springer-Verlag Company
Series Editors
Werner A. Muller
Martina Bihn

Editors
Professor Alberto Quadrio Curzio
U ni versita Cattolica
CRANEC
Via Necchi, 5
1-20123 Milano
[email protected]

Professor Marco Fortis

Montedison
Piazzetta M. Bossi, 3
1-20121 Milano
[email protected]

ISBN 978-3-7908-1471-2 ISBN 978-3-642-50007-7 (eBook)

DOl 10.1007/978-3-642-50007-7

Cataloging-in-Publicatjon Data appli\!d for

Die Deutsche Bibliothek - CIP-Einheitsaufnahme
Complexity and industrial clusters: dynamics and models in theory and practice: 8 tables /
cd.: Alberto Quadrio Curzio; Marco Fortis. - Heidelberg; New York: Physica-Verl.. 2002
(Contributions to economics)

This work is subject to copyright. All rights arc reserved. whether the whole or part of the
material is concerned. specifically the rights of translation. reprinting, reuse of illustrations.
recitation, broadcasting, reproductiun on microfilms or in any other way, and storage in data
banks. Duplication of this publication or parts thereof is permitted only under the provisions
of the German Copyright Law of September 9. 1965. in its current version. and permission
for use must always be obtained from Physica-Verlag. Violations are liable for prosecution
under the German Copyright Law.
Physica-Verlag Heidelberg New York
a member of BertelsmannSpringer Science+Business Media GmbH
© Physiea-Verlag Heidelberg 2002

The use of general descriptive names. registered names, trademarks, etc. in this publication does
not imply, even in the absence of a specilic statement, that such names are exempt from the
relevant protective laws and regulations and therefore free It,r general use.
Softcover Design: Erich Kirchner, Heidelberg
SPIN 10865729 88/2202-5 .. 3 2 I 0 - Printed on acid-free and non-aging paper
 Preface

This volume contains the proceedings of the international conference "Complexity

and Industrial Clusters: Dynamics and Models in Theory and Practice", organized
by Fondazione Comunita e Innovazione and held in Milan on June 19 and 20,
2001 under the aegis of the Accademia Nazionale dei Lincei (founded in Rome in
1604), one of the oldest and most famous national academies of science in the
world.
Fondazione Comunita e Innovazione encourages research and the dissemination
of knowledge about social, economic, cultural and civil issues. It promotes
research and innovation related to local production systems and industrial districts,
with special reference to: the interactions between large companies and SMEs
(small and medium-size enterprises), the effects of industrial districts on the
development and welfare of their communities and of neighbouring areas, the
effects of globalisation on these local systems of productions.
Fondazione Comunita e Innovazione was created in Milan in 1999. It supports
studies, publications, and events, both on its own and in cooperation with
corporations, research institutes, foundations, associations and universities. It also
grants scientific sponsorship to research that is in line with its mission, as set forth
in its by-laws. The founding member of the Fondazione is Edison (formerly
Monted:son). The other subscribing members, in historical order, are: Ausimont,
Tecnimont, Eridania, Accenture, and PricewaterhouseCoopers.
The editors of this volume, personally and on behalf of Fondazione Comunita e
Innovazione, wish to extend their heartfelt thanks to all those who have
contributed to the success of the conference and particularly to Enrico Bondi, past
Chairman of the Fondazione, Edoardo Vesentini, President of Lincei, which
granted the scientific sponsorship, to all of the speakers, whose papers are
presented following the order of the conference programme, and to David A. Lane
and Franco Malerba, chairmen of the sessions who contributed greatly to the
debate and the deeping of the issues.

Milan, December 2001 Alberto Quadrio Curzio

Marco Fortis
 Contents

Preface ....................................................................................... V

Introduction: From Specific Industrial Cases to a General Economic

Model?
Alberto Quadrio Curzio and Marco Fortis .............................................. 1

I. Complexity and Economic Dynamics

What Is Complexity?
Murray Gell-Mann ......................................................................... 13

Complex Adaptive Systems and Spontaneous Emergence

John H. Holland ............................................................................ 25

\The Industrial Cluster as a Complex Adaptive System

I' Enzo Rllllani ................................................................................ 35

II. General Models of Industrial Cluster Dynamics

Complexity and Local Interactions: Towards a Theory of Industrial

Districts
David A. Lane ............................................................................... 65

From Marshall's to the Italian "Industrial Districts". A Brief Critical

Reconstruction
Giacomo Becattini ......................................................................... 83

Diversity, Not Specialization: The Ties That Bind the (New) Industrial
District
Charles F. Sabel .......................................................................... 107

Competitive and Synergic Behaviours in the Development of Industrial

Clusters: Ecological Modelling and Empirical Evidence
Marco Fortis and Mario A. Maggioni ................................................. 123

On the Ubiquitous Nature of Agglomeration Economie~ and Their

Diverse Determinants: Some Notes
Giulio Bottazzi, Giovanni Dosi and Giorgio Fagiolo ............................... 167
V!II Contents

III. Success Cases Around the World

The Entrepreneurial Spark: Individual Agents and the Formation of

Innovative Clusters
Maryann P. Feldman and Johanna Francis .......................................... 195

From Exogenous to Endogenous Local Development: The Cases of

the Toulouse and Sophia Antipolis Technopoles
Christian Longhi .......................................................................... 213

Biotechnology Development in Germany: The Case of Nordrhein-WestfaIen

Francesco Salamini, Anke Sohn and Hartmut Thomas .............................. 239

Support for Technology-Based Firms: The Role Played by Property-Based

Science Parks
Don Siegel, Paul Westhead and Mike Wright ........................................ 249

High-Tech Industrial Clusters in the Asian Countries: The Case of the

Electronics Industry in the Hsinchu-Taipei Region
Kling Wang ........ ........................................................................ 267

Financial Markets, Industrial Clusters and Small and Medium-Size

Enterprises
Angelo Tantazzi ........................................................................... 291

List of Authors ............................................................................ 301

Introduction: From Specific Industrial Cases
to a General Economic Model?

Alberto Quadrio Curzio and Marco Fortis l

1. Foreword

It seems fitting to begin this publication, which contains the proceedings of the
international conference described in the preface, with a brief mention of two
institutions: Fondazione Comunita e Innovazione, which organized the
conference, and Accademia Nazionale dei Lincei, which gave the scientific
sponsorship to it.
Both of these institutions are important for their different traditions in the fields
of scientific and technological research.
Fondazione Comunita e Innovazione is closely associated with the historical
roots of Edison, founded in 1884, of Montecatini, founded in 1888, and, later on,
of Montedison. This is true primarily, but not exclusively, because these two
companies were able to build on their "local" civil and economic base and then
expand nationally and internationally by constantly focusing on technological
innovation. It is a well-known fact that the discovery of polypropylene, for which
Giulio Natta received the Nobel Prize in 1963, was made possible by the support
of Montecatini in an outstanding example of collaboration between industry and
research. Many other individuals throughout Montecatini's and Edison's histories
deserve mention for their ability to combine entrepreneurial spirit and scientific
and technological innovation, most notably Guido Donegani, Giacomo Fauser and
Giuseppe Colombo.
Accademia Nazionale dei Lincei, which honoured our conference with its
confidence and scientific sponsorship, was founded in 1603, making it the oldest
Academy in the world. It is worth mentioning that its founding members included
Galileo Galilei, who in 1613 published his Historia e dimostrazioni intorno alle
macchie solari as part of the Academy's proceedings. This is not the place for a
lengthy and detailed presentation of the extraordinary scientific merits of the
Accademia dei Lincei and of its contributions to Italian and international scientific
history. Interestingly, Giulio Natta, Nobel Prize, was a member of the Academy in

lOur heartfelt thanks to Mario A. Maggioni for his help in planning the conference and for
his useful remarks about this essay. Many others who have contributed to the conference
should be thanked, especially Biancamaria Frondoni for the organization of the conference
and for the general editing of this volume. We also thank David A. Lane and Nicola W.
Palmieri for the connections with the Santa Fe Institute, Monica Carminati, Cristiana
Crenna and Franca Sapienza for their help in editing this book.
2 A. Quadrio Curzio and M. Fortis

1947, as were Giuseppe Colombo, member since 1888 and Giacomo Fauser,
member since 1948.
It seems to us, then, that the holding of this international conference and the
publication of this book provide an ideal historical perspective linking these two
institutions, which share a similar vision, even though their objectives may be
different. And this vision is that science and technology must cooperate, that
industry cannot prosper without research. It is for this reason that the next two
paragraphs of this introduction present the points of view of two highly respected
personalities in their own field: Emico Bondi, at that time the Chairman of
Fondazione Comunita e Innovazione and c.E.O. of Montedison, and Edoardo
Vesentini, Chairman of the Accademia Nazionale dei Lincei. With each one of
these personalities the editors of this book had numerous exchange of ideas while
preparing this conference. And therefore the editors will refer their points of view,
obviously taking some liberties in interpreting their thoughts, which are shared by
the editors of this volume.

2. General Models, Economic Analysis and Applications

In the preface we have already said that the conference and this volume were
made possible by the initiative of Fondazione Comunita e Innovazione, which for
years has been promoting studies and research on the evolution of Italian
industrial districts. Therefore, we will use this section of the introduction to
present Enrico Bondi's point of view. The idea to connect complexity and
industrial districts belongs to him. But here we will explain the facts about Italian
districts that have much interested Enrico Bondi convincing him that this
phenomenon was worthwhile of deep analysis also from the point of view of a big
company.
While it may seem peculiar to begin the introduction to this book, which
presents the contribution of eminent theoreticians (some of them Nobel laureates),
with data on the Italian industrial system, we chose this approach to help the
reader understand how a specific situation can be used to develop a general
scientific approach.
In Italy, there are more than 200 main industrial districts, which account for
over two million jobs, or about 40% of Italy's total manufacturing employment
and 60% of all jobs in companies that make the most familiar made-in-Italy
products (fashion, home furniture and furnishings, food and traditional mechanical
engineering products). Made in Italy products are recognized worldwide for their
top quality, and industrial districts are the leading manufacturers of these quality
products.
As a result, these districts, which almost always enjoy global leadership in their
respective industries (from textiles and high fashion to ceramics and machine
tools) account for more than one-third of all Italian exports. The largest districts,
such as Sassuolo for ceramics or Prato for textiles, generate revenues in excess of
five billion euros. But there are dozens of districts with revenues of between 500
 Introduction: From Specific Industrial Cases to a General Economic Model? 3

million euros and one billion euros and dozens more with revenues of 100 to 500
million euros.
There are also extremely interesting examples of cooperation between industrial
districts, which contain small and medium-size enterprises, and large companies.
The most striking example in this area is the cooperation between the "Fluorine
Valley" district and Montedison. This district is located in the province of
Bergamo along the border with the province of Brescia, with its local production
system that manufactures specialized products using elastomers and fluorinated
polymers. Another and more famous case of the collaborative relationship is that
between the Maranello mechanical engineering district, which is host to such a
high-technology marvel but small firm as Ferrari, and the big automobile firm that
is Fiat.
Therefore, a study of local manufacturing systems and industrial districts, and of
the interaction between small and medium-size enterprises, must also take into
account the role of large companies and the relationship that these local systems
have with universities and centres of research in order to achieve technological
innovation, since these factors are essential in helping enterprises grow and
compete internationally.
Enrico Bondi emphasized this theme in his interesting remarks at the beginning
of the conference. Agreeing especially with the views of two Italian economists
that he often met, Giorgio Fun. and Giacomo Becattini, and through the discussion
with the editors of this volume, he emphasised that the Italian district model is a
spontaneous one based mainly on the self-organizing ability of local communities,
which in many cases is rooted in the artisan traditions of the free city-states of the
Italian Renaissance. The Italian model is that of a district consisting of small and
medium-size enterprises located away from large metropolitan areas, primarily in
the heart of some of Italy's northern and central provinces. It is a development
model that evolved without the support of public industrial policies.
In order to understand the question contained in the title of this section of the
introduction, it is important to remember that new industrial districts, or clusters,
have developed in other countries, and have become important engines of
economic growth. Just think of Silicon Valley or Hsinchu-Taipei. These clusters
are agglomerations of non-Fordist enterprises strongly rooted in local society and
supported by a network of social and economic relations with strong links with a
major university or research centre. Two are thus the relevant (and intertwined)
questions. The first: can we consider these innovative clusters a sort of "second
generation" industrial districts? The second: will the Italian industrial districts be
able to make the transition from traditional products to high-technology ones?
In conclusion, the system of Italian industrial districts appears to have produced
a "model of capitalism" that is different from the Anglo-Saxon and Rhine Valley
ones. However, this model (which we can define as "horizontal or network
capitalism") seems not to be limited to Italy, for new district systems, often
involving highly sophisticated technologies, that are emerging throughout the
world seem to point to the fact that this new type of capitalism does not possess
limited national connotations.
4 A. Quadrio Curzio and M. Fortis

3. Can General Theoretical Models Be Used to

Understand Specific Economic Cases?

A reference to the other institution that lent its prestige to this conference is also
important in placing this book within the context of the methodological input that
Edoardo Vesentini, Chairman of the Accademia N azionale dei Lincei, provided to
Alberto Quadrio Curzio and which has been presented by the latter in the
introduction to the conference.
These introductory remarks, after noting that the Accademia Nazionale dei
Lincei was glad to support the conference because it deals with economic
development and technological innovation and, therefore, has to do with aspects
of scientific research in which Lincei are interested, will review the three main
topics of the conference and of this book, paying special attention to certain
methodological issues.
The first issue has to do with the following question: can general models be
used to understand specific economic cases? This question was addressed during
the first session of the conference and is dealt with in the first part of this book.
The issue is whether or not general theories of complexity can be used to develop
models that can, in turn, be used to analyse economic dynamics, particularly those
that affect the various types of industrial districts. Vesentini, while not addressing
the specific issue of the link between complexity and industrial districts, pointed
out that, frequently, general scientific theories and mathematical models have
found important applications in the economic field. Therefore, further studies in
this area are always important for the scientific progress and valuable because
economic theory has been a major beneficiary of mathematical methods. The most
striking example in this area is that of John von Neumann. However, it is
necessary to proceed with some caution in order to avoid misapplication of the
theory in general and the economic application in particular.
The second question, which was addressed during the second session of the
conference and is discussed in the second part of this book (which is more closely
related to economics because it presents several models of district dynamks), is
the following: are these alternative or complementary models? Vesentini pointed
out that here the analysis focuses on the relationship between the theoretical
approach and the empirical approach. This interaction has been extremely
important for the success of economics, giving also rise to a separate branch, that
of quantitative methods and econometrics, that was recognized with the first
Nobel Prize ever awarded for economics. In this area, Vesentini offered a
suggestion: do not try to force a complex economic reality into a single model,
since complementary interpretations are often one of the remarkable advantages of
economic research.
The third issue, which was discussed during the third session of the conference
and is reviewed in the third part of this book (which analyses different types of
industrial districts within different economic-institutional and industry contexts),
has to do with the development of national paradigms of district dynamics.
Vesentini pointed out that in this area the analysis has to deal with the relationship
 Introduction: From Specific Industrial Cases to a General Economic ModeP 5

between economic research, research on institutions and social research, which is

the third and equally important methodological component of economics. In other
words, he emphasized that the conference, and this book, must use the three
different methodological approaches that, in the field of economics, have
produced indisputable gains.
Having provided these general introductory remarks, we shall now review the
indi vidual parts of this 'Volume.

4. Complexity and Economic Dynamics

The first part of this book, which contains three papers by Murray Gel\-Mann
(What Is Complexity?), John H. Holland (Complex Adaptive Systems and
Spontaneous Emergence) and Enzo Rullani (The Industrial Cluster as a Complex
Adaptive System), deals with the relationship between complexity and economic
dynamics and industrial districts in an effort to develop a general model.
The study of complexity is based on the notion that there is a general law that
governs the formation of orderly configurations within systems in dynamic
balance. Complex systems are formed by independent agents that interact in a
non-linear fashion, adapting, evolving and developing a form of self-organization
that enables them to acquire collective properties that are not attributable to
specific agents.
These complex self-organizing systems are adaptive in the sense that they do
not react passively to events, the way a stone would tumble in an earthquake. On
the contrary, they make a deliberate effort to turn all circumstances to their
advantage.
This theory has been applied in the most diverse areas, ranging from the natural
sciences to such fields as meteorology and urban planning. In economics, the
theory of complexity has been used to study stock market cycles and is now being
considered with great interest by some of those who are studying industrial
districts.
It is also important to remember that industrial districts of different types now
exist in many part of the world. In most cases, they reflect self-organizing models.
Therefore, it was reasonable to believe that the theory of complexity could
provide important general guidelines for understanding the birth and dynamics of
these districts when they are viewed as a general manifestation of self-organizing
systems.
With this in mind, top priority was given to investigating the relationship
between complexity and economic dynamics. This task was entrusted to Murray
Gell-Man and John Holland, key representatives of the Santa Fe Institute, where
mathematicians, computer scientists, physicists, neurobiologists, evolutionary
biologists, economists, environmentalists, political scientists, historians,
archaeologists, linguists and many other scientists work. At the Santa Fe Institute,
Gell-Mann and his colleagues, among them John Holland (the father of "genetic
6 A. Quadrio Curzio and M. Fortis

algorithms" and one of the world's leading computer scientists), take an

interdisciplinary approach in studying the meaning of simplicity and complexity.
The purpose of the first part of the book is to stimulate a broad debate on the
ways in which the analytical methods, developed by physics and biology, can be
transferred to economics, where the phenomenon of Italian industrial districts
appears to offer to complexity scholars an extremely fertile ground for analysis.
The issue that, more or less implicitly, is being addressed here is to what extent
the general theories of complexity and adaptive systems can provide a theoretical
basis to explain the economic dynamics that characterize industrial districts.
The papers produced by these two researchers on complexity are particularly
relevant with regard to this issue. It is not by accident that, even though one is a
physicist and Nobel laureate and the other a computer scientist and the developer
of genetic algorithms, their contribution has proven to be extremely valuable in
analysing the economic phenomena that are the subject of this conference. It
would seem appropriate at this point to specifically mention again John von
Neumann, who, though not an economist, provided a fundamental contribution to
the development of the economic sciences with his general theory of growth and
his theory of games. This convergence underscores the unity of all sciences and
the fact that often we perceive only with a considerable lag the extent of the cross-
fertilization effort - as exemplified by this book - entailed by the courage to
innovate.
The obvious relationship between complexity and districts, as seen by an
economist, is the subject of the paper contributed by Rullani, an expert on Italy's
industnal districts. As such, he was able to abstract universal principles from
specific cases by projecting them onto the canvas of general typologies, without
losing sight of their individual characteristics. Because of this, his paper can also
be seen as an introduction both to the second and third parts of this book. Rullani' s
paper provides an excellent explanation of the effort to combine economic
sciences and complex adaptive systems by drawing attention to the binary
paradigm of exploiting what is known and exploring what is unknown.
Exploitation has to do with refinement, choices and efficiency, while exploration
means searching for something new that cannot be codified beforehand.
Exploration entails complexity, which brings to the realm of economic research a
whole series of variables that previously had been excluded precisely because they
did not fit within the logical guidelines of economic computations based on just
two main variables: price and quantity.

5. General Models of Industrial Cluster Dynamics

The second part of this book contains the following five papers: Complexity and
Local Interactions: Towards a Theory of Industrial Districts, by David A. Lane;
From Marshall's to the Italian "Industrial Districts". A Brief Critical
Reconstruction, by Giacomo Becattini; Diversity, Not Specialization: The Ties
That Bind the (New) Industrial District, by Charles F. Sabel; Competitive and
 Introduction: From Specific Industrial Cases to a General Economic Mode!? 7

Synergic Behaviours in the Development of Industrial Clusters: Ecological

Modelling and Empirical Evidence, by Marco Fortis and Mario A. Maggioni; On
the Ubiquitous Nature of Agglomeration Economies and Their Diverse
Determinants: Some Notes, by Giulio Bottazzi, Giovanni Dosi and Giorgio
Fagiolo.
The second part deals with the general models of district dynamics. It also has to
do, more or less explicitly, with the issue of whether district interpretation models
are complementary or alternative. Lane's introductory paper, which develops a
taxonomy of district operating logics in an attempt to provide all researchers of
industrial districts (and, perhaps, also enterprises) with a common language and
logic, is followed by four modelling studies that offer different analytical-
interpretative systems.
We shall begin by mentioning the paper written by Becattini, who, using
Marshall's approach as a starting point, develops a personal and original
interpretation, viewing districts as the dual expression of social cohesion and
technological innovation and as new systems of network companies that
irreversibly overcome the Fordist paradigm by viewing the division of labour as a
complementary relationship among district companies.
Sabel, who views districts through the lens of an approach where sociological
and institutional approach also matters, provides a more pragmatic-descriptive
approach to explaining how districts operate.
Fortis and Maggioni apply an original theoretical framework, derived from
population ecology, to the analysis of the development path of Italian industrial
districts. The paper underlines the interplay of agglomeration economies and
diseconomies in the growth process of an industrial cluster, distinguishes three
main phases of its development process and stresses the complex and different
(i.e. synergic, competitive, etc.) interactions that exist between different industries
within the same area, and between different areas within the same industry. An
econometric exercise allows the authors to estimate the values of specific
parameters and to forecast the long run evolution of some industrial di8tricts in
Italy.
Dosi et al. in their work attempt to study the multiple drivers of agglomeration
phenomena in contemporary economies and propose a tentative taxonomy where
the conditions of knowledge accumulation, often specific-to-specific locations and
specific sectors, play a paramount role. They discuss the achievements and
limitations of current theorizing on spatial location of economic activities,
proposing a simple model, which is based on Italian data, that highlights the rich
intersectional diversity of agglomeration forces, together with, in a few cases, the
lack of them.

6. Success Cases Around the World

The third part of the book contains the following five papers: The Entrepreneurial
Spark: Individual Agents and the Formation of Innovative Clusters, by Maryann
8 A. Quadrio Curzio and M. Fortis

P. feldman and Johanna Francis; From Exogenous to Endogenous Local

Development: The Cases of the Toulouse and Sophia Antipolis Technopoles, by
Christian Longhi; Biotechnology Development in Germany: The Case of
Nordrhein-Westfalen by Francesco Salamini, Anke Sohn and Hartmut Thomas;
Support for Technology-Based Firms: The Role Played by Property-Based
Science Parks, by Don Siegel, Paul Westhead and Mike Wright; High-Tech
Industrial Clusters in the Asian Countries: The Case of the Electronics Industry in
the Hsinchu-Taipei Region, by Kung Wang.
The theme of the third part of the book is therefore the study of cases that have
an emblematic value at the global level. This section includes five papers that
analyse an equal number of systems that, while not all of the district type, can
provide insights on technological innovation and entrepreneurship that are
territorially based rather than concentrated in large companies. In this case as well,
the implicit questions are: which elements are common to all of these success
stories and which differentiate them, and can the clusters of high-tech companies
that have sprung up all over the world as alternatives to large companies be
classified as types of second-generation districts? This question therefore relates to
the case of Silicon Valley in USA, the case of Toulouse in France, the BioRegio
system in Germany, the British scientific parks and the Taiwan electronics story.
Feldman and Francis outline the development of a regional industrial cluster in
the U.S. Capitol region, through the efforts of entrepreneurs who adapted to both
constructive crises and new opportunities. The paper examines the initial spark of
entrepreneurship and how it influences the formation of high-technology clusters.
The perspective taken here is that entrepreneurs are a critical element in the
formation of clusters, and their actions are important to the analysis of clusters as
complex adaptive systems.
Longhi's paper analyses two French technopoles, Toulouse, in the southwest
region of Midi-Pyrenees, and Sophia Antipolis, in the southeast French Riviera,
both considered the result of the French national system of innovation and process
of decentralisation initiated in the 1970s. The first part of the paper draws on
recent developments in complexity, which emphasize openness, non-linearity and
path dependence, providing an analytical framework that makes it possible to
explain the development of these "technopolitan areas" (defined as areas
developed around several knowledge elements, including but not limited to
science parks). The second part of the paper analyses the cases of Toulouse and
Sophia Antipolis along these lines. The recent development of these complexes
allows us to consider their histories and the "small events" that have resulted in
self-reinforcing expansion, although along highly distinctive paths.
Salamini et al. in their paper review the history and the players that have been
influencing the development of biotechnologies in Nordrhein-Westfalen (NRW):
the BioRegio project, the role of the Bio-Gen-Tec-NRW agency and the first steps
toward biotech development in NRW.
Westhead et al. compare the performance of Science Park and off-Park firms in
the U.K. with regard to several indicators, and explore the role played by the
Science Park manager/director in the development of firms located in "managed"
and "non-managed" Science Parks.
 Introduction: From Specific Industrial Cases to a General Economic Model? 9

Finally, Kung Wang mainly discusses Taiwan innovation system (TIS), that is,
the operation system by enterprises, government, universities and public research
laboratories of science and technology (S&T) industry development, and also the
most representative characteristics that can explain TIS, that is: the technology
development program (TDP) of the Ministry of Economic Affairs, the Industry
Technology Research Institute, the Hsinchu Science-based Industrial Park and
bringing back overseas scholars.

7. Conclusions

The last paper, Financial Markets, Industrial Clusters and Small and Medium-Size
Enterprises, by Angelo Tantazzi, is somewhat separate from the rest of the book,
in that it opens a new area of study and provides guidelines for developing it on
another occasion. In particular, he discusses the links that exist between the
financial markets and small and medium-size enterprises within different
economic and institutional contexts.
In conclusion, we would like to point out that the goals of firm's growth and
profit are always paramount for firms irrespective of the organizational system
adopted by an enterprise. Generally speaking an organizational system does not
exist for its own purpose: it is an operational tool, a method of achieving
economic objectives, two of which are, obviously, growth and profit.
However, districts pursue another objective, which, while not always explicit. is
clearly evident in certain historical and social contexts. It is the goal of
propagating across the generations the district community as a self-organizing
system, even when this requires district participants to sacrifice a proportionate
share of growth and profits. This justifies an interest in districts not only by
economists and sociologists, but also by researchers of complexity, in light of
districts' nature as complex adaptive systems in which enterprises and institutions
interact, accumulating experiences that increase the efficiency of the districts. As
we enter the 21 st century, we can forecast that it wi II be characterized by two great
new phenomena: globalisation and complexity. And perhaps in both cases the new
type of industrial and economic organization that goes under the name of
"indutrial clusters" or "industrial districts" will have an increasing role in its
capacity to combine innovation and flexibility, to combine systems of small-
medium size firms with giant corporations. The reality and the researches of the
next years will tell us which is the answer.
I. Complexity and Economic Dynamics
What Is Complexity?

Murray Gell-Mann

Abstract. It would require many different concepts to capture all our notions of
the meaning of complexity. The concept that comes closest to what we usually
mean is effective complexity (EC). Roughly speaking, the EC of an entity is the
length of a very concise description of its regularities. A novel is considered
complex if it has a great many scenes, subplots, characters, and so forth. An
elaborate hierarchy can contribute to complexity, as in the case of nested industrial
clusters each composed of a great variety of firms and other institutions. In
general, though, what are regularities? We encounter in many different situations
the interplay between the regular and the random or incidental: music and static on
the radio, specifications and tolerances in manufacturing, etc. But ultimately the
distinction between the regular and the incidental depends on a judgment of what
is important, although the judge need not be human or even alive. For instance, in
the case of songs of a male bird in the nesting season, the identification of
regularities is perhaps best left to the other birds of the same species - what
features are essential in repelling other males from the territory or attracting a
suitable female'? A technical definition of EC involves the quantity called
algorithmic information content (AIC). The description of an entity is converted to
a bit string and a standard universal computer is programmed to print out that
string and then halt. The length of the shortest such program (or, in a
generalization, the shortest that executes within a given time) is the AIC. The AIC
is expressed as the sum of two terms, one (the EC) referring to the regularities and
the other to the random features. The regularities of a real entity are best expressed
by embedding it conceptually in a set of comparable things, the rest of which are
imagined. The EC can then be related to the AIC of the set, the choice of which is
restricted by the conditions imposed by the judge. Theorists like to study highly
simplified models of complex systems, often by computer modelling. What can be
claimed for such models? Overall agreement with observation is hardly to be
expected. However, in many cases simple regularities can be found in both the
observational data and the model, which may then be helpful in understanding
those regularities. Examples are given, involving scaling laws and also
"implicational scales".

It would take a great many different concepts to cover all our intuitive notions of
what is meant by complexity (and its opposite, simplicity), but the concept that
agrees best with the meaning in ordinary conversation and in most scientific
discourse is effective complexity (EC). Roughly, the EC of an entity is the length
14 M. Gell-Mann

of a very con..:ise description of its regularities. Thus we would call a novel

complex if it had many different subpluts, scenes, and characters. Likewise, an
international conglomerate firm would be complex if it had many different
branches in different countries with different products, management styles, and so
forth. A hierarchical structure that takes a long time to describe would also
cuntribute to complexity. It is easy to see that an industrial cluster comprising
many different firms and other institutions in diverse locations within a region,
perhaps grouped into sub clusters (which may, in turn, be composed of sub-sub
clusters) is complex in this sense.
An amusing exercise consists of looking at the patterns of neckties to see which
ones are simple (for example those with regimental stripes) and which ones are
complex (for instance most hand-painted ones). Note that we are looking at how
long it takes to describe the regularities, not the features treated as random or
incidental. Of course the description length will typically depend on the "coarse
graining" (level of detail), which depends in turn on how far away the tie is when
viewed with the unaided eye.
We are constantly observing in the world around us the interplay between
regularity and randomness. When we hear music and static on the radio, we
identify the music as regular and the static as random noise, and we try to make
the signal-to-noise ratio as high as we can. In cooking, we may use a recipe but
there are always some variations in the quantities of ingredients, in cooking times
and temperatures, etc. In manufacturing, we encounter specifications and
tolerances.
But the distinction between the regular and the random is not absolute. Many
decades ago scientists at the Bell Telephone Laboratories investigated the origins
of static and found that some of it comes from sources in particular places among
the constellations in the sky. That was the origin of radio astronomy, and hence a
permanent association of some static with important regularities. When we looked
at neckties, we were concerned only with the pattern, and we neglected po~sible
wine or food stains, but to a dry-cleaner the stains might be the significant
regularities, while the patterns are of little concern. We conclude, then, that what
is a regularity depends on a judgment of what is important and what is not. Of
course the judge need not be human, or even alive.
Suppose we are concerned with the regularities in the repeated, but somewhat
variable territorial songs of a male bird of a particular species in the nesting
season. The other birds of the same species would make a suitable judge. Their
behaviour can reveal what features of the song serve to repel other males from the
territory or perhaps to attract a suitable female.
The crude definition of effective complexity given above in terms of minimum
description length needs to be refined and made more technical. First, one should
recognize that the effective complexity is necessarily context-dependent in a
number of ways. To begin with, the entity in question is described at some level of
coarse graining and in some language, and so we have immediately two sources of
context dependence. It is also true that some degree of knowledge and
understanding of the world is assumed. To illustrate the importance of this point,
imagine that you are entering the village of a group of hitherto uncontacted
 What Is Complexity~ IS

Indians in the Brazilian jungle, and that you know their language from experience
with another group speaking the same tongue. Now explain to your hosts the
meaning of a tax-managed mutual fund. It should take quite a while.
Of course as we get more technical we must think in terms of a computer. The
description of the entity is coded into a string of bits (zeroes and ones) according
to some coding scheme, and then a particular standard universal computer is
employed (we have here two more sources of context-dependence). The
algorithmic information content (AIC) of the description is then defined as the
length of the shortest program that will cause the computer to print out the bit
string in question and then stop computing. In fact, we shall generalize that
definition by considering a variable time T and asking for the length of the shortest
program that will perform this task within time T. The AIC is the limit of this
quantity as T goes to infinity.
In discussing K, the AIC of the entity, or the corresponding quantity K(T) for
finite values of T, we are not yet dealing with the effective complexity. The EC
has to do with the length of a very brief description of the regularities of the
entity, not the features treated as random or incidental. Thus we need to exhibit the
AIC (or its generalization to time 7) as the sum of two terms, one applying to the
regularities and the other to the features treated as random.
The best way to represent particular regularities of an entity is to embed that
entity conceptually in a set of comparable things. all the rest of which are
imagined, and to assign probabilities or weights to the members of the set. Such a
set, with probabilities for the members, is called an ensemble, and it embodies the
regularities in question.
The probabilities allow us to define a measure I of "Shannon information" or
ignorance for the ensemble (These two apparent opposites are really two facets of
the same thing: the amount of ignorance about the contents of a letter before one
receives it is the same as the amount of information learned when it is received
and read).
Since the ensemble serves to describe the regularities of our entity, the AIC of
the ensemble (call it Y) is a candidate for the effective complexity. We just have to
find the right ensemble. It turns out that the prescription for that is a
straightforward one:
- First, minimize the "total information" Y+/. The minimum value can easily
be shown to be about equal to K, the AIC of the entity. Thus K is exhibited as
the sum of two terms referring to the regular and random features
respectively. That is just what we needed.
- Second, following a famous principle, maximize the ignorance / or - what is
the same thing - minimize Y, subject to the conditions imposed by the judge.
The imposition of those conditions, based on a judgment of what is
important, is an essential part of the prescription. One can show that
otherwise Y would always be very small and the ignorance / would
correspondingly use up almost all of K. Every entity would be simple.
In the arts, we are accustomed to the use of the imagined alongside the real, as
in fiction and drama, and we are not surprised that light is often thrown in that
way on the regularities of the real.
16 M. Gell-Mann

The break-up of K - the AIC of the entity - into the effective complexity Yand
the ignorance I can be conceived in terms of a basic program and a lot of data to
be fed into that program. The effective complexity is related to the length of the
basic program, while the ignorance measures the information content of the
additional data needed to specify the entity with which we are dealing.
The famous computer scientist, psychologist, and economist Herbert Simon
used to call attention to the path of an ant, which has a high AIC and appears
complex at first sight. But when we realize that the ant is following a rather simple
program, into which are fed the incidental features of the landscape and the
pheromone trails laid down by the other ants for the transport of food, we
understand that the path is fundamentally not very complex. Herb said, "I got a lot
of mileage out of that ant". And now it is helping me to emphasize the difference
between total AIC and effective complexity.
A significant advantage of generalizing the measures K and Y to K(T) and Y(T)
is that we can then discuss situations such as the following:
When I was a graduate student, more than fifty years ago, we wondered what
kind of dynamics underlay the structure of atomic nuclei. It appeared that the laws
that govern the energy levels of nuclei might be extremely complex. Today,
however, we have the theory of quantum chromodynamics, the field theory of
quarks and gluons, supplemented by quantum electrodynamics, which governs the
electromagnetic interaction. We now believe those simple theories are correct (and
many of their predictions have been verified by experiment). They must yield the
energy levels of nuclei to an excellent approximation, but the calculations are so
elaborate that even today's machines and techniques can't handle them properly.
The standard universal computer would take a very long time to execute the short
program corresponding to these simple theories and thus calculate the energy
levels. If we insist on execution in a much shorter time T, then the program would
have to be exceedingly long. So, the energy levels of nuclei have very little
effective complexity Y in the limit of T going to infinity, but huge values of Y if T
is taken to be small.
There can be no finite procedure for finding all the possible regularities of an
arbitrary entity. We may ask, then, what kinds of things engage in identifying and
using particular sets of regularities. The answer is complex adaptive systems,
including all living organisms on Earth.
A complex adaptive system receives a stream of data about itself and its
surroundings. In that stream, it identifies certain regularities and compresses them
into a concise "schema", one of many possible schemata related by mutation or
substitution. In the presence of further data from the stream, the schema can
supply descriptions of certain aspects of the real world, predictions of events that
are to happen in the real world, and prescriptions for behaviour of the complex
adaptive system in the real world. In all these cases, there are real world
consequences: the descriptions can turn out to be more accurate or less accurate,
the predictions can turn out to be more reliable or less reliable, and the
prescriptions for behaviour can turn out to lead to favourable or unfavourable
outcomes. All these consequences then feed back to exert "selection pressures" on
the competition among various schemata, so that there is a strong tendency for
 What Is Complexity? 17

more successful schemata to survive and for less successful ones to disappear or at
least to be demoted in some sense.
Take the human scientific enterprise as an example. The schemata are theories.
A theory in science compresses into a brief law (say a set of equations) the
regularities in a vast, even indefinitely large body of data. Maxwell's equations,
for instance, yield the electric and magnetic fields in any region of the universe if
the special circumstances there - electric charges and currents and boundary
conditions - are specified (we see how the schema plus additional information
from the data stream leads to a description or prediction).
In biological evolution, the schemata are genotypes. The genotype, together
with all the additional information supplied by the process of development - for
higher animals, from the sperm and egg to the adult organism - determines the
character, the "phenotype" of the individual adult. Survival to adulthood of that
individual, sexual selection, and success or failure in producing surviving progeny
all exert selection pressures on the competition of genotypes, since they affect the
transmission to future generations of genotypes resembling that of the individual
in question.
In the case of societal evolution, the schemata consist of laws, customs, myths,
traditions, and so forth. The pieces of such a schema are often called "memes", a
term introduced by Richard Dawkins by analogy with genes in the case of
biological evolution.
For a business firm, strategies and practices form the schemata. In the presence
of day-to-day events, a schema affects the success of the firm, as measured by
return to the stockholders in the form of dividends and share prices. The results
feed back to affect whether the schema is retained or a different one substituted
(often under a new CEO).
A complex adaptive system (CAS) may be an integral part of another CAS and
it may, in turn, contain smaller complex adaptive systems. In fact, a CAS has a
tendency to give rise to others.
On Earth, all complex adaptive systems seem to have some connection with life.
To begin with, there was the set of prebiotic chemical reactions that gave rise to
the earliest life. Then the process of biological evolution, as we have indicated, is
an example of a CAS. Likewise each living organism, insofar as it can learn, is a
CAS. In a mammal, such as a human being, the immune system is a complex
adaptive system too. Its operation is something like that of biological evolution,
but on a much faster time scale (if it took hundreds of thousands of years for us to
develop antibodies to invading microbes, we would be in serious trouble). The
process of learning and thinking in a human individual is also a complex adaptive
system. In fact, the term "schema" is taken from psychology, where it refers to a
pattern used by the mind to grasp an aspect of reality. Aggregations of human
beings can also be complex adaptive systems, as we have seen: societies, business
firms, the scientific enterprise, and so forth.
Nowadays, we have computer-based complex adaptive systems, such as "neural
nets" and "genetic algorithms". While they may sometimes involve new,
dedicated hardware, they are usually implemented on conventional hardware with
special software. Their only connection with life is that they were developed by
18 M. Gell-Mann

human beings, such as that outstanding human John Holland. Once they are put
into operation, they can, for example, invent new strJ.tegies for winning at games
that no person has ever discovered.
It is probably helpful to mention at this point that John Holland and I use
somewhat different terminology. In fact, we both heard at a meeting some years
ago that a scientist would rather use someone else's toothbrush than another
scientist's nomenclature. What I call a schema is called an internal model by John
Holland, and what I call a complex adaptive system is close to what he calls an
adaptive agent. He uses the expression "complex adaptive system" to mean a
loose aggregation of complex adaptive systems adapting to one another's
behaviour, for example investors in a market or organisms in an ecological
system.
According to my system of nomenclature, an industrial district would have to
have a schema of its own to qualify as a complex adaptive system. From what
little I know, I would say that may not generally be the case. Using John Holland's
meaning of complex adaptive system, we would ask whether the component
institutions (such as firms or universities) have schemata, and in general they do.
In my notation, then, an industrial district mayor may not be a complex adaptive
system itself, but it certainly is an aggregation of complex adaptive systems,
adapting to one another and to external int1uences.
Science fiction writers and others may speculate that in the distant future a new
kind of complex adaptive system might be created, a truly composite human
being, by wiring together the brains of a number of people. They would
communicate not through language, which - Voltaire is supposed to have said - is
used by men to conceal their thoughts, but instead through sharing all their mental
processes. My friend Shirley Hufstedler says she would not recommend this
procedure to couples about to be married.
The behaviour of a complex adaptive system, with its variable schemata
undergoing evolution through selection pressures from the real world, may be
contrasted with "simple" or "direct" adaptation, which does not involve a variable
schema, but utilizes instead a fixed pattern of response to external changes. A
good example of direct adaptation is the operation of a thermostat, which simply
turns on the heat when the temperature falls below a fixed value and turns it off
when the temperature rises above the same value. It just keeps mumbling "It's too
hot, it's too hot, it's just right, it's too cold", and so forth.
In the study of a human organization, such as a tribal society or a business firm,
one may encounter at least three different levels of adaptation, on three different
time scales.
1) On a short time scale, we may see a prevailing schema prescribing that the
organization react to particular external changes in specified ways. As long as
that schema is fixed, we are dealing with direct adaptation.
2) On a longer time scale, the real world consequences of a prevailing schema
exert (in the presence of events that occur) selection pressures on the
competition of schemata and may result in the replacement of one schema by
another.
 What Is Complexity') 19

3) On a still longer time scale, we may witness the disappearance of some

organizations and the survival of others, in a Darwinian process. The
evolution of schemata was inadequate in the former cases, but adequate in the
latter cases, to cope with the changes in circumstances.
It is worth making the elementary point about the existence of these levels of
adaptation because they are often confused with one another. As an example of the
three levels, we might consider a prehistoric society in the U.S. Southwest that had
the custom (1) of moving to higher elevations in times of unusual heat and
drought. In the event of failure of this pattern, the society might try alternative
schemata (2) such as planting different crops or constructing an irrigation system
using water from far away. In the event of failure of all the schemata that are tried,
the society may disappear (3), say with some members dying and the rest
dispersed among other societies that survive. We see that in many cases failure to
cope can be viewed in terms of the evolutionary process not being able to keep
pace with change.
Individual human beings in a large organization or society must be treated by
the historical sciences as playing a dual role. To some extent they can be regarded
statistically, as units in a system. But in many cases a particular person must be
treated as an individual, with a personal intluence on history. Those historians
who tolerate discussion of contingent history (meaning counterfactual histories in
addition to the history we experience) have long argued about the extent to which
broad historical forces eventually "heal" many of the changes caused by
individual achievements - including negative ones, such as assassinations.
A history of the U.S. Constitutional Convention of 1787 may make much of the
contlicting interests of small states and large states, slave states and free states,
debtors and creditors, agricultural and urban populations, and so forth. But the
compromises invented by particular individuals and the role that such individuals
played in the eventual ratification of the Constitution would also be stressed. The
outcome could have been different if certain particular people had died in an
epidemic just before the Convention, even though the big issues would have been
the same.
How do we think about alternative histories in science? Is the notion of
alternative histories a fundamental concept?
The fundamental laws of nature are:
(1) the dynamical law of the elementary particles - the building blocks of all
matter - along with their interactions and
(2) the initial condition of the universe near the beginning of its expansion some
ten billion years ago.
Theoretical physicists seem to be approaching a real understanding of the first
of these laws, as well as gaining some inklings about the second one. It may well
be that both are rather simple and knowable, but even if we learn what they are,
that would not permit us, even in principle, to calculate the history of the universe.
The reason is that fundamental theory is probabilistic in character (contrary to
what one might have thought a century ago). The theory, even if perfectly known,
predicts not one history of the universe but probabilities for a huge array of
alternative histories, which we may conceive as forming a branching tree, with
20 M. Gell-Mann

probabilities at all the branchings. In a short story by the great Argentine writer
Jorge Luis Borges, a character creates a model of these branching histories in the
form of a garden of forking paths.
The particular history we experience is co-determined, then, by the fundamental
laws and by an inconceivably long sequence of chance events, each of which
could turn out in various ways. This fundamental indeterminacy is exacerbated for
any observer - or set of observers, such as the human race - by ignorance of the
outcomes of most of the chance events that have already occurred, since only a
very limited set of observations is available. Any observer sees only an extremely
coarse-grained history.
The phenomenon of chaos in certain non-linear systems is a very sensitive
dependence of the outcome of a process on tiny details of what happened earlier.
When chaos is present, it still further amplifies the indeterminacy we have been
discussing.
Some years ago, at the science museum in Barcelona, I saw an exhibit that
beautifully illustrated chaos. A non-linear version of a pendulum was set up in
such a way that the visitor could hold the bob and start it out in a chosen position
and with a chosen velocity. One could then watch the subsequent motion, which
was also recorded with a pen on a sheet of paper. The visitor was then invited to
seize the bob again and try to imitate exactly the previous initial position and
velocity. No matter how carefully that was done, the subsequent motion was quite
different from what it was the first time. Comparing the records on paper
confirmed the difference in a striking way.
I asked the museum director what the two men were doing who were standing in
a corner watching us. He replied, "Oh, those are two Dutchmen waiting to take
away the chaos". Apparently, the exhibit was about to be dismantled and taken to
Amsterdam. But I have wondered ever since whether the services of those two
Dutchmen would not be in great demand across the globe, by organizations that
wanted their chaos taken away.
Once we view alternative histories as forming a branching tree, with the history
we experience co-determined by the fundamental laws and a huge number of
accidents, we can ponder the accidents that gave rise to the people assembled at
this conference. A fluctuation many billions of years ago produced our galaxy, and
it was followed by the accidents that contributed to the formation of the solar
system, including the planet Earth. Then there were the accidents that led to the
appearance of the first life on this planet, and the very many additional accidents
that, along with natural selection, have shaped the course of biological evolution,
including the characteristics of our own subspecies, which we call, somewhat
optimistically, Homo sapiens sapiens. Finally we may consider the accidents of
genetics and sexual selection that helped to produce the genotypes of all the
individuals here, and the accidents in the womb, in childhood, and since that have
helped to make us what we are today.
Now most accidents in the history of the universe don't make much difference to
the coarse-grained histories with which we are concerned. If two oxygen
molecules in the atmosphere collide and then go off in one pair of opposite
directions or another, it usually makes little difference. But the fluctuation that
 What Is Complexity') 21

produced our galaxy, while it too may have been insignificant on a cosmic scale,
was of enormous importance to anything in our galaxy. Some of us call such a
chance event a "frozen accident". It produces substantial effects, if only in a
limited region of space and time. Once it happens and its outcome is determined,
an accident no longer belongs to the realm of the random. Frozen accidents,
affecting significantly things judged to be important, give rise to regularities. Of
course the fundamental laws also contribute to regularities, but those laws are
thought to be simple, and so effective complexity comes mainly from frozen
accidents.
We can take an example from modern human history to illustrate the idea of a
frozen accident. It was mentioned earlier that a few historians are willing to
discuss "contingent history", in which one asks "What if ..... '1". One of their
favourite incidents occurred when Buffalo Bill's Wild West Show was touring
Europe in 1889. Of course a star attraction of the show was the female
sharpshooter Annie Oakley, who would ask for a male volunteer so that she could
knock the ash off his cigar with a bullet. Normally there were no volunteers, and
her husband, himself a famous marksman, would step forward and stand still
while Annie shot, hitting the ash but not her husband. On this occasion, however,
there was a volunteer, the Kaiser - Wilhelm der Zweite - with an expensive
Havana cigar. Annie was a little worried, having drunk heavily the night before,
but she fired and we know the result. The Kaiser survived, fired Bismarck,
cancelled the reinsurance treaty with Russia, engaged in competition in naval
construction with Great Britain, and in other ways laid the groundwork for the
First World War.
The dominant role played by frozen accidents in generating effective
complexity can help us to understand the tendency for more and more complex
entities to appear as time goes on. Of course there is no rule that everything must
increase in complexity. Any individual entity may increase or decrease in effective
complexity or stay the same. When an organism dies or a civilization dies out, it
suffers a dramatic decrease in complexity. But the envelope of effective
complexity keeps getting pushed out, as more and more complex things arise.
The reason is that as time goes on the results of frozen accidents can keep
accumulating. If that process outstrips the erasure of the consequences of frozen
accidents, then more and more effective complexity can arise. That is so even for
non-adaptive evolution, as in galaxies, stars, planets, rocks, and so forth. It is well
known to be true of biological evolution, where in some cases higher effective
complexity probably confers an advantage. And we see all around us the
appearance of more and more complex regulations, instruments, computer
software packages, and so forth, even though in many cases certain things are
simplified.
The tendency of more and more complex forms to appear in no way contradicts
the famous second law of thermodynamics, which states that for a closed
(isolated) system, the average disorder ("entropy") keeps increasing. There is
nothing in the second law to prevent local order from increasing, through various
mechanisms of self-organization, at the expense of greater disorder elsewhere.
(one simple and widespread mechanism of self-organization on a cosmic scale is
22 M. Gell-Mann

provided by gravitation, which has caused material to condense into the familiar
structures with which astronomy is concerned, including our own planet).
Here on Earth, once it was formed, systems of increasing complexity have
arisen as a consequence of the physical evolution of the planet over some four and
half billion years. biological evolution over nearly four billion years, and - over a
very short period on a geological time scale - human cultural evolution.
The process has gone so far that we human beings are now confronted with
immensely complex problems and we are in urgent need of better ways of dealing
with them (the role of industrial clusters and their potential constitute, of course,
an example of a complex issue). When we attempt to tackle such difficult
problems, we naturally tend to break them up into more manageable pieces. That
is a useful practice, but it has serious limitations.
When dealing with any non-linear system, especially a complex one, it is not
sufficient to think of the system in terms of parts or aspects identified in advance,
then to analyse those parts or aspects separately, and finally to combine those
analyses in an attempt to describe the entire system. Such an approach is not, by
itself, a successful way to understand the behaviour of the system. In this sense
there is truth in the old adage that the whole is more than the sum of its parts.
Unfortunately, in a great many places in our society, including academia and
most bureaucracies, prestige accrues principally to those who study carefully some
aspect of a problem, while discussion of the big picture is relegated to cocktail
parties. It is of crucial importance that we learn to supplement those specialized
studies with what I call a crude look at the whole.
Now the chief of an organization, say a head of government or a CEO, has to
behave as if he or she is taking into account all the aspects of a situation, including
the interactions among them, which are often strong. It is not so easy, however, for
the chief to take a crude look at the whole if everyone else in the organization is
concerned only with a partial view.
Even if some people are assigned to look at the big picture, it doesn't always
work out. Some time ago, the CEO of a gigantic corporation told me that he had a
strategic planning staff to help him think about the future of the business, but that
the members of that staff suffered from three defects:
1) They seemed largely disconnected from the rest of the company.
2) No one could understand what they said.
3) Everyone else in the company seemed to hate them.
Despite such experiences, it is vitally important that we supplement our
specialized studies with serious attempts to take a crude look at the whole.
I am sure that observation applies to industrial clusters.
At many research institutions, theorists are busily engaged in constructing
simple models of quite complex real systems. Today, those are usually computer
models. How, though, does one rate the success of such a model in helping us
learn about the real world? Normally, a theory is compared to observation and the
value of the theory is judged by the degree of agreement with observed facts,
along with criteria such as elegance, consistency, and compatibility with well-
established principles. However, when we are dealing with a highly simplified
 What Is Complexity? 23

model of a very complex system, detailed agreement with observation could well
be embarrassing rather than encouraging.
At the Santa Fe Institute, we have had a number of discussions of this question:
what can be claimed on the basis of a simplified model? One outcome of those
discussions is the conclusion that one should look for rather simple regularities in
the data and see if the model predicts some or all of them. This semi-empirical
"middle level theory" can be enormously valuable. Often, the regularities in
question will hold along the whole spectrum from the complex real world down
through successive simplifications of that world to the model being examined. If
that is the case, then the reason for the regularity in the simple model may be easy
to understand and it can throw light on how the regularity arises in the real world.
Examples of such simple regularities certainly include scaling laws, in which
one quantity of interest appears as a power of another such quantity. These laws
crop up in all fields of science: physical, biological, and behavioural. A striking
example is the rule, verified with increasing confidence over most of a century,
but unexplained until recently, that the metabolic rate of a mammal species goes
like the three-quarters power of the mass. The mammal in question can range from
a tiny shrew to an elephant or a whale. If one mammal weighs ten thousand times
as much as another, it will process energy at a rate one thousand times higher.
Similar laws (with the same power but different coefficients) apply to birds and to
higher plants.
This law was explained by Geoffrey West, a theoretical physicist at the Los
Alamos National Laboratory, and James Brown, an ecology professor at the
University of New Mexico in Albuquerque, along with his student Brian Enquist.
These researchers met and collaborated at the Santa Fe Institute, in a perfect
example of how that institute is supposed to function.
They modelled (analytically rather than on a computer) the cardiovascular and
respiratory systems of the animals and the vascular system of the plants. They
based their model on three-dimensional fractals, and they used two facts:
1) in the branchings of the tubes, areas are preserved and
2) the thinnest tubes all have about the same diameter, independent of the mass.
The scaling law then followed, with the fraction 3/4 arising from the number of
spatial dimensions divided by that number plus one.

Sometimes scaling laws are laws of distribution, for example the famous Zipf's
law - found in fact by Auerbach in 1913 - for the populations of "cities" (actually
metropolitan areas). It states that if we rank the cities in order of decreasing
population, then the population is roughly inversely proportional to the numerical
rank. The data are a bit ragged for the very largest cities, but they agree quite well
with the law for the next several hundred cities. Thus the 300 lh city has about one-
third the population of the 100th city.
Recently Xavier Gabaix published an explanation of Zipf's law starting from
Gibrat's law, which is known to be true and states that the mean growth rate and
mean square deviation of growth rates for metropolitan areas are independent of
population to a good approximation. Gabaix showed that the steady state
distribution of populations then obeys Zipf's law.
24 M. Gell-Mann

Besides scaling laws, we can look at other examples of simple middle level
semi-empirical laws, for instance implicational ladders. Around 1940, when the
USA was still not a belligerent in the Second World War, some sociologists were
asked to measure the attitudes of Americans toward Great Britain. They
formulated a long list of questions with yes or no answers, and those questions
were posed to a representative sample of American citizens. The results showed a
fascinating regularity. If the questions were arranged in a certain order, then most
of the returned questionnaires showed responses that were affirmative down to a
certain point and then negative after that. Thus most of the sets of answers to the
questions could be characterized by a single parameter, the point in the list where
the answers switched from positive to negative. As far as the questionnaire went,
attitude toward the United Kingdom could be summarized in a single number.
Such a result is often called an implicational scale (not to be mixed up with
scaling laws, as described earlier). To avoid confusion, I like to call it an
implicational ladder.
Another example can be cited from linguistics. There are some countries, such
as Haiti and Jamaica, where a Creole language is spoken that gets its words
mainly from the official language of the country. In Haiti, nearly everyone speaks
a French-based Creole language at home, while the official language is, or was
until recently, French. Likewise, in Jamaica, the official language is English bUl
many people speak an English-based Creole at home (this is in contrast to the
situation on the island of St. Lucia, for example, where the official language is
English but the Creole is French-based).
In 1961, the linguist DeCamp investigated the situation in Jamaica, where,
under the influence of official English in the schools, there is a tendency for
certain people to substitute usages from regular English for those in the broadest
or deepest Creole. He chose six cases of such usages, four involving vocabulary
and two involving pronunciation. He discovered that those usages could be
arranged in a particular order such that some speakers used the standard English
version in all six cases, some in the first five, some in the first four, some in the
first three, and so forth. Other patterns were negligibly rare. Thus, out of all the 64
possible patterns for choices in the six cases between Creole usage and Standard
English usage, only 7 patterns existed with appreciable frequency among the
population, and those seven formed an implicational ladder.
If I had to model industrial clusters, I would search carefully for "middle-level"
semi-empirical rules that might persist all the way from the real world down to the
highly simplified model, and test the model by seeing if it clarifies how the rules
arise.
Complex Adaptive Systems
and Spontaneous Emergence

John H. Holland

Abstract. This paper introduces concepts associated with complex adaptive

systems (cas), linking those concepts at some points to economic planning. The
paper begins (section 1) with an informal description of the notion of a cas and
then (section 2) discusses the critical role of "building blocks" in understanding
cas. Using these ideas, the paper goes on (section 3) to discuss the phenomenon of
"emergence", wherein the whole of the system's behaviour goes beyond the simple
sum of the behaviours of its parts. The body of the paper (sections 4 through 6)
looks at the role of modelling in predicting the behaviour of cas, examining the
kinds of model that will serve this purpose. The paper concludes (section 7) with a
brief discussion of the relevance of these ideas to economic planning.

1. Complex Adaptive Systems

A complex adaptive system (cas) exhibits three distinguishing characteristics:

(i) A cas consists of a large number of interacting components, usually called
agents. The agents may range from firms in an economy or participants in a
market to antibodies in the immune system or signalling proteins 111 a
biological cell.
(ii) The agents in a cas interact in non-additive (non-linear) ways. The
interactions can be specified by associating a set of condition/action rules
with each agent, where each agent's rules describe its strategy for
interacting with other agents (the counterpart of a strategy for playing a
game like chess). Such rules range from a simple stimulus-response form,
IF stimulus x THEN make response y, to message processing rules, IF
message x THEN send message y. Because message processing rules can
implement any program that can be written for a general-purpose computer,
this way of specifying interactions is fully general. Often, the rules are
organized to form an internal model of the agent's external world, allowing
the agent to anticipate the future.
(iii) The agents in a cas adapt or learn. That is, they modify their rules as
experience accumulates, searching for improvements. Learning to play a
complicated game, such as chess, provides an example: as one learns to
play, some rules begin to implement sub goals ("develop your pieces during
the opening stages") and look ahead ("if I move that bishop, I will lose a
26 1. H. Holland

rook to a knight's fork three moves from now"). As we will see (section 2)
discovering new, plausible rules depends upon finding appropriate
"building blocks" for describing parts of the cas agents.

2. Building Blocks

Building blocks are the pervasive, critical foundation of an ability to act with
insight in a complex world. Human perception, for example, consists primarily in
combining well-known, simple components to describe familiar phenomena.
Different trees are described by different arrangements of familiar parts: leaves,
stems, branches, and trunks. Similarly, human faces are composed of variants of
standard parts - hair, forehead, eyebrows, eyes, nose, and so on. A little thought
shows that we approach all objects, familiar and unfamiliar, via combinations of
familiar building blocks.
We can also extract building blocks as the generators of the dynamic situations
that characterize the interaction of agents. Consider the rules of chess: the rules are
building blocks that generate a miniature artificial world in which two agents
interact. Though chess is simply defined - less than a dozen rules suffice - it is an
artificial world of perpetual novelty. We never see the same game played twice
unless some previous game is deliberately recorded and repeated. The game is so
complex that, after centuries of study, we have no idea of what a "best strategy"
would be for playing the game. Nevertheless, in this perpetually novel world,
planning and anticipation are possible, indeed essential.
Planning in chess depends upon extracting and exploiting certain patterns
(configurations of pieces) that occur repeatedly in the play of the game. These
patterns become higher-level building blocks from which to construct strategies
(plans) for playing the game. Over the centuries, chess players have repeatedly
discovered new repeating-pattern building blocks that make possible ever more
sophisticated strategies. For example, early in the 20th century, chess players
discovered that certain interlocking pawn structures could exhibit strong influence
on the play of the game. As a result the pawns in today's game do a lot more than
they did in the 19th century. By adding this pawn-structure building block to the
repertoire, a present-day master chess player can easily outplay a 19th century
player of equal rank.
Lest this discussion of games seem frivolous, note that the rules of a game are
not different in kind from the axioms that generate Euclidean geometry, or the set
of computer instructions that define a flight simulator for a new aircraft. Indeed if
we look to that most complex of organizations, the biological cell, we see layer
after layer of building blocks. At a low level we have the 4 nucleotides that form
the building blocks from which all chromosomes are constructed (Note that a
physicist would add several levels of building blocks below this level, ranging
from atoms down to quarks). One level up we have the 20 amino acids, coded by
triplets of nucleotides, that form the building blocks for proteins. The proteins, in
 Complex Adaptive Systems and Spontaneous Emergence 27

turn, have substructures such as alpha-helices and beta-sheets that determine their
three dimensional structure. From there we go to small interacting groups of
proteins, such as the Krebs cycle, on to building blocks for membranes and
tubules, to organelles, and on up through the hierarchy of living organisms.
If we turn from organisms to artefacts, we see that most major innovations in
science and technology amount to new combinations of old building blocks. For
example, the internal combustion engine, that great source of mobile power that
transformed the 20th century, was simply a new combination of building blocks
that had been around for a long time: the carburettor was a modification of
Venturi's perfume sprayer, the spark plug was presaged by Volta's sparking
device, gear wheels and pistons had been known from antiquity, and so on. The
innovation was the combination of these known building blocks in a way that
provided new interactions between them.
While innovations typically come from new combinations of familiar building
blocks, once in a while there is a genuinely new building block. Though such
events are rare, they can have dramatic impact: the transistor revolutionized the
design and production of electronic circuits. However, even the transistor is a new
combination of well-known building blocks one level down; there was a theory of
semi-conduction many decades before that idea was combined with other ideas to
yield this new switching device. New building blocks often arise through "cross-
breeding" successful lower-level building blocks, blocks drawn from previously
unassociated areas or disciplines.
In sum: whether it be natural or artificial systems, building blocks are almost
always the basis of understanding and innovation. Moreover, the resulting
structures almost always have a hierarchical format, with selected combinations of
building blocks at one level serving as building blocks one level up.

3. Emergence

It is a commonplace that a small number of well-chosen building blocks can be

combined to generate a vast array of interesting, non-random structures. The 26
letters of the English alphabet suffice to generate the vast literature in English, past
and future. The 20 amino acids generate the unending array of proteins that form
the basis of life. Chess, defined by fewer than a dozen rules, offers new patterns of
play after centuries of study, and the geometry defined by Euclid's five axioms
surprises us with new theorems after two millennia of study. This perpetual novelty
generated by finite means puts us at the threshold of the phenomenon called
emergence.
There is no accepted definition of emergence, even among scientists, but few
who have seriously studied such phenomena believe it to be an "eye-of-the-
beholder" effect. Indeed, it is possible to list criteria that go far toward
distinguishing some observation as emergent, regardless of the observer and the
time of discovery.
28 J. H. Holland

First among these criteria is the requirement that the phenomenon be a repeating
pattern in a system that exhibits perpetual novelty. This repeating pattern is a
feature or property common to some subset of the unending building-block
combinations. In mathematical terms, the repeating pattern is an equivalence class
within the subset of combinations. Let us call such a repeating pattern a regularity.
In the dynamics of a cas, regularities typically distinguish commonalities among
some of the agents. When these regularities are reinforced by interactions among
the agents, they become persistent, providing possibilities for "speciation" and
selection. For example, the Krebs (citric acid) cycle is a regularity common to all
living aerobic organisms. This dynamic cycle, mediated by eight enzymes (rule-
like building blocks), underpins the production of all the basic components of the
cell: amino acids, lipids, and the like. The persistent regularity we call the Krebs
cycle becomes an emergent building block making possible higher levels of
organization, ultimately yielding the organelle-cell-organ hierarchy common to all
metazoans.
This example suggests the second criterion: emergent phenomena exhibit a
hierarchical organization wherein selected combinations of building blocks at one
level become building blocks at a higher level of organization. Herb Simon's
famous "watchmaker" parable (Simon 1996) makes clear the great advantages
hierarchical organization confers on dynamic systems undergoing selection. The
"adaptive" part of a cas assures the prevalence of this kind of organization under
emergence.
Once we allow levels of organization, it becomes natural to ask if, or how, the
different levels interact. Clearly living organisms and complex artefacts, such as
computers or airplanes, exhibit such organization, with interactive signalling in
both bottom-up and top-down directions. For example, a single gross movement of
the control column in an airplane sends signals that affect dozens of devices
throughout the airplane (a top-down effect), while the malfunction of a single
critical part, say a warning light, can ground the plane (a bottom-up effect).
Similarly, though an overall market indicator (say the Dow-Jones average) is an
average of the actions of individual investors (a bottom-up effect), a change in that
indicator can exert a top-down influence on the actions of individual investors
(agents). In general, agents at higher levels in a cas hierarchy are composed of
agents at lower levels. Here then is a third criterion: the overall form and
persistence of an emergent regularity depends upon both bottom-up and top-down
effects.
The fourth criterion sounds mysterious, but is quite simple when understood: the
whole emergent regularity is more than the sum of its parts. This criterion contrasts
with the simple reductionist strategy for parsing phenomena: study the building
blocks, and then sum their behaviours to get the behaviour of the whole.
Throughout the history of science this simple reductionist strategy has worked over
and over again, from the description of gases as colliding particles to the
superposition of waves of different frequencies to describe complex sounds.
Trends, polls, and averages, as predictors, all exemplify this additive approach.
However, the non-linear interactions of the agents in a cas short-circuit simple
 Complex Adaptive Systems and Spontaneous Emergence 29

reductionism. We cannot add up the behaviours of the agents to get the behaviour
of the whole because the interactions are not additive. Reduction still works, but it
is not simple because the interactions are not simple. The behaviour of the whole
cannot be obtained by simply adding together the behaviours of the parts.
These four criteria go a long way toward separating the phenomena I would like
to call emergent from most other phenomena. The criteria are subject to explicit
testing procedures, so emergent phenomena satisfying these criteria are not an eye-
of-the-beholder effect.

4. Models and Prediction

Though modelling is the subject of an extensive literature with rigorous

mathematical underpinnings (homomorphism), I will only comment informally on
a few aspects relevant to the modelling of cas and agents.
A first principal: models, like theories, suggest where to look for answers to
some previously formulated question or conjecture. Accordingly, anything
considered irrelevant to the question should be considered a detail that can be
eliminated from the model. A model much like a political cartoon, makes its point
by exaggerating certain features while eliminating incidentals. And a model, like a
map, is only serviceable when it contains a limited amount of detail.
When we model cas, the models are perforce-dynamic models. The object is to
find a few mechanisms and laws (similar to the pieces and rules of a game) that
generate a description of the cas's changing states (configurations). A typical
guiding question will concern the occurrence of some set of regularities (see
section 3) in these unfolding configurations. For example, how does a vaccine
confer lifetime immunity on the continually changing immune system? Or, more
generally, why do almost all cas exhibit lever point phenomena, where small
"inexpensive" inputs cause major directed effects in the cas dynamics?
It is an art to select a level of detail that allows simple laws. Such laws, when
selected with insight, generate state-sequences (configuration-sequences) that
faithfully describe, or at least approximate, the behaviour of the system at that
level of detail. In weather forecasting, discovering the laws of movement for
regularities called fronts made possible dynamic models yielding predictions an
order of magnitude more accurate than previous statistical models.
The quintessential exhibits of. this art form are the physicist's gedanken
experiments. Einstein's formulation of thought experiments illustrating the difficult
questions in quantum mechanics are classic (Jammer 1974). In such models the
details are pared to the point that the model can be examined by thought alone.
Cas are generally too complex to allow such elegant formulations, but we can
build computer-based models that serve a similar exploratory purpose. Computers,
of course, allow the construction and execution of models of remarkable detail.
However, this facility also poses a great danger. Because the computer can handle
the detail, there is a strong impulse to add detail upon detail. It is possible to wind
30 1. H. Holland

up in the posltlon of Borges' king (Borges 1964) with a map the size of the
country. The essence of elegant science is providing just enough detail to explore
the question posed.
Once a question has been posed, it is typical to start with a very simple model,
quite likely an over-simple model. You then see how far the model will take you
toward answering the question, only adding a new mechanism or law (see pieces
and rules in a game) when there is a clear need. This process of going as far a you
can with a simple model, in order to see what's missing, can be quite productive.
And it does produce models that are just detailed enough to answer the question.

5. Modelling Agents

In a cas each agent is surrounded by, and interacts with, a multitude of other
agents. It is convenient to think of all of an agent's environment as represented by
agents, some of which are passive and do not adapt (such as a rock or a chair).
There is no real loss of generality in such a conceit.
Each adaptive agent receives information about its environment via a set of
detectors (receptors, gauges, or the like). We can think of the output of a set of
detectors as a standardized packet of information, called a message. An agent's
internal information processing can also be treated in terms of messages (much
like the memos circulated in an office). Some messages may be directed to other
agents or to effectors (machines, muscles, or the like), provi~ing the interactions
that change the agent's environment. In short, we can look upon an agent as a
message processing device. In consequence the overall cas becomes a complex
signalling network.
One formalization of this outlook is provided by the adaptive rule-based systems
called classifier systems (Holland 1995,2000). The building blocks for a classifier
system are IFrrHEN rules called classifiers. The condition (IF) part of the rule
"looks for" certain kinds of messages; when the rule's conditions are satisfied, the
action (THEN) part specifies a message to be sent. For example, if the agent has a
visual system, one of the rules might be IF there is (a message from the detectors
indicating) an object left of centre in the field of vision, THEN (by issuing a
message to the effectors) cause the eyes to look left.
In a classifier system many rules may be active simultaneously because
additional active rules simply means more messages being issued at a given
instant. From a computational point of view, it is convenient to think of the
messages present at a given instant as collected in a list called a message list. As a
rough analogy, think of the rules in a classifier system as standard operating
procedures. Then think of an agent as an office containing many desks, where each
desk (rule) is responsible for executing a particular operating procedure. In this
analogy, the message list is analogous to a bulletin board that contains all the
memos (messages) directing the activities of the office that day. The action of each
desk, when there is a memo for it on the bulletin board, is to produce a memo for
 Complex Adaptive Systems and Spontaneous Emergence 31

the next day. In classifier system terms, each rule is looking to the message list, to
see if there are messages satisfying its conditions (the IF part). If the rule's
conditions are satisfied by messages on the list, it generates a new message (the
THEN part) to be added to the list.
It is often desirable to direct a message to a given rule or set of rules, much like
the header on an email message selects a set of destinations for the email message.
To select destinations in a classifier system we use tags. Tags for this purpose are
quite common in the natural world. For example, in molecular biology, we have
amino acid sequences that bind to a specific locus in DNA. The DNA loci are the
counterparts of rules that are repressed (cease to act) by proteins (messages)
having particular tags. Tags play similar roles in identifying antigens for the
immune system, they also serve to define active sites in an enzyme, and tags serve
as "headers" for the signalling proteins that mediate reaction cascades and
complicated feedback loops in biological cells. Rules using tags may represent
DNA loci that are repressed by signalling proteins (messages) having particular
amino acid sequences (tags) that provide binding action, or rules may serve as
steps in a cascade wherein a protein initiates the formation of a new protein in the
cascade (sending a signal), or the rules may interact in much more complicated
feedback loops.
In a classifier system using tags rules typically have the form IF (signal with
apropos tag present) THEN (send signal with new tags).
The condition part of a classitier rule will often specify several conditions, so
that the rule requires the simultaneous presence of ditferent tagged messages
before it acts. If an agent has a modular, hierarchical structure, the rules in each
module can be directly identified by a common tag. That is, to send a message to a
module you attach to it the identifying tag of the module. An appropriately tag-
linked collection of classifier rules can implement any subroutine or program that
can be written for a computer. As a result any hierarchical cas can be implemented
as a classifier system.

6. Cas Models Based on Classifier Systems

Classifier systems address three basic problems in the study of cas:
(1) Parallelism and coordination.
Should an agent use a single, monolithic rule for each situation it could
encounter, the demands on its capacity to store rules would be
overwhelming. No system can afford to retain enough rules to handle
unique situations such as "a red Saab by the side of the road with a flat
tire". On the other hand, such a situation is easily handled by
simultaneously activating rules for the building blocks of the situation:
"car", "roadside", "flat tire", and the like. In short, a rule-based system can
handle a broad range of novel situations if it can act on them with
combinations of "building block" rules. Combinatorics then work for the
32 1. H. Holland

system instead of against it. Moreover, by definition, useful building blocks

are frequently used, so they are frequently tested and confirmed. The
object, then, is to provide for the simultaneous interaction and coordination
of well-confirmed rules that, together, describe the new situation.
Classifier systems address this coordination problem by restricting rule
action to the emission of tagged messages. Complicated logical questions
about "rule consistency" and "consistency maintenance" are avoided
because messages only serve to activate rules. Messages have no other
intrinsic meaning. More messages simply mean more active rules, and vice-
versa.
(2) Credit assignment.
There is no secure way to determine a priori which rules will serve as good
building blocks for generating system behaviour. This means there must be
some means of rating the usefulness of rules as the system accumulates
experience. This is a difficult problem when many rules are active
simultaneously, some helping and some obstructing. The problem is
exacerbated when the environment requires stage-setting actions in order to
achieve some desired outcome (as when one sacrifices a piece in chess to
set the stage for the capture of a major piece later). Solution of the credit
assignment problem is a sine qua non for adaptation.
In realistic situations, exhaustive exploration of all possible action paths is
not feasible, so the credit assignment scheme must be "local". Samuel's
early work (Samuel 1959), using prediction of the value of future options,
points the way, but few have exploited his insights. In classifier systems,
credit assignment is handled by setting up a market situation.
Each rule in the classifier system is assigned a strength that is a measure of
its usefulness. In the credit assignment procedure, this strength serves as a
kind of cash-in-hand. When the condition part of a rule is satistied it makes
a bid to become active, a bid proportional to its strength. If the rule is a
winner in the bidding process, its strength is reduced by the amount of the
bid, and the strengths of rules sending activating messages to it, its
"suppliers", are increased by that amount. That is, the newly active rule
makes a payment that increases the strength of its "suppliers". Then the
newly active rule stands to profit from bids for its newly posted message,
the bids of its "consumers". A rule increases its strength if it makes a profit
in this market (its payments to its suppliers are less than the payments it
receives from its consumers).
In effect, a rule acts as a go-between (broker, middleman) in rule-chains
leading from the current situation to (possibly favourable) future outcomes.
At intervals, some rules have their strength increased when the system
acquires direct payments from the environment (reward, payoff,
reinforcement - e.g., "food", or some payment from an "ultimate
consumer"). The newly strengthened direct-paid rules will make larger bids
the next time this particular chain is executed. These larger bids will
strengthen supplier rules in the chain that set the stage for the direct-paid
 Complex Adaptive Systems and Spontaneous Emergence 33

rules. And so on back through the chain on successive executions. As rules

become stronger they are more likely to win the bidding process, so their
messages are more likely to influence system behaviour. In short, rules
which make a profit in rule-chains leading to direct environmental
payments will come to control the system.
(3) Adaptation via rule discovery.
Rule discovery (or its equivalent) is the most recondite problem the
adaptation of classifier systems. It was not even well handled in Samuel's
remarkable work. We know that rules receiving little credit (low strength)
should be replaced, but random generation of new rules can only work for
the simplest problems. The key to effectiveness is use of past experience to
generate plausible new rules for situations as yet poorly understood.
Classifier systems exploit a genetic algorithm's (Mitchell 1996) ability to
discover and recombine building blocks. Classifier systems have "building blocks"
at two levels: (i) the parts (called schemata) from which the condition and action
parts of individual rules are constructed, and (ii) the rules themselves, as
components of the overall system. A genetic algorithm is designed to "crossbreed"
well-rated (strong, fit) elements in a population, so it is wellsuited to working on
this task at both levels. The strength of a rule determines its likelihood of being a
parent in the crossbreeding process.
Each of the mechanisms used by the classifier system has been designed to
enable the system to continue to adapt to its environment, while using its extant
capabilities to respond instant-by-instant to that environment. In so doing the
system is constantly balancing exploration (acquisition of new information and
capabilities) with exploitation (the efficient use of information and capabilities
already available).

7. Relevance to Organizations (such as Industrial

Districts)

If we think of the agents in a cas as building blocks, then the discussion of

hierarchy in section 3 suggests that we should see hierarchical organization in
most, if not all, cas. Certain combinations of agents, because of efficient
exchanges of resources and signals, acquire a greater persistence. In effect they out
compete other combinations, absorbing resources that might otherwise be
available to those other combinations. Persistent combinations of agents (say
firms) become candidates for agents at a higher level of organization (say
industrial districts).
Agents in a cas typically are not indefinitely persistent, a fact that strongly
affects the organizing process. For example, consider the mammalian body as a
cas: every atom in a mammal is turned over in less than two years, so that its body
is a persistent pattern imposed upon a flow of atoms. This persistent pattern is a
complex version of the standing wave imposed on a flow of water molecules past a
34 1. H. Holland

rock in a white-water river. The same is true of larger human organizations, be

they firms or cities. Firms typically persist over time spans much longer than the
time of employment of an average employee, and cities persist over spans of time
much longer than the lifespan of their inhabitants. A cas, rather than being a
"bricks and mortar" structure, is primarily a pattern imposed on flows of resources
and signals.
Under this view, adaptation in a cas is facilitated by a redirection of t10ws to
provide enhanced persistence. Resource exchanges with surrounding agents (the
equivalent of mutualism and symbiosis in ecosystems) provide for exploitation of
strategic discoveries already made, while the modification of extant agents, and the
creation of new agents, provides for further exploration and discoveries. Because
the agents in a cas, at whatever level, are continually adapting, innovation is a
regular feature. This continual innovation makes the balance between exploitation
and exploration a critical factor in the persistence of agents.
One of the least understood, but pervasive, cas properties is the "lever point". A
"lever point" is a kind of flow redirection that causes long-term directed changes at
a small expenditure of resources. For example, an inexpensive vaccine can cause a
lifelong change in the mammalian immune system. All cas that have been
examined closely to date exhibit lever points. However, we have no principled way
of searching for lever points. It is here that good models and, ultimately, theory
should tell us "where to look". As with vaccines, lever points offer opportunities
for making substantial changes in "locked-in" situations. Without such knowledge,
we often wind up "throwing money" at a cas, with little or no effect on the
outcome.
In economic planning, knowledge of the relevant lever points can make the
difference between resounding success and stagnation or failure.

References

Borges, J. L. (1964), "On Rigor in Science", Dream Tigers, Austin, U. Texas Press.
Holland, J. H. (1995), Hidden Order, Reading, MA, Addison-Wesley.
Holland, J. H. (2000), "What is a Learning Classifier System?", in: Learning Classifier
Systems. Lanzi, P. L., Stolzmann, W. and Wilson, S. W. (eds), Berlin, Springer, pp. 3-6.
Jammer, M. (1974), The Philosophy of Quantum Mechanics, New York, Wiley.
Mitchell, M. (1996), An Introduction to Genetic Algorithms, Cambridge, MA, MIT Press.
Samuel, A. L. (1959), "Some Studies in Machine Learning Using the Game of Checkers",
in: Computers and Thought, Feigenbaum, E. A. and Feldman, J. (eds), New York,
McGraw-Hill.
Simon, H. A. (1996), The Sciences of the Artificial. Cambridge, MA, MIT Press, 3rd ed ..
The Industrial Cluster
as a Complex Adaptive System

Enzo Rullani

Abstract. In the economic debate of recent years, the rediscovery of industrial

clusters is concurrent with the rediscovery of complexity.
And with good reason.
These two issues regained popularity at the same time, after years of neglect,
because they both underscore, in different ways, the failure of the idea of
modernity derived from Fordism and the need for a new vision in which reason is
unshackled from the deterministic prejudice that results from the extraordinary
success enjoyed until now by a system based on the control of technology and
financial success. Neo-modern reasoning, or post-modern reasoning, as some
prefer to call it, should salvage the worthwhile aspects of modern tradition -
rational criticism, dialog communications and universalism of values and rights -
integrating these values with an attitude that is more open to the exploration,
interpretation and invention of possibilities that, while outside our control, can be
important and yield the benefit of a rational exploitation of what we already know
and control (Rullani 1996).
The notion of modernity that shaped the imaginary edifice of Fordism was the
expression of a deterministic reasoning that sought to exploit to the fullest what
was already known. The step that is being proposed today, which brings
complexity back into play, is to develop alongside this line of reasoning another
one, more dangerous but more promising, based on the exploration of the world of
possibilities (Axelrod and Cohen 1999, p. 43).
According to J. March (1991, p. 71):
"exploration includes things captured by terms such as search, variation,
risk taking, experimentation, play, flexibility, discovery, innovation.
Exploitation includes such things as refinement, choice, production,
efficiency. selection, implementation, execution".
In order to govern complexity, we must explore and capture what is new, but also
exploit the results of what has been learned each time. If we were to just explore,
we would have to bear all the costs of the acti vities needed to research, interpret
and test that which is new, but we would reap only a fraction of the benefits that
could be obtained from the resulting innovations. The choice of focusing only on
exploitation, with a minimal interest in exploration, would be equally
unsatisfactory, because if we were to limit ourselves to exploiting rationally what
we already know, we would find ourselves trapped in some sub optimum
equilibrium zone from which we could not escape even if we were confronted
with exceptional opportunities.
36 E. Rullani

Between exploitation (of what is known) and exploration (of what is new) there is
a border that neo-modern reasoning must cross over repeatedly as it seeks useful
opportunities, without ever stopping at that methodological borderline beyond
which is a land that until a short while ago was marked "Danger! Unknown".
The relationship with complexity - and, as it concerns us here, with industrial
clusters - brings into relief a basic alternati ve between rationalizing that which is
old and exploring that which is new one that all systems of learning must· face
sooner or later and that, not accidentally, represents a "place" (topos) that is
typical of the more general studies of complexity (Holland 1988, 1995).
During the past century, when deterministic reasoning was prevalent, the focus
was primarily on computing and optimising what was already known, reducing the
exploration of that which was new to a routine that administered and controlled all
innovative acts, neutralizing their subversive energy. In this sense, the century of
Fordism was characterized by hostility to complexity, which was seen as a source
of uncertainty and risk and, therefore, an obstacle to rationalization. Increasingly
codified and rigid forms of forecasting, planning and control were developed in
order to minimize the exposure to complexity, which artificially reduced
opportunities and interest in exploring that which is new. In both theory and
practice, the emphasis was on rationalizing that which existed by relying on
bureaucracies and automatisms. Complexity was tackled by breaking it up into
fragments and forcing these fragments into predetermined computation
(technology, markets), control (plans) and negotiation (management) procedures.
To do this, economic behaviour, which could be put through computation and
optimisation procedures more easily than other activities, was separated from
everything else. Economics became the science of efficiency, specialized in
exploitation and removed from exploration. Its complexity was drastically reduced
by assigning all its anthropological, social and institutional variables to the realm
of exogenous variables, which left economics with just one task: maximize the
efficient use of known means to achieve equally known ends.
In this context, the territory was inevitably pushed to the margins of economics
studies. Since it is an elective place where economics, society and history
intersect, the territory contains information and complex relationships that exceed
the sphere of calculable actions. One could even say that thinking in the past
century, because it was consumed with reducing complexity, was ultimately also
an enemy of the territory, which was perceived as a system that creates, preserves
and reproduces the complexity of history and experience.
The reason why we are now re-examining with such great interest industrial
clusters and other local phenomena that are rooted in geographical locations and
are rediscovering old studies of territorial economies that were gathering dust in
local archives is because we need to understand how to tackle high levels of
complexity without first breaking it down with computations, controls and
negotiations, since time has shown all these methods to be blunt and increasingly
ineffective instruments.
The rediscovery of complexity entails the rediscovery of the territory and vice-
versa. Choosing to take a pragmatic, experimental approach that allows rationality
to explore the open world of possibilities also means finding a foothold in the
 The Industrial Cluster as a Complex Adaptive System 37

stratified compiexity of an archipelago of many and varied territories (Sabel 1989,

1999). The reference to the plurality of experiences contained in the different
types of territories helps reject a reductive idea of rationality that is limited to the
ability to forecast, compute and control. It also opens the way for the spirit of
exploratory rationality, thereby contradicting the legacy of deterministic
reasoning from the Fordist century.
In hindsight, we can say that the ambition to reduce the economic world to a
computable tool was an illusion from the start. Complexity, which was being
pushed out the door, re-entered shortly afterward through the window. Out of the
limelight, complexity continued to exist unhindered, despite Fordism, in the
economic reaiity of daily life, even though, unquestionably, it was mostly ignored
by economic theory, which relegated it to the margins of attention, labelling it
deviation, exception, error, anomaly, uncertainty, etc. Complexity, although
viewed as a negative factor, continued to playa role in the practical history of
businesses and individuals, actually becoming a problems' problem, the beginning
and end of the concerns of a techno structure that denies its existence and fears its
impact at the same time.
The true story of modernity has two faces: an official one, which shows the work
of players who understand their goals and are rational in the computation of their
means, and an unofficial one, in which complexity resurfaces, pushing through the
meshes of computation and escaping the control of planning. When we look at the
history of great organizations, of business systems, of the webs of alliances and of
territorial formations - even during the long season of Fordism - the dynamics of
these relationships clearly show the complex nature of an experimental type of
reasoning that does not go from goals to means or from computation to result,
choosing instead to intertwine goals and means in a maze of relationships and
circularities in which intentional action follows the path of evolutionary learning.
which proceeds in a zigzag pattern instead of acting in accordance with
predetermined objectives and computations (Di Bernardo and Rullani 1991).
Today, both complexity and clusters are the subject of attention and d;scussion
because one season of modernity has ended and another one is beginning. The new
modernity entails a reconciliation with complexity and with distributed-
intelligence forms of organization, such as clusters, that allow the exploration and
governance of complexity at a reasonable cost.

1. Reducing Complexity: The Goal of Deterministic

Reasoning

The main goal of deterministic reasoning, as it developed during the long reign of
Fordism, was the efficient exploitation of what was known. From this standpoint,
the first order of business was to neutralize the world's complexity, since it
hindered computation and rationalization efforts. In other words, it was necessary
to narrow the spaces occupied by variance and indeterminateness, since they were
the reasons for the unpredictability of natural events, individual choices, meanings
38 E. Rullani

perceived by different people and organizations and behaviours occurring in the

course of economic and social life.
It was clear from the outset - in the years straddling the eighteenth and
nineteenth centuries, which saw the birth of modernity - that these spaces, which
in the premodern world were relatively wide and unregulated, were incompatible
with the yearning of modern reason for full control of destiny.
Consequently, they were regulated and narrowed with the use of the tools
created by the industrial revolution.
With the use of the science and technology incorporated in the machines and
technical systems of factories, which were characterized by the repetitiveness and
standardization of procedures and products.
With reliance on the automatism of the marketplace, which made price the
universal measure of everything and, therefore, forced productive and social life
through a powerful process of quantification and abstraction.
During the 20th century, Fordism accelerated the drive toward technical
rationalization, adopting work methods that were consistent with the scientific
organization developed by F. Taylor and generalizing the use of machines in the
production process in all areas of business, even the most complex.
Thanks to this approach, even inherently complex activities could be
modernized because complexity could be efficiently minimized with the technique
of segmentation: a complex task could be broken down into a sequence of basic
steps, each sufficiently simple to be delegated to a computational algorithm or a
machine. An assembly line or production program could then reassemble the
fragments, increasing complexity only to the level strictly necessary, which,
however, was significantly lower than it had been before (Di Bernardo and Rullani
1991).
In a large Fordist enterprise, planning, by reassembling a puzzle, generated an
artificial environment of reduced complexity. This environment had the
fundamental advantage of making the behaviour of all players controllable, since
the consequences of their actions were computable and the relationships between
the different actions programmable ahead of time.
In this way, computation became wedded to power: what can be computed can
be easily controlled. The reverse is also true: what can be controlled constitutes an
excellent input - free of ambiguities and surprises - for computational procedures.
Modern reasoning thus became technocratic reasoning, which combines the
powers of computation and control. Thanks to these twin powers, technocratic
modernity was able to bring a growing range of events and behaviours into
compliance with predetermined standards and computable algorithms. In those
areas where variance or indeterminateness remained too high, forecasting and
control tools were supplemented with negotiation procedures between organized
interests. To achieve this goal. working against the backdrop of managerial
capitalism triumphant, as it existed until the 1970s, great public and private
institutions ensured the stability and predictability of behaviours using the power
of scientific and technological knowledge, the power of ownership control and the
more flexible and adaptive power of negotiations with stakeholders to reduce
complexity (Galbraith 1967).
 The Industrial Cluster as a Complex Adaptive System 39

2. What Does Territory Mean?

The notion of modernity in which the ability to produce is based on reducing the
"natural" complexity of human and social behaviour has very little in common
with the notion of territory, which represents a synthesis, sedimented at one
location, of history, culture and interpersonal relationships. This synthesis was too
rich in variances and had too many levels of indeterminateness to fit into the
framework of modernity and, therefore, was ignored or completely removed from
the horizon.
For all intents and purposes, the idea of territory as a complex synthesis of life
stratified in one place disappeared from the theory and practice of modern
economics. In its place, modernity offered an artificial space devoid of
complexity where players can easily perform opportunity computations.
Obviously, this metamorphosis caused the notion of territory to lose all the
relevance it had gained in premodern times. This was also because, in the Fordist
vision of modernity, the complexity of local societies could not be used in any
way to create economic value. On the contrary, it was an obstacle and an
impediment. The modernity we have been familiar with until now attempted to
render abstract the available physical space, thereby suppressing the uniqueness,
identity and history of different locations.
When viewed through this filter, the territory becomes an easily fungible space,
indifferent to the quality of the places it contains. In other words, it becomes
something that can be measured in acres of space and miles of distance.
The distribution of activities over the territory - and the very idea of industrial
cluster - thus becomes the product of a localizational algorithm that produces
geographical agglomerations and dispersions that mirror the distribution of costs
and revenues over the different locations, in accordance with a routine
computation of opportunities that is indifferent to the subject matter being
optimised. There is no indication of the existence of places with economic
identities and histories that are different from those of all other comparable places.
The disappearance of complexity from the ideal and technical horizon of
modernity also marked the disappearance of the concept of industrial cluster and
the notion of territory that it entailed. This was unavoidable, but we must be aware
of the theoretical and practical consequences of such a disappearance.
A territory without complexity is a place without qualities. It is just another
place, just another geographical agglomeration or dispersion produced by a
computational algorithm. Consequently, it is easy to see why the scientific and
practical program of modernity would virtually ignore the existence of industrial
clusters despite the prodding of an influential author such as A. Marshall and a
wealth of relatively successful territorial clusters or similar areas that developed
on repeated occasions during the past century.
Industrial clusters were viewed as little more than a geographic oddity or a
sitting anomaly: "typical" areas, or system areas, that for some reason had
spawned a mono-industry culture instead of a well-balanced and diversified multi-
industry economic system. The hidden bond that, in actuality, linked the quality of
40 E. Rullani

the cluster to the quality of the site remained hidden and was basically
uninteresting. At best, it aroused the curiosity of a few, anomalous, theoretical
economists (in Italy, mainly Giacomo Becattini and Sebastiano Brusco).
Today, we know that this small group of scholars was not studying an irrelevant
or backward-looking phenomenon, but that it had uncovered one of the anomalies
that made it possible to observe complexity escaping from the net of deterministic
reasoning. Clusters did not have an authority or official belief that limited the
opportunity space available to their members. On the contrary, everyone was
allowed, at his own risk, to explore a segment of the opportunity space. Each
member, operating within an environment characterized by a great deal of
variance and indetermination, learned to manage complexity whenever it arose,
without attempting to minimize it beforehand. Industrial clusters, with their
minute specializations and strong interactions, can metabolise complexity,
transforming it into learning, rules and collective identities.

3. Rediscovering Industrial Clusters

An industrial cluster develops when a significant number of businesses in the

same area of manufacturing, or in related areas, cluster at the same location, using
their territorial contiguity as a means of interaction and exchange. The territory,
viewed as the place where the shared culture, history and institutions of the
cluster operators are being sedimented, functions as a relational and
communicational frame that can integrate thousands of decentralized and
interdependent intelligences that interact with each other in an aggregate pattern of
behaviour that is organized and efficient - efficient enough to make the clusters
competitive in numerous industries in the modern economy.
The most successful industrial clusters occurred in what used to be called
traditional industries: textiles and garments, footwear, furniture, tiles, light
mechanical engineering (production of machines, equipment and components,
metalworking systems, surface treatment systems, micro-mechanical engineering).
In reality, these industries are far from "traditional" because they do not embody
methods and meanings handed down through tradition and preserved largely
unchanged until the present. Today, considering the technologies and
organizational models employed, we could easily classify these areas of industry
as innovative businesses and as complex businesses, i.e., businesses open to
unpredictability and ready to experiment with the communications and symbols
they offer their customers. Think, for example, of the density and volubility of the
messages that fashion trends associate with objects of daily use that are an integral
part of people's lives.
It is important to clear up a misunderstanding caused by the habitual taxonomies
of industrial economics. Even though industrial clusters helped revitalize artisan
types of professional endeavours and businesses that are far removed from the
Fordist stereotype, they represent a kind of industrial modernity that is rooted in
the contemporary world, even when clusters manufacture products that hark back
 TIle Industrial Cluster as a Complex Adaptive System 41

to consumer needs that already existed in the pre-industrial world or use

manufacturing techniques based on an intensive use of human beings
(entrepreneur or worker) and their skills.
It is totally misleading to associate industrial clusters with traditional forms of
production. Clusters are not competitive in those areas in which tradition is
repeated with unchanging and aproblematic forms, because in the modern world,
production of this type is handled more cost effectively by large companies that
can lower costs by standardizing their products and optimising their
manufacturing cycles. Clusters are effective in those areas in which it is difficult
to standardize and plan production far in advance. They excel in those areas where
operators must adapt quickly and creatively to changes that they cannot control,
exploring new meanings and building around them a structured network of
specializations, transactions and exchanges. Clusters are efficient at this task not
because there is a mastermind that decides what to do and orders the necessary
behaviours, but because they are able to develop decentralized forms of
exploratory and constructive learning as a result of attempts, imitations, mutual
adaptations and communications among the different independent intelligences
that comprise the cluster. At each production phase or function, competitors work
independently but are capable of learning (copying, imitating, duplicating) from
each other. Suppliers organize experimental and flexible relationships with their
customers along the value chain, using the relational channels sedimented locally
over time. Even though it entails a certain amount of "evolutionary waste"
(unsuccessful attempts, redundancies, lock-in circuits), the result is that the web of
relationships is quickly reshaped and remodelled each time, as it adjusts to
environmental changes and to endogenous learning dynamics.
Cluster companies are "extended companies" in the sense that, because they are
part of a broad learning circuit involving the supply chain and local society, they
optimise the use of internal intellectual and financial resources, specializing in the
performance of narrowly defined functions that cannot be performed without a
cluster-type learning (Varaldo and Ferrucci 1997).
The self-organizing that occurs through cluster dynamics does not confer
significant competitive advantages when the task at hand is the repetition of
traditional models and forms. It is useful when it becomes necessary to respond to
unexpected situations, explore unknown paths and follow untested intuitions. To
sum it up: industrial clusters, even when they are rooted in traditional consumer
needs and craftsmanship, constitute a modern way of producing and competing
that is particularly suited to conditions of high complexity.
After all, cluster formations (territorial aggregation and integration of
distributed intelligence) have also occurred in such high-tech industries as
information technology, telecommunications and biotechnology, with the
flowering of new enterprises around large centres to produce the intellectual
capital (higher research and learning) needed to fuel growth and innovation in
these industries.
In Italy, almost one third of manufacturing activity is carried out in ways that, to
varying degrees, reflect the cluster model. More importantly, the prevailing trends
42 E. Rullani

of international trade are leading to a further specialization of Italian industry in

those fields where industrial clusters are more present.
In other countries, industrial clusters are less common. But cluster-related issues
are equally important, because communities everywhere are interested in deriving
the maximum benefits from the connection between production and territory,
between proprietary business resources and resources that are not proprietary but
can become accessible through territorial localization. This is true for the
availability of infrastructures, highly qualified or particularly inexpensive human
resources, or regulatory or tax incentives or advantages. These considerations are
especially true for knowledge, a resource that is of key importance for modern
competition. An important part of the knowledge used in manufacturing has a
localized nature, i.e., it cannot be transferred from one place to another. It is
embedded in a specific place (or, more often, in a specific society), so that it can
be accessed only by establishing a direct relationship, of variable intensity, with
that place (CorD 1998).

4. Out of the Limelight, an Economic Theory of Industrial

Clusters Begins to Develop

The development of industrial clusters, as a physical event, was not recognized for
quite some time. There was an even longer delay in translating the "teachings of
facts" into theoretical awareness.
There is a basic reason for the failure to develop a theoretical explanation for the
economic growth visible in real life in developing clusters. The growth of
industrial clusters is triggered by mobilizing resources that are doubly anomalous
for standard theory: first of all, because they are local resources, unique and very
difficult to duplicate and, secondly, because they are complex resources that
represent the summation of history, culture, relationships and interactions with
unpredictable results and, therefore, create patterns of behaviour that are far from
simple and virtually impossible to calculate.
To acknowledge that industrial clusters are part of the modern economy and not
vestiges of the pre-industrial economy is not an easy thing to do, because such an
acknowledgment entails a conclusion difficult to accept: that modernity can take
complex forms that escape instrumental reasoning and deterministic calculations.
At first, the handful of pioneers in this field received limited attention from
mainstream economists. For some time, clusters were seen as empirical and
theoretical variants that existed only in Italy. Later, it would become apparent that
these variants carried the seeds of a conceptual revolution that was much broader
in scope and extended beyond Italy.
The first clear theoretical demarcation, separating the economy of clusters,
which is location-based, from the classic economy of industries appeared in 1961
in a work by Giacomo Becattini on the relationship between economic value and
industries. Becattini criticized the theory of value based on the notion of industry
because it lacked all those concrete references - relationships with society, history
 The Industrial Clustc:r as a Complex Adaptive System 43

and culture - that exist in localized production processes. Industries are

abstractions that enshrine the primacy of technology over all other determinations:
all factors of production become useful inputs to the extent that they can be
utilized by the technology employed to generate the industry's output. All else is
irrelevant, a mere accident or obstacle to manufacturing efficiency. But real
history suggests a totally different interpretation: the quality of the people, the
places where they live and the societies that shape their actions are of fundamental
importance in suggesting which product should be manufactured and which
process should be employed. Looking beyond the notion of industry, Becattini
saw a capitalism that is different from the one described by conventional theory, a
capitalism that produces value by mobilizing the intelligence of individuals and
their ability to interact within appropriate contexts.
Since then, Becattini has never relinquished this critical approach, challenging
the vast majority of the academic and political establishment. At the centre of his
theory is the industrial cluster, seen as a vital form that facilitates a synthesis
between economy and society. The empirical proof of these theories is provided
by a rise in competitiveness, which, beginning in the 1970s, produced an increase
in the importance and wealth of almost all Italian industrial clusters.
But the hidden subject of this dispute was, in reality, another one. The question
was whether or not its was ,appropriate to include within the boundaries of
economic theory phenomena such as the industrial cluster that, instead of being
just a system for the efficient allocation of resources, represents an organized form
of learning in action that explores complexity and evolves as a result of this
exploration. Clusters are fuelled by complexity, which they regenerate and
increase, adding new varieties and possibilities.
So the question is whether this is still an economic process or one concerned
with history, sociology, geography and that oblique form of knowledge that has to
do wi th "totali ty".
Clearly, regional economies regained importance because the territory, with its
economic, social and cultural complexity, once again became a resource for the
creation of value and competitive advantages (Sabel 1989).
It was precisely the circular nature of the process that creates the cluster as a
total system that encouraged the study of the different industrial clusters with their
array of relationships and histories. At the same time, this circularity was also the
reason for the resistance offered by orthodox economists. The circular process of
self-production (we could, perhaps, call it autopoiesis) by the cluster cannot be
limited to the economic sphere, since the process does not apply just to machines
and professional skills, but also to a social context, institutional rules and the
values and attitudes that are inside people's heads. Self-production is an issue that
is viewed with suspicion not only because, by virtue of its circular nature, it
escapes static optimisation, but also because society and the people who populate
it can no longer be measured with given exogenous variables, becoming instead
both actors in and products of the economic process.
The main appeal of the industrial cluster is not its differential efficiency nor its
competitive success, but the fact that the industrial cluster brings to the fore the
nexus through which "that which is economic-productive and that which is
44 E. Rullani

sociocultural fuel and condition each other's development" (Becattini 2000c, p.

16), placing at the centre of economic analysis the people that create enterprises
and those who work at those enterprises. These people, even when they seem to be
following a merely efficientistic and instrumentalistic logic, have, in fact,
intertwined their life goals inextricably with the work they perform at their
enterprises, thereby creating a type of economy that may seem perfectly normal at
a given location, but could not exist anywhere else (pp. 19-21). An industrial
cluster is a local system with characteristics that are different from those of the
global economy: even though it creates a steady exchange of goods and means of
production with the rest of the world, a cluster reflects its borders and local
identity when it has to regenerate the conditions that underpin its existence as a
complex system, correcting and reinventing the balance of relationships it
inherited from the past (Becattini 2000b, p. 107). The problem with reproducing a
given cluster at another location stems from the fact that what makes up a cluster
is not just institutions, contractual rules, factories or material infrastructures, but
also people of flesh and bones, with their anthropology, way of life and
interactions. The Prato experience cannot be exported because the people from
Prato do not exist anywhere else, and it would be hard to create a cluster without
its main raw material (Becattini 2000a).
Sebastiano Brusco (1982, 1989) agrees that an in9ustrial cluster is a place wher<!
the causes of manufacturing efficiency, social balance and knowledge creation
coexist in a dynamic equilibrium. A cluster must reproduce not just the efficiency
model that impacts the costs and revenues of its businesses, but also the
"environmental" conditions that allow the model to function at a given location
but not elsewhere. The environment includes institutions, social culture. fiduciary
relationships between people, networks of family members and friends, and local
identities and ethics - in other words, all that the cluster has generated during its
history to keep in balance the delicate relationships that exist between economic
and social forces, individual and collective needs, the past and the present.
Within industrial clusters, history proceeds with the rhythms and ways of
evolutionary learning, including breaks that occur when one of the original points
of equilibrium vanishes or there is a sudden change in the competitive
environment to which a cluster is linked. Local systems, when they are able to find
sufficient energy to take up the challenge, answer these developments by
cumulating a thousand small steps, a thousand minuscule explorations that, when
combined, end up producing a visible change. It is the so-called "development
without fractures" (Fua and Zacchia 1983), which does not mean without breaks,
dramas and big or small revolutions, magnified by the narrow confines of a local
environment.
The works published by Becattini, Brusco and Fua paved the way for
innumerable theoretical and empirical essays that viewed industrial clusters not
just as examples of local economies, but as manifestations of a complex and self-
organizing system that provides a "key to understanding capitalism" and opens up
for exploration certain aspects of reality that had been closed to conventional
theory (Becattini 2000c, p. 16).
 The Industrial Cluster as a Complex Adaptive System 45

We thus discover that there is a "Third Ital y" (Bagnasco 1977) studded with
industrial clusters; that, in recent years, clusters are starting to appear in Southern
Italy as well (Viesti 2000a, 2000b); that a significant portion of the "Made in
Italy" products that give strength to the Italian balance of trade are manufactured
in industrial clusters (Fortis 1996, 1998); and that all industrial clusters are
different and must be studied individually. This realization gave rise to a new
branch of economic literature consisting of an increasingly large number of
empirical studies (I will not even attempt to produce a list of Italian cluster "cases"
because it would be too long and would inevitably be incomplete). By becoming
an "ideal type", in Weber's sense, the cluster economy is transmogrified into a
flexible specialization - an alternative to mass production (Piore and Sabel 1984) -
thereby achieving a level of international recognition that was underscored by the
success of Silicon Valley in the United States (Saxenian 1994, Zagnoli 1991). This
evolution has progressed to such an extent that recent broad-based studies have
focused on comparing the performance of economies organized into industrial
clusters with other economic forms (Signorini 2000).
In recent years, the issues surrounding industrial clusters have received
international attention not just because clusters represent a type of organization
that is attractive in all those cases where the goal is to trigger the proliferation of
neo-entrepreneurship, but also because they can help explain the localized
formations created by economic development. With the theory of clustering
(Porter 1989) and the new "economic geography" of Krugman and Venables
(1990, 1995), economic theory took a closer look at the territory, with the goal of
assimilating the complex geometry of its forms and histories. Even though these
formal approaches contained an echo of the old economy of localization that paid
scant attention to the anthropological and social quality of locations 1, a connection
between the Italian theory of clusters and the new economy of the territory that
has gained acceptance on an international scale was firmly established.

5. Meta-theory of Clusters: Self-organized Complexity

The distinguishing feature of industrial clusters is their extraordinary ability to

allow an efficient organization of production, with the full range of specializations
and exchanges that are typical of a modern economy, to coexist with high levels of
variety (in space), variability (in time) and indeterminateness (in the realm of
possibilities). In other words, clusters represent the practical refutation of the
"central dogma" of modernity, which postulates the incompatibility of
organization and complexity and opts for organizations that are doomed to
maintain at an artificially low level the complexity that can exist within them and

IOn this issue see the discussion that appeared in Economia e Politica Industriale,
(2000), Nos. 106 and 107, following the publication of an article by Viesti
(2000a).
46 E. Rullani

in their transactions with the outside world, the only alternative being a situation
devoid of organization in which complexity increases freely and chaotically.
Industrial clusters offer a practical example of how to achieve a reasonable
synthesis of organization and complexity. Their market success is proof positive
that a good level of organizational efficiency can be achieved even in the absence
of artificial systems that use techniques, command and negotiations to reduce
complexity. In other words, clusters deal with complexity in another way (without
reducing it beforehand) and with other systems (different from techniques,
command and negotiations).
What are these systems?
Industrial clusters metabolise complexity through a dynamic process of
distributed learning that mobilizes the intelligence of many separate players who
communicate and interact with each other (Micelli 2000). None of these players
has sufficient technical, regulatory or negotiating power to force the others to
behave in accordance with a common action framework. Instead, each player
learns in the course of action how to explore his/her segment of complexity,
accumulating personal experiences and at the same time benefiting from the
experience of others. The result is a circuit of self-organization (Silverberger, Dosi
and Orsenigo 1988) that slowly, through repeated attempts, "learns" how to
develop a system with an efficient internal organization that can respond flexibly
to the demands of the outside world.
The organizational tools that can be used to manage complexity are those that
allow:
• the autonomous and intelligent exploration of a fraction of complexity
(small business);
• the taking of personal risk with respect to the results of the exploration
(entrepreneurial spirit);
• the ability to learn from the experience of others through the circuits of
communication and local emulation (imitation of competitors, integration of
customers and suppliers, switching of work from one company to another,
spin-offs of new businesses, etc.).
The behaviour of a cluster as a whole is the product of the combination of a
series of partial organizational choices that let each enterprise find its position vis-
a-vis the other enterprises in accordance with classic swarm dynamics (Kelly
1994, 1998). The sum of these semi-independent behaviours of limited rationality
produces an orderly result that, in many cases, is more efficient than the result that
could have been obtained through centralized planning.
From this standpoint, industrial clusters have all of the characteristics typical of
a complex adaptive system, or CAS (Axelrod and Cohen 1999), because:
i. they are ruled by self-organizing processes that occur as a result of changes,
selections and retentions of the variables that are best suited to each
contingent situation and mayor may not be the product of an intentional
design and the work of conscious players;
ii. in the development of alternatives and the selection and retention of these
alternatives, clusters use an intellectual and relational capital sedimented
 The Industrial Cluster as a Complex Adaptive System 47

through experience and assimilated in the specific and original structures of

each cluster (languages, logistics circuits, fiduciary and guarantee systems);
iii. they are characterized by non linear processes of growth and dynamics, due
to the multipliers that disseminate throughout the clusters the most effective
solutions and block those that are less successful;
iv. they harbour a wealth of environmental variances and niches that is
significantly greater than is required to meet current needs, thereby creating
a reservoir of diversity that permits flexible and creative responses to all
contingencies.

6. Rediscovering Complexity

As I have already stated, after years of neglect, the physical and social sciences
rediscovered complexity. In this field of learning, as in many others, there is a
tendency to generalize the work of a few pioneers who laid the foundations for
future advances.
The theory of complexity has put forth certain very general formal principles
and noted the existence of complex adaptive ~ystems in many different fields, from
physics to biology, from social evolution to economics (Axelrod and Cohen 1999,
p. 16).
The key element that justifies accepting the very notion of complexity in the
physical, biological and human-social world is the idea that the formal structure of
complexity has certain common characteristics that make it possible to frame it
within a general, transdisciplinary context.
There are many definitions of complexity, and this is not the place to review
them. For our purposes, we can say that a field of variance that exists within a
space of possibilities can be deemed to be complex if:
- the space of possibilities in question is broad and undetermined, and
- the space of possibilities is not flat or just plain chaotic; it is curved, due to
the impact of attractors, breaks, slopes and the intertwining of different
phenomena that facilitate or hinder learning.
In a complex situation, the space of possibilities can be described as an
extremely rugged and non-linear landscape, where small causes can have big
effects or, quite to the contrary, major forces can implode without any apparent
consequences, and as an environment in which there are levers that can trigger
chain reactions with unpredictable effects at remote spots on the playing field.
Consequently, any attempt to explore complexity will have to deal with the
effects of magnification (positive feedback circuits) or implosion (negative
feedback circuits), with rising and descending paths and with catastrophic events
that render the progress made irreversible.
In order to identify useful or efficient solutions, any exploratory work must take
advantage of favourable curvatures while resisting unfavourable forces. This
approach entails complex forms of learning that can help overcome the most
48 E. Rullani

obvious attractors (local optimum points that block the path to remote optimum
points) and the ability to guess the positions of the most favourable possibilities.
The general idea of complexity is shaped by the varied and undetermined nature
of the space of possibilities that is being explored and by the "curvature" of that
space. In a flat or perfectly casualistic space, the exploration can proceed
stochastically on the basis of the degree of probability assigned to the different
possible events.
The general model for the exploration of the curved space of possibilities is
derived from the evolutionary paradigm and "tested" through computer
simulations (Dosi and Marengo 1993, Axelrod and Cohen 1999). It includes a
sequence of variation, selection and differential reproduction for the variables that
are found to be suitable (having a minimum degree of adaptability) and allows a
large number of different modalities in the execution of these activities. The use of
this form of learning - which is not the only one, but is the one that is more
general in scope because it does not presuppose any previous knowledge of the
space of possibilities that is being explored - is important because adaptive
learning alters the frequency and structure of the chosen systems. In other words,
the complexity of the environment becomes "imprinted" in the internal structure
of the organism or organization, thereby ceasing to be an exogenous variable that
is fundamentally unpredictable and uncontrollable. Evolutionary learning creates a
link, a structural coupling of the (adapted) internal structure and the possibilities
that exist in the external environment. The complexity of the environment is thus
metabolised by the system as it learns and evolves into an organization, with an
implicit match between what is internal (complex system) and what is external
(environmental complexity).
We can then speak of complexity with respect to the environment (latent form of
the space of possibilities) or the system that explores it (its internal structure that
metabolises the complexity of the external environment and makes it
manageable). These two definitions are different but convergent: we expect the
exploration of a complex environment to produce the construction of a structure of
equal complexity within the system, and the internal complexity of a system to be
useful only if it refers to an environment that is in some way related to the
structure of the system.
The learning process removes free complexity from the environment and
metabolises it into the structure of the system as governed complexity, i.e., as
variance, indeterminateness and curvature of the space of possibilities that the
system can control with variances or adjustments that are quick and inexpensive
but not deterministic (Di Bernardo and Rullani 1990).
The internal structure is organized by levels, based on the emerging properties
that, for example, differentiate a macro-system from its elemental components.
The discovery of emerging properties that, for example, are typical of cells and
not of molecules, of social groups and not of the individuals within the groups,
permits the allocations of the governance of complexity among different levels of
the organization, from micro to macro. The governance of complexity can be
achieved by regulating the behaviour of the elemental components. However, the
control of all those aspects that concern properties that emerge at the macro level
 The Industrial Cluster as a Complex Adaptive System 49

can become very difficult and expensive unless governance mechanisms based on
the principle of functional equivalence are established at the macro level.
Elemental components that perform the same function, because they do not alter
properties that emerge at the macro level, are deemed to be equivalent and,
therefore, interchangeable. Control and feedback mechanisms that correct any
variances, regulations and standards that guarantee a high level of process and
product consistency, and shared meanings that permit the joint testing of virtual
possibilities ensure that all of the parties involved are functionally equivalent,
thereby preserving the emerging properties of the macro-system. As a result, the
macro level becomes largely independent of the fate of its elemental components,
which can be easily exchanged or replaced with others, when needed.
Complex systems cannot be reduced to the sums of their component parts
because the ability to maintain the emerging properties depends more on the
services that guarantee functional equivalences than on the behaviour of individual
elemental components.

7. Different Forms of Complexity

Clearly, beyond this general outline, there are profound differences between the
different levels of organization and the different fields in which complexity
occurs. A different !Jystem of variation/selection/differential reproduction is
adopted for every level or field.
Because of this, the complexity of economics is different from the complexity
of, for example, physical or biological sciences. This occurs because:
- the mechanisms of variation that explore what is new are specific
(deliberateness, innovation);
- the mechanisms of selection are tied to the internal choices of businesses and
the competitiveness of the marketplace;
- the mechanisms for the expanded reproduction of the "winning" or "correct"
variances are tied not only to the competitive success of the winning
companies, but also to processes involving contractual transfers (licensing,
franchising), reputation (brand, image), copying, imitation, communication
and learning.
Therefore, even in the area of complexity, there is an economic specificity that
cannot be overlooked. On the economic plane, the cluster has its own specificity
as a special complex adaptive system.
In the field of economic studies, the theory of complexity has been applied to
the solution of specific problems, almost all of them involving the "invisible
hand" of the market and the changes that occur in response to attempts at finding
equilibrium between prices and quantities.
The market can be considered an example of CAS and was treated as such by a
tradition of research that viewed it as a cognitive mechanism (Hayek 1937;
Rizzello 1997).
50 E. Rullani

The marketplace represents the summation of the behaviours of many

independent operators who try to maximize their profits. As they move and line up
on the sides of demand or supply they help identify the unknown variable: the
price that can level demand and supply simultaneously in all markets.
A special field for implementing the ideas derived from the theory of
complexity is that of the financial markets, which often display dynamics of
research that are the summations of independent valuations by different operators
who supply and demand funds. According to the theory of repeated games, the
rules of the game that develop among the players incorporate the links of
reciprocity and the expectations that are necessary for the proper operation of the
markets by metabolising the complexity that develops on each separate occasion.
Another field suitable for the use of complex adaptive systems is that of the
transformation of the rigid forecasting and planning systems used by large Fordist
companies, which have become inefficient due to the inescapable complexity of
the environments in which they have to operate. For example, in the organization
of distributed logistics, when setting up appropriate systems to respond to market
changes, manage the production cycle and handle all problems that exist in a
nonreducible space of variance and indetermination, experimental learning paths
that adjust choices to the results achieved can prove more effective than
conventional systems, which are based on a predetermined representation of
reality and unchangeable selection procedures.
The superiority of adaptive solutions, which do not require drastic measures to
reduce complexity, can be tested with computer simulations that show the greater
efficiency of on-the-job adaptation compared with planning systems based on
rigid forecasts or commands that cannot learn from experience.
In all of these cases, a circuit of learning that proceeds by trial-and-error uses
repeated experiences to generate a structure that can respond to complexity with
research procedures that are adaptive without being deterministic. In such an
environment, market operators or nodes in a logistics network can transform the
limited intelligence of each operator into a form of collective intelligence that
represents the summation of a plurality of experiences and develops solutions that
can ensure a reasonably high level of efficiency within a given range of variance.

8. Peculiarities of Clusters as Complex Adaptive

Systems

Compared with other examples, industrial clusters present certain peculiarities

that make them especially interesting as a means of organizing and managing
complexity. In particular:
1) Industrial clusters are natural self-organizing systems. They are not technical
or computational devices designed by someone to use self-organization as a
problem-solving tool, but are themselves products of self-organization. As
opposed to what occurs in the market or in large organizations during
periods of change, the rules of the game by which clusters govern
 The Industrial Cluster as a Complex Adaptive System 5I

complexity are not dictated "from above". On the contrary, they form
spontaneously over time without any intervention. In addition, these rules
are updated through a dynamic process of collective learning that follows the
swarm model rather than being preordained or mandated by a higher order-
making power.
2) Clusters are localized ecosystems of relationships created through history
and organized in accordance with an internal logic that is largely
inaccessible to observation. Much like neural networks, clusters are black
boxes that metabolise experience and develop specific integrations of
resources and relationships that are unique and nonreproducible because
they escape logical representation, even though they exist in the different
visions of the individual players. In the same way that a neural network
modifies the weight and form of the relationships between the different
nodes, cluster ecology continuously modifies the structural coupling
established between individuals and the existing meanings, physical assets
and products. In their activity, clusters do not just "process" information and
technical solutions; they develop social structures, collective identities and
value scales, and, ultimately, shape anthropologies and mindsets that exist
only at one location and that are virtually impossible to re-create elsewhere,
even when very similar learning paths are followed.
3) Industrial clusters are living systems, not just more or less efficient problem-
solving devices. Or, rather, clusters acquire a problem-solving ability only
when they can reproduce themselves as living systems consisting of
thousands of people who use these cluster environments to develop not just
efficient and profitable solutions, but also answers to their need to interact,
communicate and feel, which goes beyond a narrow economic vision. All of
these people live permanently within each cluster and lack an alternative
world where they can satisfy their noneconomic needs. Therefore, cluster
micro societies cannot be efficient only from an economic standpoint; they
must also meet meta-economic and extra-economic needs that are equally
important to ensure the system's ability to reproduce over time.
Because they are natural self-organizing systems, localized ecosystems, living
systems that are required to reproduce society in its entirety and not just an
instrumental portion of it (the economy), clusters do not fit easily into the classic
categories of economic science. For example, their anthropology is broader, by
definition, than the one applicable to homo oeconomicus and their rationality is
always more than just instrumental. The ties that link clusters to the history,
culture and mind set of the people who populate them are never contingent or
superficial, binding instead deep structures and distinctive differences. A cluster is
not just a logical category; it is an organism made of past experiences, collective
identities and relationships based on mutual recognition. When studying cluster
dynamics, there is a limit to which extent the evidence grid can be dissected
analytically, because if the dissection is not done with sufficient care, one runs the
risk of separating that which is inseparable and loosing sight of the fact that
clusters function like systems of integrated parts that interact with each other and
52 E. Rullani

not like complexes of isolated elements that are joined together strictly for
contingent reasons.
Clusters and the economy belong to two different regions of knowledge: the
former belong by full right to the region of complex phenomena and systems; the
latter belongs methodologically to the region of computing and determinism.
Consequently, clusters can be rightfully included in the realm of economics only if
the field of interest of economics is enlarged, leaving determinism behind and
entering the open sea of complexity.
It is only by taking this step, which many economists still refuse to do, that
clusters become meaningful examples, serving as prototypes and living
laboratories for systems in which society metabolises complexity, using it as a
productive resource capable of creating economic value and competitive
advantages. In a certain sense, clusters are in vitro displays of the learning system
of an economy that has discovered the most effective way of moulding people,
economic developments and social organizations for the purpose of harnessing
complexity (Axelrod and Cohen 1999).

9. Clusters as Cognitive Systems

The localized nature of learning, and of the resulting knowledge, is not the product
of a historic accident or a geographic anomaly. On the contrary, it derives from the
very way in which knowledge is produced, each piece of knowledge being
generated in a context that includes a web of personal, business, local and industry
relationships, each of which can be essential in order to obtain the desired effect.
Consequently, each piece of knowledge retains its validity only as long as its
context remains the same, losing validity (and value) as the conditions of the
original context begin to change. Obviously, there are certain activities -
abstraction, codification, standardization, measuring - that serve the specific
purpose of extending the validity of knowledge beyond the original context to a
much broader field. However, it is also true that these activities involve only a
portion of the knowledge acquired through experience. The greatest portion of
knowledge remains embedded and sedimented in personal abilities, business
routines and territorial contexts (Cora 1998).
People, businesses and territories know more than they can make explicit or
transfer to others. More importantly, they know how to do more than they can
explain because they use the embedded knowledge that is sedimented in the
contexts. In complex situations, where there are many and undetermined
variances, the processes of abstraction, codification, standardization and
measurement are too slow and costly to produce significant quantities of this
implied knowledge and make it transferable from one context to another. When
complexity is high, conditions change so rapidly that they require a <.:Ontinuous
process of interpreting and modifying the existing knowledge. As a result, there is
not enough time or potential benefit to justify extracting from the existing contexts
a transferable knowledge that can become obsolete in a short time. It becomes
 The Industrial Cluster as a Complex Adaptive System 53

more effective to mobilize people, businesses and territories in a process of

invention and productive organization, because they are bearers of a practical
knowledge sedimented into experience (Becattini and Rullani 1993).
In this way, complexity develops an organic relatiunship with industrial clusters:
thanks to the sharing of the context of experience, which involves many people,
businesses, social participants and institutions gravitating around a single
productive hub (furniture, footwear, tiles, household appliances, etc.), clusters use
a type of organization that is particularly suited to mobilizing the embedded
intelligence of many independent actors for the purpose of accomplishing a
difficult but rewarding task: making governable the complexity of modern
industrial procuction, thereby transforming a problem - the problem of
unpredictability, uncertainty and risk - into a valuable competitive resource.
This is true in part for any form of territorial organization, but in industrial
clusters the sharing of context between people, businesses, social participants and
institutions is particularly meaningful because at the cluster level, everyone, or
nearly everyone, can draw from experience gained from life, work and
relationships within the same industrial sector. There is a centre around which the
different learning curves coalesce and become integrated. The sharing of the same
territorial context is also the sharing of ideas, experiments, expertise and
innovations, and this sharing makes it easy to integrate individual fractions of
experience quickly, because they derive from a common matrix.

10. Will Clusters Survive the Increase in Complexity that

Is Bound to Occur in the Near Future?

For the foreseeable future, the economy will continue to display a growing level of
complexity, if for no other reason than that market globalisation and
communications networking will inevitably multiply connections and inter:.lctions,
thereby increasing significantly the variety of nodes and the indeterminateness of
their behaviour. Any research into the type of organization that will be best suited
to handle this growing complexity must take into account the cluster - i.e., the
localized specialization of a nucleus of autonomous businesses that interact with
each other - as one of the systems capable of providing an effective answer
(Grandinetti and Rullani 1996, Vacca 1997).
In this endeavour, the cluster will have to compete with other solutions,
including three alternative systems that must be reckoned with in assessing how
competitive the cluster system will be (Rullani 2000):
The market, i.e., a decontextualized relationship among businesses that does not
presuppose the existence of a context of shared experience and that, therefore, can
apply to exchanges and dissemination of the highest possible multipliers.
Hierarchy, i.e., the development of forms of shared ownership of knowledge
and skills, organized around the common membership in a single corporate
organization.
54 E. Rullani

Networks, which are based on interaction and the sharing of skills by different
businesses that develop away from each other, at different locations, with the help
of modern communication technologies.
There is one specific feature that distinguishes the cluster from these three
systems: in a cluster, the communication and interaction medium linking all
parties involved (customers, the full range of suppliers, entrepreneurs, employees,
banks, local institutions, etc.) is in fact the territory and whatever became
sedimented in the territory through history.
In the market, it is price that provides an interface between the parties. In a
hierarchy, the links between the various companies are provided by
communications, logistics and assurance systems that must be built and
maintained through a collaborative effort that entails sharing investments, risks
and, if nothing else, beneficial behaviours.
Obviously, these are very different relationships and exchange media.
In a competitive comparison with other systems, the cluster is aided by the
advantages provided by territorial contiguity. These advantages are not
insignificant, considering that most of our complexity-governing skills are the
product of localized learning dynamics that are strongly influenced by the context
in which the learning takes place.
In successful clusters that have steadily increased sales to their terminal
markets, territorial continuity has produced a marked division of labour amung
different businesses characterized by:
Mutual speCialization, which allows companies to focus their investments and
risks on a narrowly defined core business, relying on other local specialists
(subcontractors, components manufacturers, providers of specialized services and
production, logistics suppliers, distributors, consultants, etc.) for everything else.
Weaker entry barriers, because new companies can be established easily and
enter the market by specializing in very narrow fields of expertise, thereby
minimizing capital requirements and risk, and relying on the skills, manufacturing
capacity and services of outsiders that are available locally.
The continuous birth of new companies, the creation of a pool of specific
professional skiffs (related to a specific industry) that are both highly developed
and plentiful, the organization of flows of exchange among specialists and the
resulting interdependencies.
The growth of a cluster is fuelled by the innovations made by its enterprises,
provided that these innovations are at the same time inclusive internally (open to
other workers and entrepreneurs within the cluster) and exclusive externally.
Proprietary innovations developed by large companies that invest in assets that
cannot be "socialized" within a cluster (for example, proprietary brands,
commercial chains, etc.) fuel the growth of individual companies and not of local
systems.
Codified innovations, which can be incorporated into easily transferable
machines or technologies, can provide only a minimal contribution to the growth
of a cluster. Innovations of this type can be quickly disseminated outside a cluster
because they are not exclusive externally: after a short while, they become
accessible to competitors located outside the cluster.
 The Industrial Cluster as a Complex Adaptive System 55

11. Inclusion and Exclusion: A Battle on Two Fronts

The innovations that are best suited to producing an increase in the knowledge that
becomes sedimented within a cluster and fuels the growth of the businesses that
use this knowledge are those that have to do with products and production
processes. A cluster works best when the task at hand is identifying niche needs,
meeting latent consumer requirements and bringing innovations to the production
process incrementally through on-the-job field tests. In all of these cases,
innovations are based on an unspoken knowledge that is not easily transferred
outside the cluster. As a result, these innovations have the characteristics typical of
localized innovations: they are inclusive internally (they can be easily imitated)
and exclusive externally (they cut out competitors located outside the cluster).
The growth of a cluster occurs along the often-narrow divider that separates
localized innovations from proprietary innovations (non inclusive) and codified
innovations (non exclusive). In order for a cluster to replicate its competitive
position and protect the quantity and quality of its product supply from
competitors, the flows of innovations that occur within the cluster at a given
moment in time must be sufficient to maintain the balance of this twin frontier -
both internal and external. When innovations serve to disseminate good ideas at
the local level while at the same time excluding potential external competitors, a
cluster expands. If, on the other hand, there are obstacles that hinder internal
propagation and external defence, a cluster contracts, and if it shrinks below a
certain critical threshold, it can even implode to the point of disappearing as an
organized system of learning and localized production.
Therefore, the future of industrial clusters will be predicated on a critical factor:
the manner in which they handle the growing complexity they will be faced with,
i.e., the innovations that will emerge in the economy in the near future (Arora and
Gambardella 1993; Arora, Gambardella and Rullani 1997). In the case of codified
innovations, the division of labour could use the market to bypass the cluster. In
proprietary innovations, the division of labour within a major enterprise could be
more effective than the one occurring externally, within the cluster. Under these
new conditions, the survival of clusters is endangered simultaneously on two
fronts: clusters risk losing business activity to the market, on the one hand, and to
delocalised or multilocalized large companies, on the other.
Nevertheless, in the specific field of innovation, clusters continue to enjoy
important competitive advantages in the division of innovative labour in
noncommoditized fields that resist codification, but can, at the same time, be
imitated, transferred and reproduced at the cluster level. For example, the work of
exploring what is new, which will ultimately result in new products and new
processes, is often carried out in ways that permit good permeability at the local
level, but provides a relatively solid defensive barrier against external imitators.
Clearly, this type of work can be carried out by cohesive local networks that
include many independent operators capable of working together. However, it is
also important to keep in mind that companies are required to make certain types
56 E. Rullani

of investments that, for a number of reasons, pull in opposite directions. For

example:
a) Investments in technology, mechanization and process standardization tend
to expand the market for commoditized products that will be sold in the
market at low cost and in large quantities. Here, too, it will be unlikely that
clusters will playa major role, considering that the resulting innovations do
not exist exclusively in the local domain and can easily be transmitted to
external competitors, who will tend to gravitate toward locations with low
labour costs.
b) Investments made in marketing to expand downstream along the value chain
tend to fuel the growth of proprietary transnational systems that are
controlled directly or gravitate strongly around a leader company. The
produced innovations increase the division of labour within a single
enterprise (large company) and can easily be duplicated or imitated within
the cl uster.
In the first case (commoditizing), there will be an expansion of the global
markets that will result in increasingly impersonal transactions between producers
and users, who may be separated by great distances. The share of the market
segments held today by industrial clusters will tend to contract, paving the way for
distribution systems that operate from multiple locations within the global
economy that are occasionally linked for trading purposes. From this perspective,
clusters will survive if new complexity can develop on the cusp between
manufacturing innovation and consumption, as those areas that are better known
and controllable become standardized and commoditized. When complexity
decreases in certain areas of production and consumption, it can be recreated in
other areas if the quest for innovation drives producers and consumers toward
exploring new possibilities. There are good reasons to believe that the resources
saved through commoditization will be used by producers and consumers to
explore new forms of production and new forms of consumption. This will allow
clusters to survive the pressure of commoditization, provided they position
themselves decisively on the cutting edge of innovation, where complexity is re-
created every time a portion of it is neutralized and absorbed by the market.
In the second case mentioned above (proprietary investments), large
multinational companies that set up manufacturing systems in different countries
generally prevail over other types of economic organizations. This occurs because
they are better prepared to assume investment risks, plan interdependencies and
link production cycles that are carried out at different, distant locations. Large
post-Fordist companies have gained experience in managing complexity through
their ability to forecast, invest and take risks. Clusters will be able to continue to
perform these functions competitively only if local societies are able to:
a) Invest in reproducing and expanding the social (intellectual and relational)
capital required to tackle increasing levels of complexity. Today, this is one
of the main shortcomings of the cluster system, which until now has relied on
spontaneous learning processes to make up for the lack of investments and
limited propensity to assume collective responsibility. To compete with large
multinational companies in complex innovations, clusters must learn to rely
 The Industrial Cluster as a Complex Adaptive System 57

less on the spontaneity and individualism of the past, choosing instead to

leverage their ability to share projects, risks and identities.
b) Maintain a greater intellectual and relational vitality than the one produced
by the direct investments of large enterprises. In this area, a critical
contribution will come from those knowledge, trust and communication
resources that in rich and vital local societies exceed basic needs and help
create a wealth of variety and social intelligence that can be used to perform
new tasks and explore unusual possibilities. Clusters can compete with large
multinational companies if, for example, local societies become more open to
the multicultural environments that have become prevalent in the global
economy, more effective in interpreting what is different and surprising, and
more tolerant of transgressions, anomalies and errors.
In the economies of the near future, the ability to manage complexity will be the
decisive factor in the survival of industrial clusters. Clusters will not survive on
the strength of "external economies" or on the magical effect of economies
produced by "externalities" and "rising returns". They will survive if they are able
to field their key resource: an intelligent local society, open to innovative
experimentation and willing to take risks and assume collective responsibilities.

12. Clusters and Networks: Competition and

Complementarity

As [ have explained before, clusters manage the complexity inherent in the

organization of innovative labour using as an "aggregant" the sharing of the same
local context by companies that work alongside each other and belong to the same
experience system. In this sense, clusters are nothing more than localized
networks that use the territory as a relationship medium to facilitate the division of
labour and the dissemination of successful ideas.
On the other hand, trans local networks, which cannut use the territory as a
relationship medium, must build their own relationship media, investing resources
in communication systems, logistics channels and assurance systems to link a
certain number of companies that, in this way, are able to work together and
overcome the distance handicap. The new information and communication
technologies available today can play a decisive role in making remote
relationships quite rich and helping translocal networks become competitive with
those that develop at the local level and use the territory as an aggregant.
Building a communications, logistics and assurance network that links distant
companies requires a significant amount of resources and time because it entails
designing and operating an artificial system. This is not the case for industrial
clusters that use a "natural platform" and have developed spontaneously, without
special investments and deliberate efforts.
Nevertheless, a network has many advantages over spontaneous (local) systems
for the division of cognitive work:
58 E. Rullani

A network can use a pre-existIng system of skills and specializations,

combining the types of knowledge and relationships that are required for each
situation without the handicap of physical contiguity that can significantly restrict
the range of available possibilities.
A network can control what companies are added to or removed from an
organized context of relationships and exchanges, setting forth explicit and
enforceable conditions for the sharing of knowledge and the use of common
resources.
Networks can proliferate in localized form whenever physical contiguity proves
to be useful or can expand into trans local systems (the limit of which is the global
system) whenever it becomes necessary to capture useful varieties and differences
that are distributed on a vast scale.
Clusters and networks are two ways of managing complexity through the
interaction of different companies. To succeed they must overcome the
competition of multinational and/or market hierarchies. They can do this when the
task at hand is organizing complex solutions that, with the help of contextual
knowledge (of the location or of the network), can be disseminated within the
system and, at the same time, protected/rom external competitors.
Both clusters and networks must defend their competitive positions on two
fronts: from commoditization (market) and from proprietary control (hierarchy).
In view of the fact that the innovations that are prevalent today are already moving
in these two directions, the result of which is the codification of the production
knowledge (which favours the markets) and the creation of a need for proprietary
investments (which favours hierarchies), it is easy to see how, over the coming
years, clusters and networks will have to reposition themselves away from today's
production and systems organization.
Competition between clusters and networks will also increase. The competitive
space gained at the expense of the market and the hierarchy will provide a venue
for comparing the effectiveness of a managed sharing of the territory (physical
contiguity) with a managed sharing of remote interaction (virtual contiguity). In
the first instance, the act of sharing is a by-product of contiguity and can even be
involuntary (when an entrepreneur copies other entrepreneurs and is copied by
them). In the second instance, sharing occurs only when it is planned and
supported by adequate investments. Up to a certain point, a cluster can survive
even without an organized system. A network cannot.
However, as I have said before, a trans local network does have the advantage of
being able to multiply the frequency and opportunities of experience by linking
different locations and cultures.
The decisive element in comparing these two systems is the possibility of
gaining experience in the sharing 0/ innovations. The shared context of a cluster
(local circuit) and the shared context of the network (remote cooperation) can be
complementary resources - for example, when a cluster establishes a series of
relationships with the delocalised nodes of a broader experience sharing network -
or competing resources, when one can replace the other.
The most probable evolution is a mutual repositioning based on the assumption
that existing clusters will use a whole host of functions available through the
 The Industrial Cluster as a Complex Adaptive System 59

trans local networks to which they are linked. At the same time, networks could
become denser at certain locations, creating quasi-clusters that exchange goods
and knowledge over a finely meshed network that comprises light and highly
mobile nodes with minimal territorial roots.

13. Conclusions

The most realistic possibility is that all four of the systems I have discussed
(global market, hierarchy of the multinational corporation, cluster-type local
network, translocal remote interaction network) will occupy different regions of
complexity, in accordance with their respective characteristics.
In order to survive, clusters must avoid becoming prisoners of the businesses in
which they have engaged thus far, embracing without reservation or delay the
changes that are starting to reshape post-Fordist complexity. For example, they
must quickly surrender to other systems those areas that are indefensible,
concentrating instead on that for which they are best suited: managing complex
solutions that can be developed with the resources of intelligence and self-
organization possessed by local societies.
We cannot tell whether clusters will account for a large or small percentage of
total production and employment, since we cannot know if they will increase in
number and importance or shrink quantitatively compared with competing
transterritorial systems (market, multinationals, remote interaction networks).
What we do know is that in order to foster the competitive vitality of the
existing clusters, we must invest in localized innovations (inclusive internally and
exclusive externally) that take advantage of the skills of interpretation, invention
and sharing that are typical of local societies.
The future of industrial clusters is tied first of all to their ability to exploit
promptly - without being excessively defensive - the strengths of the markets,
hierarchies and networks, borrowing the functions that give these systems
irresistible competitive advantages. It will also be predicated on the ability of local
companies to manage complexity better than the multipliers of global markets, the
investments of large multinational corporations and the selective communications
of trans territorial networks.
The destiny of industrial clusters and of complexity is intertwined, as is their
past. It will be up to our intelligence to discover how to continue forward on a
path already taken, giving new meaning to roots and identities sedimented in the
places we inhabit.

References

Arora, A. and Gambardella, A. (1993), Division of Labour and Inventive Activity, H. John
Heinz III School of Management, Carnegie Mellon, 93-3, January.
60 E. Rullani

Arora, A., Gambardella, A. and Rullani, E. (1997), "Division of Labour and the Locus of
Inventive Activity", Journal of Management and Governance, No. I, Autumn.
Arthur, W.B. (1988), "Self-reinforcing Mechanisms in Economics", in: The Economy as an
Evolving Complex System, Anderson, P.W., Arrow, K.1. and Pines, D. (eds), Santa Fe
Institute, Studies in the Sciences of Complexity, Redwood City, CA, Addison-Wesley.
Axelrod, R.. and Cohen, M.D. (1999), Harnessing Complexity: Organizational
Implications of a Scientific Frontier, New York, The Free Press.
Bagnasco, A. (1977), Tre Italie. Problematiche dello sviluppo italiano, Bologna, II Mulino.
Becattini, G. (1961), II concetto di industria e la teoria del valore, Torino, Boringhieri.
Becattini, G. (2000a), Il bruco e la farfalla. Prato: una storia esemplare dell'Italia dei
distretti, Firenze, Le Monnier.
Becattini, G. (2000b), Il distretto industriale. Un nuovo modo di interpretare it
cambiamento economico, Torino, Rosenberg e Sellier.
Becattini, G. (2000c), Dal distretto industriale allo sviluppo locale. Svolgimento e difesa di
un 'idea, Torino, Bollati Boringhieri.
Becattini, G. and Rullani, E. (1993), "Sistema locale e mercato glob ale", Economia e
Politica Industriale, 80, December, pp. 25-48.
Brusco, S. (1982), "The Emilian Model: Productive Disintegration and Social Integration",
Cambridge Journal of Economics, 6, 2, pp. 167-184.
Brusco, S. (1989), Piccola impresa e distretti industriali, Torino, Rosenberg e Sellier.
Brusco, S. and Paba, S. (1997), "Per una storia dei distretti industriali italiani dal secondo
dopoguerra agli anni novanta", in: Storia del capitalismo italiano dal dopoguerra a oggi,
Barca, F. (ed.), Roma, Donzelli.
Cora, G. (1998), "Distretti e sistemi di piccola impresa nella transizione", in: Il
postfordismo. Idee per il capitalismo prossimo venturo, Rullani, E. and Romano, L.
(eds), Milano, Etaslibri.
Di Bernardo, B., and Rullani, E. (1990), Il management e Ie macchine. Teoria evolutiva
dell'impresa, Bologna, II Mulino.
Dosi, G., and Marengo, L. (1993), "Some Elements of an Evolutionary Theory of
Organizational Competences", in: Evolutionary Concepts in Contemporary Economics,
England, R.W. (ed.), Ann Arbor, University of Michigan Press.
Fortis, M. (1996), Crescita economica e specializzazioni produttive, Milano, Vita e
Pensiero.
Fortis, M. (1998), II "made in Italy" - Quando stile e creativitii non sono solo moda,
Bologna, II Mulino.
Fu1i, G. and Zacchia, C. (eds) (1983), Industrializzazione senza !ratture, Bologna, II
Mulino.
Galbraith, J. K. (1967), The New Industrial State, Boston, MA, Houghton Mifflin. Italian
translation: Il nuovo stato industriale, Torino, Einaudi, 1968.
Grandinetti, R., and Rullani, E. (1996), Impresa transnazionale ed economia globale,
Roma, La Nuova Italia Scientifica.
Hayek, F.A. von (1937), "Economics and Knowledge", Economica, n.s. IV, No. 13, pp. 96-
105.
Holland, 1. (1995), Hidden Order: How Adaptation Builds Complexity, Reading, MA,
Addison-Welsey.
Holland, J. H. (1988), "The Global Economy as an Adaptive Process", in: The Economy as
an Evolving Complex System, Anderson, P.W., Arrow, K.J. and Pines, D. (eds), Santa Fe
Institute, Studies in the Science of Complexity, Redwood City, CA, Addison-Wesley,
Proceedings.
Kelly, K. (1994), Out of Control, Addison-Wt:sley, Italian translation: La nuova biologia
delle macchine, dei sistemi sociali e del mondo dell 'economia, Milano Apogeo, 1996.
 The Industrial Cluster as a Complex Adapti ve System 61

Kelly, K. (1998), New Rules for the New Economy. Ten Ways the Network Economy is
Changing Everything, Londra, Fourth Estate. Italian translation: Nuove regole per un
nuovo mondo. Un decalogo per chi vuole cavalcare Ie nuove regole e non esserne
scavalcato, Firenze, Ponte delle Grazie, 1999.
Krugman, P. (1995), Develupment, Geography and Economic Theory, Cambridge, MA,
MIT Press.
Krugman, P., and Venables, A. J. (1990), "Integration and the Competitiveness of
Peripheral Industry", in: Unity with Diversity in the European Economy, Bliss, C. and
Braga de Macedo, J. (eds), Cambridge MA, Cambridge University Press.
Krugman. P., and Venables, A. J. (1995), "Integration, Specialization and Adjustment",
European Economic Review. 40.
March, J. G. (1991), "Exploration and Exploitation in Organizational Learning",
Organization Science, vol. 2, No. I, pp. 71-87.
Micelli, S. (2000), imprese. reti. comunita, Milano, Etas.
Piore, M. J. and Sabel, C. F. (1984), The Second industrial Divide, New York, Basic
Books. Italian translation: Le due vie dello sviluppo industriale, Torino, Isedi, 1987.
Porter. M. E. (1989), The Competitive Advantage of Nations. Italian translation: fl
vantaggio competitivo delle nazioni, Milano, Mondadori, 1991.
Rizzcllo, S. (1997). L 'economia della mente, Roma, Laterza. English translation: The
Economics of the Mind, Elgar, Aldcrshot, 1999.
Rullani, E. (1996), "Complessita c informazione nella scienza economica", Pluriverso 2,
RCS Libri.
Rullani. E. (2000), "Sistemi locali e produzione di conoscenza", presentation at the OECD
Conference on Enhancing the Competitiveness of SMEs in the Global Economy:
Strategies and Policies, Bologna.
Sabel, C. F. (1989), "Flexible Specialization and the Reemergence of Regional
Economies", in: Reversing industrial Decline, Hirst, P. and Zeitlin, 1., Oxford, Berg.
Sabel, C. F. (1999), "Regionalismo sperimentale e i dilemmi della politica economica
regionale in Europa", in: Neuregionalismo. L 'economia-arcipelago, Perulli, P., Torino.
Bollati Boringhieri.
Saxenian, A. (1994), Regional Advantage: Culture and Competition in Silicon Valley and
Route 128, Cambridge, MA. Harvard University Press.
Signorini, F. L. (2000), Lo sviluppo locale. Un'indagine della Banca d'italia sui distretti
industriali, Cosenza, Meridiana Libri.
Silverberg, G., Dosi, G., and Orsenigo, L. (1988), "Innovation, Diversity and Diffusion: A
Self-Organizing Model". Economic Journal, 98, pp. 1032-1054.
Vacca, S. (1997), "Le differenze socio-culturali e istituzionali nello sviluppo delle imprese:
il ruolo del capitale immateriale e del capitale umano", Economia e Politica Industriale.
No. 94.
Varaldo. R., and Ferrucci, L. (eds) (1997), II distretto industriale tra logiche di impresa e
logiche di sistema, Milano, Angeli.
Viesti, G. (2000a), "Le strade dello sviluppo: come sono nati i distretti industriali del Made
in Italy nel Mezzogiorno", Economia e Politica industriale, No. 106, June.
Viesti, G. (2000b), Come nascono i distretti industriali, Bari. Laterza.
Zagnoli, P. (1991), I rapporti tra imprese nei settori ad alta tecnologia. II caso della
Silicon Valley, Torino, Giappichelli.
II. General Models of Industrial Cluster Dynamics
Complexity and Local Interactions:
Towards a Theory of Industrial Districts

David A. Lane l

Abstract. In this talk, I will sketch an outline of a theory of industrial districts

from a complexity perspective. The theory addresses three principle problems.
The first is ontological: what is an industrial district? I argue that a district may be
defined in terms of two kinds of structures: networks and scaffolds. At the nodes
of the networks are individual entrepreneurs/artisans and firms. The links between
the nodes consist of processes of recurring interactions. The networks carry the
competences through which the district gathers and interprets information about
products, production technologies and markets; produces and sells artefacts; and
develops concepts for new artefact functionality, new markets, and new artefact
types. The network structure undergoes constant transformation as the district
generates new products and explores new markets. The transformation processes
rely on various scaffolding structures, which may be regarded as the institutions
that provide both a meta-stable identity and the possibilities for renewal and
change to the district. I will discuss and illustrate two particularly important kinds
of scaffolding structures: interaction loci and emergent rules and roles. Then, I
will address two puzzles about districts from the viewpoint of the network-
scaffold ontology: 1) How do districts achieve coordination in their distributed
organization of production (or, as rephrased by Sebastiano Brusco, what is the
difference between Carpi and Benetton)? 2) What exactly is the role of
geographical space in the structure and functioning of an industrial district? That
is, just what kind of links in a production system's network structure need to be
constructed from physical rather than virtual interaction?
The second problem I discuss is: how do districts innovate? I begin by outlining
two perspectives on innovation that have been developed in the complexity
literature: recombination and generative processes. I then locate recombination
and generative processes in the network-scaffold framework. For recombination,
the role of spatially concentrated local interactions is fundamental, while I argue
for the importance of scaffolding structures in increasing the generative potential
of a district by organizing the processes whereby generative relations within and
beyond district boundaries may form. I then use these ideas to explain changing
innovation patterns in machines for tile manufacture.

lIn preparing this paper, I profited from discussions with Robert Maxfield, Margherita
Russo and David Stark, who kindly shared with me some of their knowledge about Silicon
Valley, Sassuolo and Silicon Alley respectively. In addition, I learned much from my
discussions over the years with my colleagues Andrea Ginzburg and Sebastiano Brusco
about the ideas presented here and their relation with district phenomenology.
66 D. A. Lane

Finally, I turn to the problem of theory development through modelling and

empirical research, and discuss two new emergent phenomena in academic district
research: the use of agent-based models to provide insight into various aspects of
district phenomenology, and the development of district-oriented ethnographic
methods.
The talk concludes with a comparison of the complexity-based approaches I have
outlined with other perspectives in the district literature.

1. Introduction

In this paper, I sketch an approach to a theory of industrial districts from a

complexity perspective. The theory would address three principal problems:
- How can we describe the organization of an industrial district?
- How can we characterize the kinds of production activities that we would
expect to benefit from a district organization?
- What kinds of processes and structures might account for the persistence or
failure of a district organization?
Proposed answers to all three of these questions are presented in the penultimate
section of the paper. The concepts in which these answers are framed are
developed in the first six sections, while the last section suggests what kinds of
empirical and modelling methods might be applied to help inform, develop and
evaluate the theory I am proposing here. In this introductory section, I provide a
road map for the rest of the paper.
What I mean by a complexity perspective is explained in section 2. I then turn to
the task of defining a district and its organization. Since I want to define a district
as a special kind of subsystem of what I call a market system, I have first to
introduce this concept. That is the task of section 3, where I distinguish market
systems from the more familiar notion of markets.
In section 4, I introduce two kinds of structures that together provide the
organization of a market: networks and scaffolds. At the nodes of a network are
agents - firms, project groups, individual human beings. The links between agents
consist of processes of recurring interactions. Networks carry competences,
through which the artefacts around which a market system is organized are
produced and sold, new attributions for artefact functionality and new designs for
artefact types and production processes are generated, and new markets
developed. As new products come into being and new markets explored, the
system's network structure is transformed. The transformation processes are
supported by various scaffolding structures, institutions that provide both a meta-
stable identity for system agents and the possibility for renewal and change for the
system itself. In my discussion of scaffolds, I will focus on two particularly
important kinds: interaction loci and emergent rules and agent roles.
Now I can define an industrial district, as a geographically concentrated
production subsystem of a market system, composed primarily of small and
medium firms, with a decentralized, non-hierarchic organization. As a market
 Complexity and Local Interactions: Towards a Theory of Industrial Districts 67

system subsystem, a district's organization can be described in terms of the

competence networks of which it is comprised and the scaffolds through which
these networks are formed and reconfigured, and the district identity is maintained
and modified.
Two interesting questions need to be addressed about the particularity of
districts in the class of all subsystems of market systems. The first has to do with
the first defining characteristic of the district: its geographic concentration.
Geographic concentration enables 2 local interactions among agents. I claim that
local interactions are essential for agents to respond when the structure of their
market system is undergoing a cascade of rapid changes, which affect both the
artefact family around which the system is organized and the network of agents
and agent relations that carries the system's economic functionality. In section 5, I
argue that effective action in situations of rapid transformation of the structure of
agent and artefact space depends upon the generative potential of an agent's
relationships with the other agents with whom it interacts. As I argue in section 6,
the generative potential of a relationship is determined by a set of characteristics
that may be strongly affected by the locality of the participating agents'
interactions. Section 7 argues that districts are one solution to the problem of
organizing effective local interactions in systems that "inherently" undergo
cascades of rapid change.
The second question concerns an important feature of a number of geographical
regions in which districts are located: the same geographical region contains more
than one district, in the sense that there are clusters of firms based in the area that
participate in different market systems. For example, both Silicon Valley and the
Washington Beltway are the homes of both an information and communication
technology cluster, as well as a biotechnology cluster, while the province of
Modena has clusters in precision machinery, textiles and tiles, among other
artefact families. How might we explain the existence of such "mega-districts". or
districts of districts?3 As I show in section 7, the concept of scaffolding structures
helps to answer this question: certain kinds of scaffolding structures provide
support services to agents that participate in a district, independently of the market
system in which the agent operates. Thus, these scaffolds can serve to promote the
creation of new firm clusters that share the services they provide with existing
clusters - and may even promote the integration of these clusters into pre-existing
market systems, as new industrial districts.

2But of course does not necessarily product the right kinds of interaction structures.
Whether these form or not depends on many factors - in particular. through the kinds of
scaffolding structures that emerge to channel and generate local interactions.
3Before taking the question too seriously. it would be a good idea to establish the empirical
validity of the proposition that districts tend to be located near other districts, rather than
having a random geographical distribution, in the appropriate sense of "random" for this
issue. I believe that districts do cluster, for the reason I provide in the text. Of course, if
they don' t, it is still the case that districts that are clustered may still share certain
scaffolding structures!
68 D. A. Lane

Characterizing district organization in terms of competence networks and

scaffolding structures may provide theoretical advantages, but it certainly
introduces empirical difficulties. Consider for example one important implication
of this idea: the firm is no longer a primary unit of analysis, but merely one type of
node in the competence networks that carry out the district's economic
functionality. In the Carpi knitwear district, a man might be a worker one year, an
entrepreneur the next, and a worker the following year, without changing in any
substantial way his productive activities. A Silicon Valley entrepreneur might
leave one company he founded, take off a year or two, then found another in a
different market system - and immediately hire the same engineering team
responsible for the success of his first company. Examples like these help us to
understand that an agent-network perspective is potentially much more powerful
in describing how districts do what they do than a firm-based perspective. The
difficulty is that most available data refer only to firms, and not to individuals,
project groups, research collaborations, strategic alliances or other organizational
forms that may be nodes in district competence networks. Moreover, the very
definition I propose for network links has a temporal dimension, as it refers to
"recurring patterns of interaction", not static relations or one-shot events. What
kinds of empirical research can shed light on competence networks and the
processes whereby they form, carry functionality, and change structure'? Similarly,
how can scaffolding structures be identified and their functions observed'l A
related issue concerns the role of modeling in developing a complexity-based
theory of districts and, in particular, examining district innovation processes. The
concluding section of this paper briefly considers two emerging answers to these
methodological problems: ethnographic methods and agent-based modeling.

2. Perspectives on Complexity

The papers in this volume by Gell-Mann and Holland start from two different
perspectives on complexity. Gell-Mann takes a theoretical perspective. The
complexity of a system, he tells us, can be measured by the length of the minimal
description of the system's regularities. Thus, if we want to address the question
of whether an industrial district is a complex system, we are forced to confront
some hard ontological issues. How do we characterize a district in system terms -
that is, what are its component entities, and through which kinds of processes do
they interact with one another'? Even harder: taking on the role of Gell-Mann's
judge, we need to decide what constitute a district's regularities - and to frame a
vocabulary in which we can describe them.
Holland's approach to complexity begins phenomenologically rather than
theoretically. For him, a complex system is one that behaves in a particular way.
Roughly, the system is composed of a number of entities that have properties that
in general differ from entity to entity; the entities interact with one another and the
environment they inhabit; as a result of these interactions, their properties, the
environment's, and even their interaction modes may change. In the examples
 Complex it} and Local Interactions: Towards a Theory of Industrial Districts 69

Holland presents, the entities' structures - that is, the set of their properties -
determine how they function - that is, the transformations they effect through their
interactions with other entities.
A complex system, then, is a particular kind of dynamical process, determined
by a succession of interactions among entities. The phenomena of interest in the
dynamics of complex systems are often characterized as emergent: they refer to
features that can be concisely described only by reference to a higher level of
aggregation than the individual entities themselves and that persist for periods of
time considerably longer than those in which individual entities' interactions are
denominated. Often, emergent structures self-organize, and this emergent
organization of the system may constrain and channel the entities' interaction
patterns. To apply the phenomenological perspective on complexity to industrial
districts, one must describe the district in terms of the agents that compose them
and their modes of interaction, then search for "global" emergent structures and
self-organization at the level of the district itself. We might even ask to what
extent the district itself might be regarded as an emergent entity.
Both Holland and Gell-Mann emphasize an aspect of complex systems that
could be described as cognitive. Holland refers to ·'adaptive agents" that adjust
their actions on the basis of what they have learned from their previous history of
interactions with their world; Gell-Mann calls the representation of the world on
the basis of which such adjustments are made a "schema", and prefers to reserve
the term "complex system" itself for an entity with a schema, which can be
thought of as its minimal description of the regularities in its world. This circle of
ideas prompts a question rather novel to the district literature: are we justified in
viewing the district itselfas what Holland calls an adaptive agent and Gell-Mann a
complex system? Or to put it another way: as we have already seen, a complexity
perspective requires that we treat a district as a time-varying structure and that we
try to understand how its structure determines its function; should we also view
the district as a cognitive entity, and understand how its cognitive processes give
rise to transformations in its structure?
Summarizing, the hallmarks of a complexity perspective include commitments
to
• process and change, not stasis and equilibrium;
• a multilevel organization of entities;
• entity function determined by entity structure;
• distributed control and information-processing; and
• emergence and self-organization.
Whether it is useful to apply a complexity perspective to industrial districts
depends on how rich an account of district phenomenology a theory founded on
these commitments can produce.
70 D. A. Lane

3. Setting the Context: Markets and Market Systems

"The market" is an abstract entity defined formally by economists and employed

informally by journalists, politicians and just about everyone else. It is a locus of
impersonal exchange activities, where agents buy and sell products with defined
characteristics, at prices that - according to standard economic theory - renect
supply-and-demand induced equilibria. Economic theory accords these prices the
role of principal communication media between agents, who use the information
prices convey to drive the actions they take in the economy. Relationships
between agents do not count for much in "the market". What matters is how each
agent separately values each artefact in "the market", values that "the market"
then aggregates into prices for these artefacts. Frequently, in popular narratives
about developments taking place in business and the economy, "the market" is
assigned the central causal role.
By a market system, I mean a set of agents that engage with one another in
recurring patterns of interaction, organized around an evolving family of artefacts.
Through their interactions, the agents produce, buy and sell, deliver, install,
commission, use and maintain artefacts in the family; generate new attributions
about functionality for these artefacts; develop new artefacts to deliver the
attributed functionality; and construct, augment and maintain new agents and
patterns of agent interaction, to ensure that all these processes continue to be
carried out, over time, even as the circumstances in which they take place are
changing in response to perturbations from inside and outside the market system
itself.
In a market system, the meaning of artefacts are up for negotiation. As a result
of these negotiations, the artefacts take on value - not just in individual heads, but
through a social process that takes place in concrete social settings. Agents learn
more from each other than they do from prices, and they do not merely exchange
information, they jointly develop interpretations. These interpretations drive action
in new and hitherto unexplored directions.
Standard economic theory starts with the concept of "the market". Of course,
something like "the market" will playa role, often an essential one, in many of the
interactions in which market system participants engage. However, in the most
interesting district phenomenology, agent relationships often hold centre stage,
and many modalities of communication, from discourse based upon shared
understandings through joint action coordinated by tacit knowledge, shape these
relationships. Moreover, district functioning relies on non-market structures - like
entrepreneur associations, user groups, trade fairs and standards organization - as
well as shared attributions about agent roles and artefact functionality and rules
that determine how agents may interact. These elements do not arise from market
transactions, but from a complex network of agent interactions, whose description
and analysis require a different set of ideas than standard economic theory has to
offer.
 Complexity and Locallnteractions: Towards a Theory of Industrial Districts 71

4. Market System Organization: Networks and Scaffolds

A market system can be viewed as a collection of transformation processes: for

example, producing, selling, installing, maintaining, designing artefacts in the
family around which the system is organized. In the course of carrying out these
processes, other processes are enacted by agents in the system - gathering and
interpreting information, setting standards, establishing new entities like strategic
alliances or trade associations. All these processes are achieved through
interactions among agents - individuals, working groups, divisions, firms,
alliances. Since these interactions taken together deliver the functionality that
permit the system to endure over time, they tend to be organized into recurring
patterns, with each pattern identifiable by its set of participants, interaction modes,
and frequency and duration of interactions. Each recurring pattern of interaction
defines a network; each network may be said to carry a system competence; as
they are enacted, these competences generate the transformation processes that
deliver the system's functionalities. Over time, of course, the transformation
processes, the competences that enact them, and the networks that carry these
competences change. But in the relatively short term, we may describe the
system's organization in agent space as the crosscutting network of these
competence networks.
If we view a market system from a somewhat longer time perspective, we noti..:e
that it is always undergoing perturbations, which may be generated from processes
taking place within the system itself or may come from outside the system, as for
example from large-scale macroeconomic shifts. In response, new networks are
constructed and others change their structure, either by replacing or adding nodes
or by altering the modes, duration or frequency of the linking interactions. Some
of these changes may seem to happen "spontaneously", but for most of them we
can identify structures that have provided the opportunity and the means for them
to happen. Thus, the fluid network organization of a market system is constructed,
renewed, and extended by a longer-lasting set of structures that serves to keep the
system functioning. We call these structure scaffolds, since they shape and support
the competence network structure of the system as it undergoes construction and
reconstruction.
Scaffolds come in two basic flavours - physical and cognitive. Examples of
physical scaffolds include user groups and trade fairs, trade and professional
organizations, and standards bodies, as well as communication media like trade
and professional journals, company and organizational newsletters, and websites.
Some of these scaffolds are generated inside the system itself, others come from
elsewhere, for example from other market systems, from professions, or from
government agencies. interaction loci are a particularly important kind of physical
scaffold. Their function is to provide a space within which particular types of
interactions may take place 4 • All the examples listed above are or provide

40f course, the space need not physically contain all the participants in the interaction.
Instead, it may be mediated by artefacts, as happens with electronic bulletin boards or
72 D. A. Lane

interaction locI. It is important to understand that all interactions are spatially

located - and the way the space in which they happen is structured can have
significant effects on the form the interactions take. Thus, interaction loci are
crucial for constructing and maintaining agent relationships within a market
system, and the kinds and organization of its interaction loci go a long way in
determining the structure of the system's agent-artefact space.
Cognitive scaffolds are a bit more difficult to define and identify than their
physical counterparts. Interacting clearly has important cognitive components:
agents intend to derive certain consequences from the actions in which they
engage; they have attributions about the identity of the agents with whom they
interact that in part determine the way in which they act and the interpretations
they make of the actions of the others, as do their attributions of the functionality
and value of the artefacts around which their actions are oriented. For a system to
function, there must be an alignment of some of these attributions and some
degree of shared expectations about the relation between intention and
consequence, so that recurring patterns of interaction can form and be maintained.
In particular, role structures emerge within a market system, by means of which
agents in the system classify one another and on the basis of which they generate
expectations for the kinds of interactions in which others may engage and for the
ways in which they may interact. Shared attributions of role structures then
channel agents into particular interaction streams, which provide some continuity
and stability to system processes. The more stable are the collective attributions of
agent roles and artefact functionality and value that emerge in a market system,
the more "rule-like" and stable are the patterns of interaction that deliver system
functionality - that is, the network structure of the system itself. Naturally, there is
a connection between cognitive and physical scaffolds: many of the processes
through which cognitive scaffolds emerge take place within system interaction
loci.

5. Generative Relationships, Generative Potential and

Complex Foresight Horizons

Attributions playa fundamental role in market system processes. Agents represent

the regions of agent-artefact space in which they operate by means of attributions.
Agents' attributions of their own and each others' identities - what the agent does
and how it does it - mediate how agents interact with one another, and their
attributions of an artefact's functionality mediate the ways in which they make and
use the artefact. We have seen how the emergence of ~hared attributions underlies
the formation of stable interaction patterns without which a market system cannot
endure for long. On the other hand, we have also seen that market systems change,

video-conferencing. As these two examples indicate, however, even such "virtual spaces"
can be identitied with physical spaces that determine the form of the interactions they
support.
 Complexity and Local Interactions: Towards a Theory of Industrial Districts 73

as new agent structures form and new artefact types are developed. In general,
new entities are preceded by new attributions. For example, a new artefact type is
developed in order to achieve a new kind of imagined functionality. So to
understand how a market system changes, it is important to consider how new
attributions come into being.
Robert Maxfield and I (Maxfield and Lane 1997) have argued elsewhere that
new attributions generally arise in the context of generative relationships among
agents. The key idea is that individual agents' attributions tend to be organized
into closed systems, within which it may be possible to recombine existing
elements to develop "better, faster, cheaper" versions of previously represented
types, but which resist extension into new classificatory dimensions. To open an
attributional system to new conceptualisations usually requires the realization that
its current representations are incomplete, a realization that can generally emerge
only in the context of discursive relationships with others, whose attributional
systems highlight different aspects of the identity or functionality of some
particular entity under discussion. Often, the realization begins with the
uncovering of an ambiguity in the sense in which different participants in the
discussion use the same terms. This ambiguity in categories that each participant
regarded as both well defined and shared opens a space for innovation in the
attributional systems of each of them. To fill that space generally requires further
joint exploration, in which each may adjust boundaries and introduce new
concepts and categories in one's own set of attributions as they all jointly
construct what they come to regard as a shared attribution for the entities that have
entered into their conversations.
Clearly, not every relationship between agents is generative in the sense
described above. Maxfield and I identified five characteristics that together
determine the generative potential of an agent relationship:
• aligned directedness - The participants in the relationship need to orient their
activities towards a common zone of agent-artefact space. For example, the
same kind of artefact might be the focus of each of their activities, a!though
the participants need not have the same relationship to the focal artefact.
• heterogeneity - Generativeness requires that the participating agents differ
from one another in key respects. They may have different attributional
systems, competences or access to particular agents or artefacts.
• mutual directedness - Agents need more than common interests and different
perspectives to form a generative relationship. They also must seek each
other out and develop a recurring pattern of interactions out of which a
relationship can emerge. Their willingness to do this depends on the
attributions each has of the other's identity. It helps, but is not necessary, for
the participants to start by trusting one another. Frequently, rather than a
precondition, trust is an emergent property of generative relationships: it
grows as participants <.:arne to realize the unforeseen benefits that the
relationship is generating.
• permissions - Discursive relationships are based on permissions for the
participants to talk to one another about particular themes in particular
illocutionary modes (requests, orders, declarations, etc.). These permissions
74 D. A. Lane

are granted explicitly or implicitly by superordinate agents and social

institutions. Unless potential participants in a relationship have appropriately
matched permissions, or can arrogate these permissions to themselves, the
generative potential of the relationship is blocked.
• action opportunities - Discourse is important, but relationships built only
around talk do not usually last long or affect deeply the identities of
participating agents. Engaging in joint action focuses talk on the issues and
entities of greatest interest -- those around which the action is organized. And
action itself reveals the identities of those engaged in it. In addition, new
competences emerge out of joint action, and these competences can change
agents' functionality and hence identity -- even leading to the formation of a
new agent arising from the relationship itself.
Generative relationships are particularly important when a market system is in a
phase of rapid cascading transformations. In such cases, it is impossible for agents
to foresee all the consequence of their actions, because some of the relevant
entities - artefacts, agents, or more general structures or patterns of interaction in
agent-artefact space - that will determine these consequences may not be visible to
the agents (or even exist!) at the moment in which the agents must act. In
situations of this kind, which Maxfield and I call complex foresight horizons, it
may be much more important for agents to decide with whom to interact, rather
than to attempt to figure out exactly what will come from the interactions in which
they engage. Clearly, the more potentially generative a relationship among agents
is, the more incentive there is for these agents to interact, to explore the new
attributions and interaction patterns that may emerge from their interactions.
Moreover, when foresight horizons are complex, agents may act to enhance the
generative potential of some of their relationships, for example by changing
permissions of subordinates to encourage cross-boundary discourse and by
creating action opportunities even when it is unclear exactly what outcome (and
profit) might be expected to result from a joint project.
Industrial districts are often characterized by complex foresight horizon:;. This is
obviously the case for the information technology and biotechnology clusters in
Silicon Valley and the Washington Beltway, as well as the internet districts of
Silicon Alley and San Francisco. Less obviously, it also holds for most of the
"made in Italy" districts, where the rapid changes in artefact style that
characterizes fashion industries make it difficult to foresee not only what designs
will be selling in the next months, but even what materials will be used in the
fabrication of products - and hence in which kinds of production processes a firm
must engage. Thus, generative relationships play a key role in district
phenomenology, and we should not be surprised to find that most successful
districts have scaffolding structures that help to promote the formation of
generative relationships among district participants.
 Complexity and Local Interactions: Towards a Theory of Industrial Districts 75

6. Locality and Generative Potential

Generative potential may be greatly enhanced by the co-locality of agents in a

relationship. To understand why, consider the following three possible
interpretations of locality. First, local interactions might be defined as those in
which the participants meet face-to-face. Such interactions differ from those in
which the participants are physically remote in two important ways. First, face-to-
face discourse has a broader bandwidth than remote speech, which in turn has
broader bandwidth than written communication. Thus, face-to-face
communication can be more nuanced, providing more space for discussions to go
in unexpected, potentially fruitful directions. Second, individuals can learn much
about and from one another by observing how each interacts both with people and
with artefacts; face-to-face interactions provide the richest context in which this
kind of learning can take place.
Locality could refer to social rather than physical space. For example, we could
describe an interaction as local if all the participants belong to the same
community of practice, even if they are physically remote. For example, the
programmers who work together on freeware projects share a language, a culture
and a project orientation that allows them to collaborate even if they live on
different continents and have never met one another. Of course, the interactions
between these programmers are local in this sense only if they stay within the
boundaries imposed by the implicit rules of their community of practice.
Similarly, we might consider any interaction local that involves participants that
inhabit the same geographical-social community, wherever and however the actual
interactions themselves take place. As with a community of practice, co-
membership in a geographical-social community implies that many kinds of
interaction modes are jointly understood and accepted and may be simply enacted
rather than negotiated. Of course, just what counts as a geographical-social
community, and which kinds of interaction modes qualify as local for a given such
community, are not obvious. Nonetheless, much of the discussion of Italian
districts like Prato or the districts of the Modenese are premised on the assumption
that the participants belong to such a community, and their co-locality underlies
the possibility and form of many of their essential interactions - face-to-face or
not. Indeed, much of the literature on Silicon Valley is based on a similar
assumption.
What all three of these interpretations of locality have in common is the idea
that participants, because of their co-locality, have a higher degree of shared
understanding about what the interaction in which they engage means, so that
there is much less need to negotiate either the context or the consequences of the
interaction event itself. Shared physical space brings this about by increasing
communication bandwidth and hence the direction and visibility of agreement,
while shared social space does it by channelling the interaction into mutually
comprehensible normative forms.
What effects does locality have on the determinants of generative potential?
Certainly locality conduces to mutual directedness. A shared social space affords
76 D. A. Lane

an a priori basis for mutual directedness, at least within the context of the
interaction modes supported by the community, while shared physical space
provides an opportunity to form either positive or negative attributions of others'
modes of operating, and so either enhances mutual directedness or brings an
unsatisfactory relationship to rapid closure.
Similarly, community values can ensure aligned directness for interactions
directed towards transformations consistent with those values, while face-to-face
communication encourages the kind of discursive exploration that can lead to the
discovery of a basis for alignment that may not have been evident a priori.
However, there is always an inherent tension between aligned directedness and
attributional heterogeneity: agents that are completely aligned may come to share
a relevant set of attributions to such an extent that the generative potential of their
relationship may decline or even disappear; and if any relationship becomes
closed to outside perturbations from other, cross-cutting networks of relations,
local interactions among participants in the relationship almost certainly will
hasten this process along. This risk is greatest when the participants are strongly
co-located in social space, and hence begin with similar attributional systems,
which the relationship may fine-tune to homogeneity. On the other hand, extended
face-to-face discourse may reveal subtle unsuspected attributional differences, the
exploration of which may provide space for generation of new attributions. Much
depends on the way in which relationship participants regard attributional
heterogeneity, whether as a threat or an opportunity for the future of the
relationship. Overall, if participants in a relationship are open to the outside and to
the potential fruitfulness of disagreement and misunderstanding, locality in all its
forms is a potent contributor to generative potential.
For this reason, every market system provides scaffolds that make possible
certain kinds of local interactions. Some of these kinds of interactions need not
recur frequently: trade fairs provide an opportunity for many market system
participants to keep up with general trends in artefact development and use
through meetings and discussions with one another, but for most market systems
one or two trade fairs a year is enough. But other kinds of scaffolded local
interactions require frequent recurrence, and these are generally possible only
when the networks that the scaffolds construct and maintain are composed of
agents that are either spatially co-located or sufficiently socially co-located that
they can maintain local interactions at physically remote sites. Scaffolds of this
type include universities and community colleges that support research
collaboration and training for system participants, users' groups, or meeting sites
for various kinds of informal face-to-face interactions like Happy Hour at the
Wagon Wheel in Silicon Valley or Silicon Alley's somewhat more formalized
cyber parties. The possibility to provide scaffolding structures of this type is a
strong incentive for geographic concentration of market system participants - and
lies at the heart of the industrial district.
 Complexity and Local Interactions: Towards a Theory of Industrial Districts 77

7. Industrial Districts

As defined in the introduction, an industrial district is a production subsystem of a

market system, characterized by predominantly small and medium firms
concentrated geographically, with a distributed or decentralized network
organization. A theory of districts ought to provide an account for how a district's
network structure arises and is maintained, explain how this structure carries the
district's production functionality and how the district's links with the rest of the
market system are organized and maintained, and characterize which market
system - locality couplings are likely to give rise to districts.
From the arguments in the previous five sections, it is clear that I would expect
to find districts forming in market systems characterized by persistent complex
foresight horizons, where generative relationships and the capacity of firms to
react rapidly to changes in the structure of the market system are much more
important than economies of scale achie'Vable by vertical integration or the kind of
strategic calculation that are advantageous in markets where impersonal
transactions for artefacts with stable and widely shared attributions of
functionality take place. In addition, entry costs for new firms in a district
production subsystem would have to be relatively low, and the value of artefacts
that the subsystem produces should be socially constructed, relatively
unconstrained by the laws of physics - as, for example, fashion and information
technology. In such market systems, local interactions are particularly important,
and if change in the agent-artefact structure of the market system occurs rapidly
enough, the advantages of both spatial and social locality should be particularly
pronounced.
Exactly where the resulting concentrations of small and medium firms would be
located depends on many factors and historical contingencies, but one obvious
prerequisite is that the geographic regions in which districts emerge must provide
scaffolding structures that support the construction and re-production of the
rapidly changing networks that must carry out the various competences the
district's activities require. At least four kinds of competence networks are
particularly important for districts as here defined: networks of information,
through which agents learn about new developments in the organization of agents
and artefacts in their market system throughout the world; networks of
interpretation, through which agents make sense out of the information they
obtain to determine the directedness of the interactions in which they will engage;
networks of production, in which they collaborate with other district agents to
generate new artefact types and produce artefacts for the market; and networks of
marketing, through which their products find buyers in markets around the world.
Two kinds of scaffolds are particularly important for creating and maintaining
these sorts of networks. The first are scaffolds that promote social cohesion within
the district, while at the same time mixing heterogeneous identities among district
agents. As Annalee Saxenian (1996) has pointed out, Silicon Valley developed a
coherent set of practices that encouraged engineers to "talk shop" with their peers
in other companies, thus providing the basis for the formation of individual-level
78 D. A. Lane

generative relationships the benefits of which redounded to the firms for which
these individuals worked (and frequently the firms they founded as the result of
some of the new attributions they generated). These practices included such
cognitive scaffolds as rules that do not penalize employees from switching jobs
from one firm to another, and physical scaffolds ranging from Happy Hour at the
Wagon Wheel to research seminars at Stanford, Santa Clara and various company
labs. In Modena, the CNA (Confederazione Nazionale dell' Artigianato) holds
frequent meetings in which entrepreneurs may learn about new markets and new
product opportunities from one another, and as the organization grew and the
production competences represented in the Modena area multiplied, its
administrators came to play a brokering role, bringing together entrepreneurs
whose firm's competences might be combined to produce an artefact filling a
newly recognized potential market niche, discovered by another Modena
entrepreneur doing his own business in, say, China or Eastern Europe.
The second kind of scaffold provides for the delivery of local services, which
make it easy to start up a new firm and enable entrepreneurs to concentrate their
resources on honing their particular design or production competences. In
Modena, both local and provincial governments and CNA provide such services,
from arranging financing to outsourcing book-keeping and legal assistance, to
supporting trade fairs at which local firms may display their products to potential
buyers recruited from around the world, to providing space for new firms to
establish their operations, to research and educational projects like PROMO,
designed to introduce local firms to best international practice. In Silicon Valley,
many of the same scaffolding services have emerged privately, with little
government or district-level organization. For example, venture capitalists,
concentrated in several blocks of Sand Hill Road, provide not only seed financing
but networking and recruiting services to prospective entrepreneurs. Local law
firms specialize in legal services for start-up companies; some of them now offer
start-up packages for as little as $5,000 per year (or shares in the new companies),
handling all the legal problems new firms encounter in their first two or three
years of business. Real estate developers build industrial parks where start-up or
expanding firms can locate. The architectural design of many of these parks
provide open space, inside and out, that facilitate the formation of discursive
relationships within and between firms located in the parks. Probably the most
important scaffold for Silicon Valley, at least in its formative years, was Stanford
University, whose engineering graduates were encouraged to start companies in
the area and whose professors were permitted to engage actively in consulting and
entrepreneurial activities of their own. Stanford, San Jose State, and Santa Clara
now all play important roles in promoting knowledge development and exchange,
through seminars, research programs, and teaching (often with instructors
recruited from local businesses) for Silicon Valley firms and professionals. While
local governments around Modena were often in the forefront in providing
infrastructural support to local districts, this was not the case in Silicon Valley,
until some of the "senior citizens" from successful firms in the district established
a lobbying group, the Santa Clara County Manufacturing Group, which became an
important scaffolding institution for the mature district.
 Complexity and Local Interactions: Towards a Theory ot" Industrial Districts 79

An interesting feature of many of the district scaffolds described above is that

they are not inherently restricted to one particular market system. Thus, once a
cluster of new firms is established in a region and scaffolds such as these are
constructed to maintain it, the same scaffolds make it easy for other clusters
operating in different market systems to develop as well. Of course, there must be
some other, market system specific scaffolds around which the new cluster may
form; but leveraging on the pre-existing scaffolds, it is much easier for second and
successive clusters to arise in the same region. This is exactly what we find in the
Modena area, as well as in Silicon Valley: the same scaffolding structures serve
more than one cluster in the region. Sometimes, these scaffolds may even provide
the means to initiate generative relationships between agents acting in different
market systems, that can give rise to hybrid artefacts and eyen, at least in
principle, to new districts or even new market systems.
Let us now return to the three questions I posed in the introduction. My
suggested answers are as follows: district organization can be described in terms
of networks of competence networks and scaffolds; which production activities
might benefit from district organization may be characterized how complex are
foresight horizons associated with the value and functionality of the artefact
family around which the market system is organized; and how successful a district
will be in organizing and maintaining its productive and innovative activities may
be determined by analysing the scaffolding structure it provides to promote
generative relationships and by examining how generative are key relationships
among district agents and those that connect these agents to the distant markets in
which the artefacts they produce are sold.
With these ideas in hand, we can now return to the two perspectives on
complexity introduced in section 2, and ask whether an industrial district qualifies
as a complex system. Certainly, districts as defined here satisfy the
phenomenological criteria for complexity: many heterogeneous agents whose
interactions give rise to interesting emergent structures (the networks that carry
the district's competences and the scaffolds that construct and maintain these
networks); in addition, these structures permit the district to adapt to both external
and internal perturbations to the market system of which it is a part. To see
whether districts satisfy the theoretical criterion for complexity proposed by Gell-
Mann, we must decide what constitute a district's "regularities". It seems natural
in this context to consider the processes through which the district's competence
networks are constructed, maintained and enact and interlink their respective
competences as the district's regularities, and it seems plausible that the minimal
length of this description would be quite long - but this is a question that will have
to wait until a full theory along the lines suggested here have been developed, and
methods for describing and empirically identifying these processes established.
I conclude this section with a proposed, if premature, answer to another question
linking districts and complexity: can a district be a complex adaptive system in the
sense proposed by Gell-Mann - that is, can a district be said to possess a "schema"
or compressed description of the world in which it operates, according to which it
generates actions in that world, and which it may modify on the basis of its
experience? I think the answer is yes, but to describe such a schema and even to
80 D. A. Lane

talk about the district acting require further theoretical development much beyond
the hints supplied here. Briefly, through its networks of communication and
interpretation, the district may build up a set of partially shared attributions about
the relevant agents and artefacts in its market system. Of course, this set is actually
distributed among district agents, but because its elements are generally shared
among these agents, they wiII produce some regularities in the way in which
district agents interact with each other and outside agents, and as a result of these
regularities, attributions that other agents make about the identity of district agents
may share some elements. In this way, attributions can arise for "a way of acting
like a Silicon VaIIey start-up company," and, at least absent strong specific contra-
indications, these attributions may be applied to any "Silicon Valley start-up",
thus reinforcing the behaviour that such an attribution predicts. In addition,
through its scaffolding structures, a district may provide certain agents with
permissions that allow them to speak or act for the district in certain kinds of
situations - for example, in testimony before Congress or regulatory agencies, or
in interviews with journalists who ask what "Silicon Valley" thinks about some
governmental policy or macro-economic context. Both the distributed set of
partially shared attributions and the set of permissions that endow district-level
agency upon particular district agents may be legitimately interpreted as
components of a district schema, and so we may regard the district itself - and nO[
merely the agents of which it is composed - as a complex adaptive system in Gell-
Mann's sense.

8. Methodology for a Complexity Perspective on Districts

What I have sketched above is only an approach to a theory of districts. To

develop a rich theory that identifies, explains and even predicts interesting district
phenomenology would require a considerable supplemental research effort. This
research wiII have to move in two directions. The first is empirical: we need to
learn how to recognize, describe and classify the network-scaffold structure of
districts, and then to examine the kinds of processes that these structures support.
There are already promising lines of work in this direction. Margherita Russo
has carried out several studies in the tile manufacturing district of Sassuolo, in
which she applies ethnographic interviewing techniques to determine how the
formation or absence of generative relationships affect the patterns of innovation
in the district. In one study, she showed how a generative relationship between the
owner of a tile manufacturing firm and a chemistry professor gave rise to a new
technique of tile production which, however, failed to be widely adopted by other
firms, a fact that she explained in terms of the network structure of the district, in
which the innovating firm was isolated from the information and interpretation
networks that linked other district agents. In another study, she showed how the
generative relationships between tile manufacturers and the producers of tile
manufacturing machinery declined, as the latter changed their networks of
relations when they began to produce turnkey plants for foreign customers
 Complexity and Local Interactions: Towards a Theory of Industrial Districts 81

wishing to emulate Sassuolo production techniques. These stories would have

been difficult to tell absent concepts corresponding to network dynamics and
generative relationships, and impossible to uncover with usual aggregate or even
firm-level data.
David Stark is leading an ambitious project to study the intra- and inter-firm
structure of Silicon Alley, the district of internet firms in New York City. He has
been fortunate in his project choice, since the group has been able to track both the
heady emergence of Silicon Alley and its spasmodic response to the crash in
internet stock prices. Using an arsenal of ethnographic techniques, including
detailed observation and recording of individual interaction events, plus the tools
of social network analysis, his group is developing a rich database on network
formation and functioning and agent-artefact and agent-agent relationships, as
well as emergent district phenomenology, including the construction and mode of
operation of scaffolding structures. As the results of the group are published, it is
to be hoped that researchers studying other districts will adopt the methods drawn
from sociology and anthropology that Stark's group uses to supplement the
economics methods of data collection and analysis that dominate most current
district empirical research.
Modelling is the second important research direction necessary to develop a rich
theory of districts from a complexity perspective. Through the use of models that
share the network-scaffold ontology it is possible to carry out observations that
may suggest interesting phenomena that emerge in such settings, which may then
be sought with appropriate empirical tools in districts themselves. It will also be
possible to use these models to carry out simulation experiments to evaluate causal
hypotheses and predictions that the developing theory generates. To implement
the ontology of entities and relations sketched in this paper, it is necessary to
develop agent-based models for districts. Agent-based modelling has become in
the last decade a powerful tool in social science research 5. At least three research
groups have already applied agent-based modelling to study aspects of district
phenomenology. Tommaso Minerva, Irene Poli and Sebastiano Brusco
constructed a cellular automaton model designed to simulate the process whereby
district agents, via their communication networks, may collectively exploit a new
market niche discovered by one of them. Through parameters that tune the rate of
new market discovery and the density of network interactions, they use the model
to predict such aggregate district features as firm-size distribution. Riccardo Boero
and Flaminio Squazzoni have developed an agent-based model 6 of districts
through which they can compare how effectively production networks are formed
in three different environments. In the first two, the only mechanism for network
formation is direct interaction between firms; in one, firms are modelled as simple
myopic optimisers, while in the other, they seek long-term relationships with other
firms. In the third, the district has two scaffolding structures that promote the

5The literature is already vast, as a web search on agent-based modelling will reveal. For a
recent collection of interesting essays on the subject. with many references. consult Kohler
and Gumerman (2000).
6Using the Swarm platform. developed at the Santa Fe Institute.
82 D. A. Lane

formation of interfirm networks. With this model, Boero and Squazzoni

investigate how effective scaffolding structures are as a function of a variety of
tunable model parameters that describe the market system in which the district
operates. Finally, Guido Fioretti has developed an interesting agent-based model?
of the Prato district, through which he tracks the persistence of the district
throughout a series of changes in the market systems in which it operates and in
the internal organization of the district itself.
Another kind of modelling enterprise may also contribute to the development of
the theory sketched in this paper. There is a virtual explosion of recent work on
the structure of networks, sparked in large part by Duncan Watts' work on small
world models 8, and on the dynamics of network formation and growth. Using this
work, it will be possible to give a much more incisive statistical description of
competence networks and the ways in which they interrelate to one another, which
will allow a more refined presentation of district dynamics than would otherwise
be possible. In addition, we might expect in the future that district agent-based
models will incorporate some of the results coming out of the work on how
networks form and grow and what kinds of processes networks with different
structures can support. For researchers interested in developing a theory that can
explain how districts are organized and how they function, the opportunity to
assimilate this emerging research on network dynamics is very exciting.

References

Kohler, T. and Gumerman, G. (2000), Dynamics in Human and Primate Societies: Agent-
Based Modelling of Social and Spatial Processes, Santa Fe Institute Studies in the
Sciences of Complexity Series, Oxford University Press.
Maxfield, R. and Lane, D. A. (1997), "Foresight, Complexity and Strategy", in: Economy
as a Complex, Evolving System II, Arthur, W. B., Durlauf, S. and Lane, D. (cds),
Redwood City CA, Addison-Wesley, Chap. 4.
Saxenian, A. (1996), Regional Advantage: Culture and Competition in Silicon Valley and
Route J28, Harvard University Press.
Watts, D. (1999), Small Worlds, Princeton University Press.

7Also Swarm-based.
8Watts' early work on small worlds is presented in his book (1999).
From Marshall's to the Italian
"Industrial Districts".
A Brief Critical Reconstruction

Giacomo Becattini 1

Abstract. This paper begins by reconstructing the historical genesis of the notion
of industrial district, focusing on the contribution of Alfred Marshall, the
economist who is thought to have discovered this system. The paper also shows
how the notion of industrial district stems from the successful wedding of
Marshall's keen observations of contemporary industrial reality, mainly in Great
Britain, with the desire to provide a way out of the difficulties inherent in classical
economic theory and answer the challenges posed by the class struggle (e.g.,
economic nations).
The economic conditions that fostered the development of industrial districts in
Great Britain recurred in roughly the same manner in Italy after the Second World
War. A few scholars studied, with an open mind, the light industrialization that
was occurring in certain regions of Italy and, after taking a fresh look at
Marshall's theories, rediscovered the phenomenon of industrial districts. The
wealth of literature they produced soon gained attention outside of Italy. This
paper, however, does not dwell on what occurred outside of Italy; it focuses
instead on the development of the industrial district phenomenon and its
theoretical underpinnings in the Italian economy. In particular, it studies the link
between the industrial district system and the manufacture of typical made in Italy
products as it seeks to explain the competitive advantage that Italian
manufacturers enjoy in many industries (for example, textiles and garments,
footwear, furniture, etc.). The third part of the paper presents some of the most
recent developments in the study of Italian industrial districts.
The main point of the paper is that, for certain types of products, industrial
districts can provide a viable alternative to a system based on a single company
(with one or more plants) or a network of companies.

We shall never fully appreciate the importance of what Adam Smith

wrote unless we understand the importance it had for him. For him it
was a matter of living truths, strengthening and arousing hope, that
encourage, almost force us, to search for further truths. And if we
wish them to be useful they must be living truths for us too.
Alfred and Mary Marshall, 1879.

I I am very grateful to several colleagues and friends, most of all to Bellanca, Bellandi,

Groenewegen, Raffaelli, Dei Ottati and Sforzi for useful comments, but, as usual, I do not
want to imply that anyone of them shares all my views.
84 G. Bccattini

1 The Marshallian Industrial District

1.1. It All Started with Marshall

Marshall's writings around 18702 contain several statements that show clearly
how his position differed from the prevailing one among economists on certain
crucial aspects of the theory of production 3 . In particular, Marshall disputed the
standard view that the factory system, in which all manufacturing processes are
concentrated under one roof with a high degree of vertical integration, was
necessarily better than production systems that were technically less integrated but
concentrated geographicalll.
Reading the descriptions of English industry produced by his contemporaries,5
analyzing the most popular works on economics 6 and observing the world around
him with a keen eye 7 (for example, Sheffield's cutlery works and Birmingham's
metal trades), Marshall quickly came to the conclusion that, at least for certain
types of production, there were two efficient manufacturing systems: the
established method, based on large, vertically integrated production units, and a
second one based on the concentration of many small factories specializing in
different phases of the same production process and operated in one location or in
a cluster of locations.
"We shall find", wrote Alfred and Mary Marshall in their 1879 economic
manual 8, "that some of the advantages of division of labour can be obtained only
in very large factories, but that many of them, more than at first sight appears, can

2See Whitaker (1975).

3The entire life and work of Alfred Marshall has been extensively and masterly explored by
Peter Groenewegen. See Groenewegen (1995).
~lt should be noted that Marshall's argument is also valid when vertical integration occurs
in a small, non-specialized business. Marshall is an early exponent of the view that at every
separable phase of the manufacturing process, a firm should assess whether it would be
better off retaining that phase within the internal manufacturing system or outsourcing it in
the marketplace.
sFor example: Cooke Taylor (1841) and Sargant (1857), which was innuenced greatly by
Les Ouvriers Europeens by F. Le Play, whose analytical method was carefully studied by
Marshall.
6Chief among them Mill (1848); Cairnes (1874): Cliffe Leslie (1888), but also Hearn
(1863). Hearn devoted some attention to the phenomenon of the industrial districts, to some
extent anticipating Marshall. See Pesciarelli (1999). The main work targeted by Marshall
was probably Fawcett (1863), commonly known as "Mill and water", because it was
considered a mere summary of Mill (1848).
7Keep in mind that Marshall was not an "armchair economist". From an early age, he
showed great interest in the technical and organizational details of the manufacturing
process. Consider, for example, the sketches of industrial facilities and equipment he used
to illustrate his survey of U.S. factories. See Whitaker (1996), vol. I, pp. 51-59 and 80-81.
8See A. and M. P. Marshall, (1881), lSI ed. 1879; from the 2nd ed. pp. 52.
 From Marshall's to the Italian "Industrial Districts", A Brief Critical Reconstruction 85

be secured by small factories and workshops, provided there are a very great
number of them in the same trade.
The manufacture of a commodity often consists of several distinct stages, to
each of which a separate room in the factory is devoted. But if the total amount of
the commodity produced is very large, it may be profitable to devote separate
small factories to each of these steps. If there are many factories, large or small, all
engaged in the same process, Subsidiary Industries will grow up to meet their
. I wants 9 .
specla
But small factories, whatever their number, will be at a great disadvantage
relative to large unless many of them are collected together in the same district lO •
Thus both large and small factories are benefited ( ... ) (by the localization of
industry). But these benefits are most important to the small factories, and free
them from many of the disadvantages under which they would otherwise labour in
competition with large factories.
And finally: "in these districts a further division of specialisation has grown up,
and separate trades have sought separate localities ( ... ). Those that work in wool
do not generally live among the Lancashire cotton workers, but are collected
together in Yorkshire; and they themselves are divided into the "wollen trade" and
the "worsted trade", and these again spread out into various branches, each of
which has a favourite district of its own"ll.
Here Marshall, in addition to offering a phenomenological description of the
manufacturing district, provides some hints for a first, clear theoretical frame of
reference based on "benefits", which he later called "external economies,,12.

1.2. The District Concept in Marshall's Thinking: a) The Problem of

the "Economic Nations"

The most logical starting point for exploring the birth of the notion of industrial
districts in Marshall's early writings can be found in the economic debates that
raged when he made his debut as an economist (1871-1873) 13, especiall y those
that arose after the publication of J. S. Mill's Principles of Political Economy and
involved primarily J. E. Cairnes l4 and T. Cliffe Leslie. These authors had

9/bid., p. 52. This is the origin of the conceptual bifurcation leading on the one hand to the
manufacturing district and on the other to the industrial sector or to the industrialfiliere.
IOHere the territory is seen as a conduit for the creation and the dissemination of the
"advantages",
II/bid, p. 47.
12Notable studies that provide a historical framework for the development of Marshall's
concept of district include those by Bellandi (1982); Loasby (1998); Loasby, (1998); and
Raffaelli (forthcoming). See also, for a good background Loasby (1989).
I3Marshall's first theoretical economic work was an 1872 review of The Theory of Political
Economy (1871) by S.levons. See Pigou (1925), pp. 93-100.
14The essay containing Marshall's first serious theoretical work is a critique of Cairnes's
critique of 1. S. Mill, See Pigou (1925), pp. 119-133.
86 G. Becattini

discovered that the professional and territorial geographical mobility of labour and
capital postulated by classical economic theory did not exist - not even in England,
the promised land of classical economic laws. In practice, British social reality
was fragmented into regional, sectorial and social compartments, which hindered
the free circulation of capital and labour that was supposed to exist in theory. The
problem was then how to detect (or circumscribe) areas or groups of agents so as
to allow the to and fro between phenomena and theory required by an empirical
science like economics. Lacking such a methodological exploration, classical
economic theory would be left floating in mid-air. This explains the cruciality of
the theme of the "economic nation", which the young Marshall derived from the
theory of "non-competing groups" and developed into two separate versions 15.
In the first version, Marshall's "economic nation" is a place, or a system of
places, characterized by such a high degree of cultural homogeneity (values and
institutions), free circulation of information and territorial continuity that the
movement of capital and labour can quickly equalize profit and wage levels. One
could then say that any geographical area of this type constitutes a sort of
"economic nation", even when it does not become fully conscious of its autonomy
and does not develop a unified system of government. A "political" nation (the
usual nation-state) consists of a series of "economic" nations, and it is entirely
possible for an economic nation to straddle the borders of several political
nations l6 •
In the second version, an economic nation is a "block of subjects" within a
nation-state (e.g., miners and owners of coal mines, the farm lobby, etc.) or
encompassing several nation-states (e.g., the working class) who recognize that
their fundamental economic interests (or an important part of those interests)
coincide and are at variance with the interests of similar blocks in the same nation.
The knowledge awareness of a commonalty of economic interests interacts over
time with other non-economic aggregants to make these social bodies coalesce
into different forms and change over time the degree of their social cohesion.
This second type of "economic nation" also differs from the first 10 that it
always has a more or less explicit form of government and a more or less
complete and coherent "foreign policy", so to speak, on the basis of which it forms
alliances or engages in confrontations with similar groupsI7.
This also means that a dialectical interaction, the nature of which has yet to be
fully understood, entails the existence in every "economic nation" of the second
type not only of a contrast between the interests of masters and men but also of a
common aggregant that enables them to stand together against the rest of society.

ISOn the young Marshall, see: Dardi (1984).

161 discussed more fully the notion of "economic nation" in my essay "Nazione economica
e nazione politica nel pensiero di Alfred Marshall" in Roggi (1994). See also, in the same
book, some brief remarks by Marco Dardi on this subject.
17 At the time, this construct, which may seem bewildering to contemporary common sense,
was not just the artifice of a scholar in his ivory tower. To understand this, all one has to do
is think of the "guild socialism" political movement, which, not accidentally, attracted
Marshall's attention. See his book (1919), Book Ill, par. XIV, 7.
 From Marshall's to the Italian "Industrial Districts". A Brief Critical Reconstruction 87

This aggregant is not necessarily the same as the one that binds the inhabitants of
a certain area or the one that is presumed to hold together the international
working class, but is probably of the same type lR .
What is the general thrust of Marshall's effort [0 develop the notion of an
economic nation separate from the nation-state? In my opinion, it expresses
Marshall's refusal to derive the basic unit of economic research from real world
entities such as nation-states and provinces. In this sense, it is difficult to define a
nation as an empirical entity. Even the use of history or geography as a means of
classification entails an effort at conceptualization or abstraction that can be
conditioned by countless and undetermined historical and geographical influences.
The unit of research must be such as to facilitate an abstract-concrete-abstract
flow. The overall action of socioeconomic forces constantly defines and redefines
the unit of research, which is at any rate no less than some sort of "economic
nation", i.e., a plurality of individuals who feel more closely linked than is
generally the case, in a grouping that can fall under either the first or the second
definition, or bothl? I believe that one of the traits that most distinguishes
Marshall, even in the classification process, is the importance he attaches to
personal empowerment, i.e. his concern for the individual and his thought process,
without making claims of absolute objectivity. It may be worthwhile to emphasize
this point.
This point of view hints at the possibility of shifting the analytical approach
typically used by economists from the study of the behaviour of individuals in the
commodities markets 20 , the identification of which heavily conditions the
conclusions drawn from economic theory, to the study of the behaviours that are
typical of representative subjects. Obviously, this approach leads to human
groupings that are represented by a limited number of agents (e.g., the inhabitants
of a certain town) and/or enterprises (e.g., a "population" of firms). In my opinion,
if this different way of "slicing" or approaching social phenomena contemplated
by Marshall in his early writings had become established 21 , it would have yielded

18The only contemporary economist who has, to my knowledge, addressed this issue is Jan
Steedman, who also makes reference to Marshall in his essay, 1986. However, he does not
seem to have followed up on the new ideas presented in his essay.
19The modernity of this notion is clearly shown in the following passages by C. p.
Kindleberger: "The optimum social area is a function of the average citizen's sense of
participation, of having a share in decisions, of counting ( ... ). While the optimum economic
area may be large, the optimum social area is clearly much smaller ( ... ). Moreover while
the optimum scale of economic activity is getting larger and larger, the optimum social
scale appears to be shrinking". Kindleberger (1989), pp. 87-88.
2°In this area, see Lange (1952), Appendix, who writes: "In economic science, however, the
classification of goods cannot be made on a purely arbitrary basis, because the laws of
economics would then be dependent on the particular classification adopted. This would
restrict the significance of the propositions of economics to a degree that would make them
useless".
21Marshall, pressured by a mainstream economics that was moving in a totally different
direction, did not have the will, or the strength, to follow this approach to its ultimate
88 G. Becattini

important consequences, in terms both of the division of scientific labour in the

study of social phenomena 22 and the ongoing development of strictly economic
thinking.

1.3. The District Concept in Marshall's Thinking: b} Humans as

Variable Entities

Alfred Marshall's ability to see districts where everyone else saw mere industrial
agglomerations, or, in other words, his willingness to apply Smith's notion of the
division of labour to entities other than the world, a nation-state or a factory (for
example, Smith's pin factory), is deeply rooted in the classifications of capitalism
developed whiie he was still young.
While this is not the place to discuss this point in detail, a few brief remarks and
references seem to be in order.
There are at least two core theoretical problems - we understood this only
recentl/ 3 - with the notion of district: the first, already mentioned, concerns
economic nations and the second, perhaps more profound, is rooted in the studies
of the human mind carried out by Marshall during his younger years. The "district
anomaly", as I like to call the complex of ideas that gravitate around the notion of
industrial district 24 , contains a transparent reference to the social and dynamic
concept wherein the process of growth of human intellectual abilities is the
specific engine that drives the human aspects of natural history. Marshall believed
that humans evolved mainly through work and, therefore, evolved differently in
different organizational and social contexts 25 • His industrial atmosphere, for
instance, is more than just the product of an uninhibited observation of the facts. It
is also a symbol of Marshall's notion of natural sociability of man and of the
historical roots of a social process that is simultaneously a process of production
of commodities and of the associated social change. In the final analysis, this
localized process is nothing more than a "fragment" of social life, taken at a given
point in time and space and viewed from the standpoint of the need to meet other
historically perceived human needs.

consequences. However, it: a) continued to have a "subterranean" inHuence on the way

Marshall approached the study of economic phenomena and b) caused profound
incompatibilities with mainstream economics.
22The division of the study of social phenomena into economics, sociology, anthropology,
social psychology etc., which became established during the second half of the nineteenth
century, would have been unthinkable.
23Especially after the studies by Raffaelli. See Raffaelli (1994).
24 For a more general understanding of the anomalies of Marshall's thinking compared with
mainstream economics, see my essay "Anomalie marshalliane", 2000a. For an English
version, see: "Marshallian Anomalies", 2000, new series.
http://www.cce.uniti.itldse/marshalllwe!come.htm.
25There are numerous passages, especially in his earlier works, where Marshall clearly
postulates this type of environmental and cultural conditioning. In particular, see the text of
the lecture entitled "Some Features of American Industry" in Whitaker (1975), pp. 355-376.
 From Marshall's to the Italian "Industrial Districts", A Brief Critical Reconstruction 89

As opposed to dogmatic economists, who on the one hand refused to give up the
"wage fund" theory and on the other were blinded by the economies of scale and
the factory system 26 , Marshall showed with a concrete example (the district
formula) that there was already, in practice at least, one necessary alternative or
perhaps a complement to the esosomatic increase in the productivity27 (more and
better machines per capita) of individual workers, i.e., an "endosomatic" increase
in productivity achieved by paying attention to values and knowledge that enhance
the compositeness of the place with which the worker himself "identifies".
Another characteristic of Marshall's analysis that dates to his early work is a
dialectic juxtaposition of routines and organizational innovations in the broad
sense, duplicating a framework the rools of which can be traced to Marshall's
early reflections on the workings of the human mind. Society, in a manner not
significantly different from the human mind, alternates routine behaviour with
innovative behaviour, which, if successful, becomes in turn routine, freeing new
energies for innovation 28 .
In this way, Marshall shifts the focus of economic analysis from the esosomatic
means of production (equipment, facilities, canals, railroads, etc.), which
incorporate routines and are acquirable and cumulable, to individual, endosomatic
tools, such as the intellectual potential of the human mind, which are constantly
open to innovation and are inseparable from their carrier, and to collective assets,
such as social, local or sectorial capital, which by their very nature cannot be
appropriated because the~ consist of delicate and shifting networks of
interpersonal relationships 2 • This position inadvertently created a wide chasm
between Marshall and mainstream economists, both classical and marginalists. It
also caused Marshall's thinking to move further away from Marx's notion of the
capitalist production system, which placed excessive emfhasis on the separation
between workers and the "material" means of production 3 .

26Conservative and a few "progressive" economists bow reverently before the so called
wages fund theory. Marx, who is not fooled by this lrick, accepts acritically - as history
later demonstrated - the claims of those who sang the praise of the factory system and the
economies of scale. In my opinion, this view of Marx's, which anticipates Fordism. harbors
the early roots of the failure of centralized economic planning.
27This distinction comes from the writings of Georgescu Roegen. See Georgescu Roegen
(1966), pp. 98-99.
28See Raffaelli (1994). But also his forthcoming (Routledge 2002) book on Marshall.
29Marshall speaks of "social credit". See Marshall (1919), Chap. IX, 1.
30It is interesting how today's crisis of Fordist capitalism, which lent itself better to a
Marxian analysis, gives new significance to the viewpoint proposed by Marshall in this
area.
90 G. Becattini

1.4. Semi-Automatic and Man-Made Districts

In his work, Marshall differentiates between two types of districts: one that
represents an evolution of the "centres of specialized skills" from the pre-
industrial age and another created intentionally and in an organized fashion as the
result of the spillover of some manufacturing and craft activities from the
industrial cities typical of English capitalism at the end of the nineteenth century.
In the first case, of which Sheffield, with its long iron-working tradition, is a
perfect example, the key feature is the interdependence between technical-
economic issues and the underlying sociocultural relationships. It is a long and
complex process in which a steady flow of socioeconomic innovations leads to a
process whereby the most efficient ones are consolidated into institutions 31 .
The second case foreshadows, in embryonic form, the role of public policy in the
formation of manufacturing districts. In a paper that became the manifesto of the
Society for Promoting Industrial Villages, which was close to the Garden City
Movement, of which, significantly, Marshall was a member, he supported "the
formation of a colony in some place well beyond the range of London smoke ( ... ).
The work of several firms, not always in the same business, might in some cases
be sent together. Gradually a prosperous industrial district would grow up, and
then self-interest would induce employers to bring down their main workshops,
and even to start factories in the colony ( ... ). It is only the first step that costs;
every succeeding step would be easier,,32.
In conclusion, Marshall's vision of the manufacturing district is not just the
concentration of manufacturing activity in a given territory (the clustering or
networking of businesses referred to by many economists today) that results from
the "natural" development of market automatisms 33 . In Marshall's vision,
economic and socio-cultural phenomena are fused into a single conceptual block,
paving the way for the implementation of cogent public policies 34 .

1.5. Paths of Industrial Development

To better understand the "district anomaly", we must consider briefly some of its
implications in terms of the understanding of the concept of industrialization 35 . If
the "district anomaly" had succeeded, industrial capitalism would have been

31Several fragments of a historical-conceptual reconstruction of the path leading from the

pre-industrial centres of specialized skill to the industrial district are scattered among
various chapters and appendices of Marshall (1919).
32See Pigou (1925), pp. 149-150.
33Marshalilikes to speak instead of "serni-automatisms". E.g. Marshall (1919) p. 599.
34This clearly differentiates Marshall's districts from any involuntarily crt!ated locals
systems, such as those obtained through an extension of Hayek's models. For example,
Parri (1997), pp. 175-190.
35For a more detailed analysis, see Becattini (2000b), p. 23. An English version of this book
is forthcoming.
 From Marshall's to the Italian "Industrial Districts", A Brief Critical Reconstruction 91

understood as having not just one development path based on accumulation and
technical progress, with firms of increasingly large average size combined in
increasingly concentrated industries (as most of the economists of the time
thought and almost all later economists would continue to believe for a long time),
but of mUltiple paths (at least two), one of which would have produced territorial
agglomerations of medium-size and small firms.
An early understanding of the existence of "multiple paths to industrialization",
which, by crisscrossing and cross-pollinating each other, coexist over time, could
have had far-reaching consequences. I will just mention a few of them: a) an
understanding of the persistence and success of independent forms of employment
and small businesses 36 in certain industries would have refuted the mechanistic
extrapolation that industrial capitalism produces a social polarization between the
two classes of owners and proletarians 37 ; b) a denial of the view that capitalism
leads to the creation of a few megalopolises would have helped focus territorial
studies toward a plurality of alternative models of territorial agglomeration; c) the
assumption that there is a plurality of interactive commonalities between
manufacturing activity and everyday human life would have refuted the vision of a
manufacturing system increasingly enclosed inside the factory and/or secluded in
"industrial quarters" that exploits and upsets the balance of surrounding socio-
cultural and natural resources 38 ; and d) a refusal of the "easy" view, supported by
socialist critics of capitalism, that the trend toward the technical, economic and
financial concentration of the manufacturing sector paves the way toward a
generalized shift from private to public property, would have spared humanity
some painful disappointments 39 •
The fundamental question raised by Marshall in his early writings about which
basic unit is best suited for a correct economic analysis casts doubts on the
acritical acceptance of de Jacto empirical compartmentalisations - such as the
concept of nation-state at a given time in history - and paves the way for the

36This, which is a recurring theme in Marshall's work, found a non-receptive cultural

environment and caused Marshall to be seen as a laudator temporis actio
37Marshall does not accept the idea that the concentration of wealth and poverty at the two
extremes of society is inevitable. He recognized that capitalism contains strong tendencies
toward that outcome, but also recognized the existence of counterbalancing tendencies,
especially in the economic domain, (for example, the tendency toward bureaucratization
among large public and private entities is contrasted by an endless supply of new initiatives
from small entities striving to succeed) but also in the moral and intellectual domains (the
trend toward more civilized behaviour in all social strata).
38It is not by accident that Marshall always viewed as "problematic" the so-called sweating
system. because, in his mind, this system was one of the manifestations of the embedment
of industry into society and could also be a cause of growth. This position made him highly
unpopular among left-wing politicians and union representatives, See, for example, Pigou
(1925), p. 225.
39See Marshall (19\9).
92 G. Becattini

current phase of territorial fragmentation-recomposition (international integration

and sub-national segmentation)4o.
The economic, cultural and political consequences of the "district anomaly",
widely interpreted, had it been accepted, could have been significant.
Unfortunately, in his later years, when Fordism was triumphant, Marshall, while
remaining critical 41 , flirted with the dominant theory and shifted his
externallinternal economies from the industrial "district", where they had been
born, to the industrial "sector", where, as it is well known, they do not fit very
we1l 42 .

2 The Italian Industrial District

2.1. The Unnoticed Decline of Fordism

Toward the end of the 1960s, many years after Marshall's death in 1924, a few
Italian economists detected some interesting developments. In particular: a) in
certain regions of Italy (Tuscany, for example)43 where large private and public-
sector companies in capital-intensive and/or high-tech industries were showing
clear signs of weakness, there was a "strange" bloom of small manufacturing
businesses, which produced increases in local income, jobs and exports 44 ; and b)
the small companies included in these clusters were adequately equipped on a
technical level - at least as well as their large competitors were - to do the things
they were doing45. These two observations were "scandalous" for those who
believed in the then dominant economic theories. As a result, for several years the
scientific and political communities dismissed these businesses as precarious and

40Kindieberger writes: "social and political disintegration ( ... ) illustrates a tendency

towards smaller units. The reaction against the multinational corporation ( ... ) is a response
to intrusion by foreign elements and the fear that no· members of the group are making
decisions affecting it". See Kindleberger (1989), p. 87.
41In line with his unpublished manuscript (Ye Machine), Marshall levels important,
criticisms at the notion of scientific organization of labour, which was accepted acritically
by most mainstream economists and was incorporated illogically and with well-known
results in the socialist ideology of centralized economic planning. See Marshall (1919).
42 I allude to the criticisms by Sraffa. A discussion of the reasons for this shift would be too
complex to tackle here. In his first criticism of Marshall, Sraffa grasps perfectly well the
alternative, but fails to explore its territorial ramifications. See Sraffa (1998), p. 346.
43 See Becattini (1978), Sforzi (1994).
44The early classifications, still based on first impressions, identify a Third Italy (Bagnasco)
and a North-East-Centre, NEC region (FUll). However, economists sensed from the start
that the reason for the efficiency of clusters of small businesses must be some factor that
links the production system with its social and cultural hinterland context. On this topic,
please see my work (1978), pp. 107-123.
4SSee Brusco (1989), pp. 59-154.
 From Marshall's to the Italian "Industrial Districts". A Brief Critical Reconstruction 93

marginal and as the alleged harbingers of an inevitable deindustrialization, leading

to the social and civil decay of those regions that were being squeezed between
industrially advanced countries with their wealth of capital, manufacturing
infrastructures and technical expertise, and underdeveloped countries with much
lower labour costs.
Despite these views, the phenomena described above continued, developing a
new and equally "strange" characteristic: the new companies were not being
created inside the industrial cities and across the full range of industrial sectors.
Instead, they were being established across a vast geographical territory -
somewhere between the regions of traditional capitalistic development and the
economically depressed areas of Southern Italy - where they were concentrated in
relatively small areas that, in most cases, should not have been attractive (because
insufficiently infrastructured or not easily accessible, for example) according to
the theories of industrial localization prevalent at that time and in industries that
most industrial economists considered "mature" and lacking in growth
opportunities (textiles, garments, footwear and leather goods, wood furniture,
etc.), using forms of business organization (small companies, primarily family
owned) considered obsolete. What was the reason behind this vigorous resurgence
of archaic and irrational forms of manufacturing in washed-up industries? Before
attempting to give a direct answer to this question, I would like to make some
parenthetical observations.

2.2. General Conditions for the Rise of Industrial Districts

Let us start with a general question: under what conditions can industrial districts
nourish and when do these conditions occur? These conditions can be divided into
two groups: "local supply" conditions and "general demand" conditions.
The local supply conditions include "the existence of countries that, throughout
the classical period of industrialization and the First and Second World War,
retained in some parts of their territory: a) sufficient cultural complexity, made up
of values, knowledge, institutions and behaviours ( ... ) that elsewhere had been
pushed aside by a generally industrial and massifying culture, and a manufacturing
system that included factories, craft or artisan workshops, home workers and
family production units; and b) a credit system ( ... ) ready to finance small but
promising businesses and help people use their increasing amount of spare time to
produce goods that could be sold in the marketplace,,46.
The general demand conditions include the progress, in many countries, of large
segments of the middle class beyond the normal standard of comfort found in each
of these countries 47 . Progress beyond this standard creates the conditions for the

46See Becattini (2000b), p. 23.

oI7For the concept of standard of corrifort, as opposed to that of standard of living, one can
refer to the author who first proposed this distinction, i.e., Alfred Marshall. See Marshall
(1961), Book 6.
94 G. Becattini

emergence of new sets of needs with a high social and "qualitative" content that
produce a highly variable demand for differentiated and personalized goods.
The combined impact of this twin set of conditions undermined the advantages
of large factories, consisting primarily of internal economies of scale in the
production of standardized goods and continuous production processes (i.e.,
processes that could not be broken up into phases), favoring instead smaller
factories with stronger ties to the surrounding community, where people worked
separately in a limited number of phases of a common (to the industrial district)
complex process. When there is an adequate "industrial atmosphere", that includes
both technical knowledge and "commercial morality", the shift from an
environment characterized mainly by hierarchical relationships between
departments - as they exist within a large enterprise - to one dominated by
transactions between independent firms (where the existing inequalities tend to
diminish and eventually, for the successful ones, to disappear) creates a positive
trust differential 48 between the participants that lowers the overall cost of the
cluster's production 49 .
Whenever these two sets of conditions come into existence in a situation
dominated by manufacturing in its industrial or preindustrial state, the district
form of production becomes relatively stronger so .
In a nutshell, my conclusion is this: Italy, during its "economic miracle" phase,
when it was part of a rapidly expanding Western world, saw the development of
conditions similar to those experienced by many British cities and towns during
the first great expansion spawned by the industrial revolution. In both cases, a
long and socially unequal income expansion concentrated wealth in the pockets of
a large segment of the middle class, who, having achieved the current standard of
comfortS I, sought increasingly differentiated and personalized goods and services
as a means of experiencing "new sensations" and achieving social prestige.
Endless series of standardized objects became increasingly less attractive to this
new, affluent class, whose demand was increasingly fragmented and variable.
These developments on the "real" side marked the birth, or rebirth, of a "new"
theory of industrial districts.

48Something of the same kind of the already mentioned marshallian "social credit".
49Marshall pointed out that the "scholastic" concept of production costs for a single product
was not realistic. A study of the production process as part of the industrialization process
inevitably leads to the conclusion that an analysis of intermediate and long-term
development must take into account the web of cost wnjunctions and connections. See
Marshall (1919), Book II, Chap. I.
50Naturally, this can be understood only if there is an economic theory that identifies this
phenomenon. Districts can be nourishing in the real world, but the prevailing theory can
fail to recognize them. It is here, at the dialectic intersection between phenomena and
theory, that the decisive importance of Marshall's break with classical orthodoxy and
neoclassical economic theory becomes fully apparent.
511n Victorian times it meant: home with garden, nice living room possibly with Morris's
wall paper, sons at the public school, etc.; in post-war Italy: town nat, home appliances,
car, vacations, etc.
 From Marshall's to the Italian "Industrial Districts". A Brief Critical Reconstruction 95

2.3. The Italian Way to Industrial Districts

An explanation of these phenomena, which came naturally to some Italian

researchers as a way to solve the impasse they were facing when applying the then
dominant theory, was the existence of a factor that increased productivity (and
innovation potential) and which had more to do with physical contiguity than with
investments in means of production. This factor of higher productivity - it should
be noted - has nothing to do with the exploitation of the inferior position of
employees, as often occurs in areas where many small firms gravitate around just
one or few large companies 52 .
A factor of this type already existed in economic theory: economies that were
external to anyone firm, but internal to an industrial sector or territorial group of
firms. Driven from the context of the industrial sector by Sraffa's criticism 53 and
little loved by mainstream economists because they provided a theoretical
justification for government intervention in economic matters 54 , external
economies, in a backward leap to Marshall's thinking in 1879, reappeared in
Italian economic thinking in a territorial context55 .
However, this was not the only path 56 followed by Italian scholars to develop
the concept of district. Sebastiano Brusco, for example, arrived at the notion of
industrial district from a theoretical basis that has nothing to do with Marshall. By

52F. Pyke and W. Sengenberger called this the "high road" to economic development, as
opposed to the "low road," which is based on exploitation. See Pyke and Sengenberger
(1992). In view of the decisive role it played in going to the roots of the industrial district
issue and publicizing it internationally, I must also mention the International Labour
Organisation (ILO) research program, The New Industrial Organization, which was
launched in 1986 and coordinated by W. Sengenberger, with the initial support of M. 1.
Piore. It produced several books: from: Pyke, Becattini and Sengenberger (1990),
Cossentino, Pyke and Sengenberger (1996).
53 1 am referring to the dispute about costs originated from a famous Clapham's article. See
Clapham (1922a). See also Clapham (1922b).
541 am referring to the area of research stemming from Pigou (1920).
55The first work where, to my knowledge, internal and external economies started again to
perform the function envisioned by Marshall in his early works is: Istituto Regionale
Programmazione Economica della Toscana - IRPET (1969). Republished in: Becattini
(1999), pp. 116-146.
56 1 am afraid I may have overlooked some of the other paths that led to the notion of

industrial clusters, and I apologize to the colleagues whose names I did not mention. I will
provide a more complete analysis on another occasion. However, I would like to mention
several books on this subject that will give the reader a fuller picture of this issue: Garofoli
(1983); Becattini (1987); Brusco (1989); Gobbo (1989); Pyke, Becattini, and Sengenberger
(1991); Nuti (1992), 2 voll; Onida, Viesti and Falzoni (1992); Bellandi and Russo (1994);
Dei Ottati (1995); Brusco and Paba (1997); Varaldo and Ferrucci (1997); Belfanti and
Maccabelli (1997); Cora and Rullani (1998); Becchetti, Belussi and Gottardi (2000);
Tattara (2001); Becattini, Bellandi, Dei Ottati and Sforzi (2001), which includes a rich
bibliography. The literature is much vaster but I think that the works mentioned above point
the way to most of the Italian and foreign work on industrial districts.
96 G. Becattini

applying his analysis to the industrial phenomena of the Emilia Romagna Region,
Brusco developed the concept of industrial district without the aid of external-
internal economies and despite what I like to call the handicap of his Sraffian
background57 .

2.4. Made in Italy Products

At the beginning of the 1990s, a major research project carried out by Michael
Porter and a group of associates 58 opened a new approach to understanding the
issues of manufacturing systems. Drawing on an in-depth analysis of world
commerce, Porter discovered, (especially in Italy but in other countries as well),
the existence of territorial clusters of small businesses that had gained significant
competitive advantages in relatively important niches in the world trade for
manufactured goods (textiles, garments, footwear, leather goods, furniture,
decorative tiles, etc.). These advantages were achieved despite the initial
technological superiority of industrial giants and the lower relative labour costs of
other competitors. How could this development be explained? In a series of
papers59, Porter and his collaborators refined the concept of business cluster,
giving it a form that closely resembled that of Marshall's industrial district. The
prestige of the center of learning where these analyses were carried out - the
Harvard Business School - and the fact that Porter's conclusions were in tune with
the views prevailing at the time - i.e., that business competitiveness is at the centre
of everything - paved the way for acceptance of the concept of business clusters in
mainstream economics and forced average economists to take a new look at the
notion of industrial districts, which until then they had found methodologically
troubling.
In Italy, this convergence of conclusions - the efficiency of local systems
claimed by the supporters of industrial districts and the competitive advantages
pointed out by Porter - led to the development of a new area of research that
focused on those products that are emblematic of Italian exports. A number of
Italian scholars (Marco Fortis, Giuliano Conti, Stefano Menghinello and others)60
sought to verify, expand and consolidate the theme, which Porter had touched on
only briefly, of the relationship between industrial districts and the success of
Italian exports. Fortis's work helped bring into focus certain industries that

57Why do I use the word "handicap"? Because the entire edifice of Sraffa's theories is
designed to prevent the entry of the entrepreneurial contribution in the core of economic
theory. It is truly remarkable how Brusco, starting from these premises, developed the
concept of industrial clusters, which are genuine incubators of the entrepreneurial spirit.
Brusco's approach to the problem is described in some detail in Brusco (1989). Peter
Groenewegen remindered to me that Maurice Dobb read a paper, in 1923, on "The
entrepreneur myth", to the economic Section of the Cambridge Club The Heretics.
58See Porter (1989).
59See Porter (1999).
60See e.g. Bagella and Becchetti (2000).
 From Marshall's to the Italian "Industrial Districts". A Brief Critical Reconstruction 97

manufacture typical Italian exports, concluding that they represent a true "system
of manufacturing sectors and sub-sectors" that are interdependent and specialize in
the production of household and personal goodS 61 . Conti and Menghinel\o studied
the relationship between export performance and territorial concentration in detail,
consolidating and expanding Porter's conclusions 62 . Becattini and Menghinello
sought to identify, within the general universe of "made in Italy" products
classified by industrial sector), a core group of products manufactured in industrial
districts that represent the cutting edge of Italian exports 63 .

2.5. The New Panorama of Italian Research on Industrial Districts

1ST AT, the Italian statistical institute, with a vital contribution from Fabio Sforzi,
divided Italy into local systems64 that included 199 industrial districts 65 . This
permitted a comparison of the many different characteristics that distinguish
district areas from non-district areas. The most important product of the studies
carried out in this area was an econometric study coordinated by Luigi Federico
Signorini for the Research Department of the Bank of Italy that confirmed the
existence of many of the relationships that are typical of the "theory of industrial
districts,,66.
Two institutions also came into existence at this time: a) the Club dei Distretti
(1996), without permanent headquarters, which sought to represent the interests of
industrial districts vis-a-vis public authorities at all levels, from local municipal
governments to the European Union and the WTO, and b) an Industrial Districts
Observatory, which was established in Milan (1997) through a collaborative effort
of Montedison Economic Research and Innovation Department and the Research
Centre in Economic Analysis, International Economics and Economic
Development (CRANEC) of Universita Cattolica, joined in 1999 by Fondazione
Comunita e Innovazione 67 .

61 See Fortis's several books, but particularly, 1996, 1998.

62See Conti and Menghinello (1996), pp. 286-303.
63See Becattini and Menghinello (1998), vol. V, n. 9.
64See Sforzi (1989), ISTAT (1997).
65See Istat (1996), pp. 262-271.
66See: Signorini (2000).
67The work of two Rome-based research institutions, Centro Studi Investimenti Sociali
(CENSIS) and Istituto Tagliacarne of the UlCCIA, on districts and local issues is also
worth mentioning.
98 G. Becattini

2.6. Principal Characteristics of Italian Industrial Districts

The recognition of the heuristic usefulness of the concept of industrial district and,
consequently, of the empirical existence of groups of phenomena that could be
effectively analyzed with such a tool (this happened more or less at the beginning
of the 1990s) was followed by a sort of terminological explosion: from the notion
of "industrial district" to system area, to local production system, to milieu
innovateur, to cluster and so on. This is not the place to review such immense and
varied literature, nor would it be useful to lump often heterogeneous concepts. I
shall just explain the path that I followed with some of my associates.
Instead of starting with a model developed by finding similarities with (and
differences from) other well-known types of manufacturing system organizations
(for example, industrial poles), which could have been restrictive and distortive of
new emerging phenomena, our group began with the assumption that, since
industrial districts follow their own peculiar productive and reproductive logic, an
understanding of their development could come only from an in-depth study of an
industrial district in action over a sufficiently extended period of time. Having
determined that Prato was an industrial district - perhaps the archetype of all
industrial districts - we concluded that a study of the Prato district would yield
significant insights into the anatomy and physiology of all of the industrial
districts 68 .
What did we find in this industrial cluster?
First of all, a population of families and businesses interacting with each other
in various ways within a well-defined territory. Looking more closely, we
discovered that the businesses could be broken down into different populations
working on different production phases of production (spinning, weaving, dyeing,
finishing, etc.)69 organized in flexible teams 70 normally headed by a finished
goods manufacturer (e.g. wool fabric mill, or impannatore) that interacted with the
external market. The labour and consumption behaviour of the families was
divided, more or less, along lines that reflected their position within the local
manufacturing process (industrial entrepreneurs and craftsmen, regular employees,
"textile sharecroppers" tradesmen, etc.).
More importantly, we were surprised to discover that the prices of semifinished
goods were determined by a system that on the one hand pegged prices to the
same (or practically the same) goods found elsewhere in the world and on the
other set prices that were strictly related to the standard of living and social status
of the main categories of families living in the district. Because of their impact on
the income of crucial socio-economic groups within the district, the relative prices
of processing tasks played a major role, together with wages levels and the profits
earned by finished-good manufacturers, In determining the system's
compatibilities.

68See Dei Ottati (1995); Bellandi and Trigilia (1991), n. 70, pp. 121-152; Becattini (2000b).
69Por an early assessment of production-phase businesses and their development after the
Second World War, see: Lazzeretti and Storai (1999).
70See Ciappei and Mazzetti (1996).
 From Marshall's to the Italian "Industrial Districts". A Brief Critical Reconstruction 99

The upshot of all this is that there must be a close, generally reproductive,
overlap between a block of economic and production relationships, occurring
within a spatially defined area, and another block, also spatially defined, of socio-
cultural relationships 71.
The next step, which is perhaps not yet fully understood, is the definition of the
synergies and antagonisms between the two engines of the Italian economy: large
and medium-size businesses in leadership positions and industrial districts 72 .
The research coordinated by Signorini that I mentioned earlier confirmed
independently and with completely different methods the concrete existence and
relevance of the relationships discussed above. It also went even further, agreeing
with a conclusion of the staunchest supporters of the district theory, who say that
the unit of research that is best suited to industrial economics is not the individual
firm, but a group of firms, possibly located within the same territory73, a statement
that clearly entails major theoretical implications.

3 VVork in Progress

3.1. From the Study of the District as Such to the Study of the Forms
of the Industrialization Process

This phase in the analysis of industrial districts opens at least two alternatives: we
either carry out an in-depth study of industrial districts as the basic unit of
research, with the goal of isolating its component parts (for example,
manufacturing system, institutional network, system of values, etc.) and seeing
how they are connected within the "local system", or move to the next level of
research, seeking those processes the combined operation of which determines
whether a certain territorial distribution of a manufacturing system is or is not
approaching the district form. I started by pursuing the first avenue 74 , but, seeing
that it led to a sort of frenetic "district hunt" in space and time, with a plenty of
"bastard" districts and an excess of taxonomic qUibbling 75 , I decided to attempt
dissecting the district phenomenon into several distinct processes that were
convergent at birth, parallel during their lifetime and divergent when the district
dissolved.
I identified the following processes: a) a gradual, targeted and self-contained
subdi vision of certain production processes and of other complementary and
ancillary processes; b) the formation and multiplication over time of dynamic
links between "complexes of specialized production skills" developed in specific

71These conclusions are articulated in: Becattini (1990).

72This topic started to be investigated by Bellandi (2001).
73See Signorini (2000), p. XXII.
74Th is phase of the research work conducted by our group is summarized in Pyke, Becattini
and Sengenberger (1990).
75See Becattini (1998).
100 G. Becattini

territorial settings and "nuclei of needs" that emerge from the general panorama of
needs 76 ; c) the sedimentation of social practices into formal and informal, tangible
and immaterial institutions that at the same time protect both competitiveness and
the social and naturalistic reproducibility of tht: local system; d) the dynamic
integration of contextual production expertise77 , which is often implied, with
codified scientific/technical expertise into the manufacturing process; e) the
development of figures and institutions (versatile integrators) that serve as
mediators between the contrasting needs of specialization and versatility; f) the
formation, consolidation, decay and dissolution of the "sense of belonging" among
the individual agents of social production; and, finally, g) constant support for
local social and professional mobility.

3.2. Inside Change: The Driving Force

The driving force in the development of district-type organizations is the trend,

triggered at any given point in time, towards a growing, focused and self-
contained subdivision 78 of a production process that is "typical" of a given cluster
of production processes into separate phases that are generally performed in
separate facilities by different enterprises. It is the Smithian-Marshallian-
Youngian principle 79 of the gradual division of labour as a wheel that, starting
from within the individual 80, moves the entire social world. However, one thing
should be quite clear: the notion of the division of labour is here not referred to a
single company, as in Smith's example of the pin factory, nor to the market in its
abstract totality, as in Young's model, but to an intermediate, meso-economic
entity that perhaps lacks a legally recognized status and aggregates and
disaggregates in its different manifestations - territorial and otherwise - in
response to overall long run changes in socio-economic relationships at both the
local and globallevel BI •
The continuous and growing "local articulation" of the manufacturing process
that is "typical" of an industrial district produces two types of effects. First of all,
it is a natural vehicle for increasing the productivity of labour within the district,

76 What I am alluding to is an attempt, perhaps still relatively raw, to formulate a theory

that could link in organic dualities nuclei of needs that are perceived as joined (for example,
clothing or, as part of it, footwear) but are constantly modi tied by disaggregations and
reaggregations of perceived needs on a global scale, with territory-based nuclei of
manufacturing expertise that have roots in the history and geography of the respective
territories.
77For a wider development of the theme, see Becattini and Rullani (1993), and Nonaka and
Takeuchi (1997).
78The most difficult, and as yet unsolved, problem is how to study the internal and external
forces that make self-containment possible.
79The key, and perhaps sole, reference point is found in Young (1928).
80Raffaelli's work, quoted earlier, is essential to understanding this connection.
81 See Bellandi (2001), pp. 189-210. But also, for different aspects, Solinas (1993).
 From Marshall's to the Italian "Industrial Districts". A Brief Critical Reconstruction 101

which is obtained by an intensive use of the existing know-how and the ability to
acquire new know-how, However, this characteristic is also the source of the
district's biggest problem: the need to find increasingly large outlets for its output.
Like capital, the industrial district either increases or perishes.
The increasing "articulation" of the main process is also important because it
permits an inter-penetration of the production process and of the daily life of the
reference community, which constitutes one of the most distinctive peculiarities of
the "district-type method of production". It is because of this peculiarity that the
"surplus" time, know-how and work energy that the local production process
generates "as it runs" tind within the same process new economic spaces and
relative competition gaps, which they can fill. By constantly evolving internally,
the production process of an expanding district creates the means of monetizing
many kinds of resources by transforming them into goods and services that can be
sold at a profit. At the same time, it embeds in the individual and collective know-
how, which can be put to new uses, the manufacturing and marketing experiences
that proved to be successful 82 ,
Viewed from this perspective, the increasing subdivision into phases, the
induction of collateral processes, the formation of open (but with a district bias)
teams of specialized businesses, the birth of markets within the cluster and of an
embryonic "local price system" are different aspects of the same process: the
gradual formation - concurrently with the unfolding of a "typical" multiphase
production process that involves a large number of autonomous, spatially
contiguous agents - of an organic plurality of local institutions centered on a grid
of local markets for semifinished goods (e.g., yarns, fabrics, etc.), new and used
equipment, and manufacturing processes (e.g., dyeing, finishing, etc.).
However, it would be impossible to understand the formation of this plexus of
internal markets, pivoting around a unique "district market" for human resources,
without also taking into account the coalescing around these markets of formal
(e.g" craftsmen's associations, unions, interest groups, technical schools,
institutions, the ultimate function of which is building in every potential
specialized contributor the confidence that he will be able to sell his specialty
easily and at a remunerative price. This local confidence-building process is just
the other side of the formation of a web of local and external markets.

3.3. Inside Change: Social Reproduction and District Governance

There is another dimension to the district mechanism that is often overlooked:

with the end of Fordism, efficiency and the ability to innovate and, consequently,
business competitiveness became increasingly dependent on the productive
performance, commitment and flexibility of the employees of those companies
that were part of the "team", For example, if a subcontractor feels bullied, he will
never perform beyond his contract minimum, and may even cheat a little, feeling

82See, e.g. Bellandi (1989), Dei Ottati (1994), Sforzi (2000).

102 G. Becattini

free to find new customers for whom to work whenever conditions allow it and
make it convenient.
In other words, an entrepreneur cannot ask his partners (workers and
subcontractors) to exert themselves excessively to help grow, or rescue, his
business when it is clear that at the first opportunity he would fire an employee
ancIJor change a supplier. In an "ideal-type district", the situation is different
because: a) the production cost of a "representative team" of firms depends
"intelligibly" on the overall functioning of the local society, which is affected to a
significant degree by the decisions (zoning, waste treatment facilities, highways,
etc.) of democratically elected officials; and b) both the "representative" worker
and subcontractor have a good understanding of the overall economic conjuncture
of the "district system". In other words, district workers and subcontractors are
well aware that they are not powerless vis-a.-vis their terminal businesses, whose
quest for cheaper labour is limited, in the environment outside the district, only by
competition. These conditions provide a basis for a genuine and realistic
negotiating process 83 . It is as if inside the heads of the representatives of the
different parties there was a common simplified model of how the cluster operates
that defines, for every degree of integration, the allowed margin of fluctuation for
the average wage and price of each phase operation. This margin is relatively
narrow and each of the parties will be careful not to destroy, together with the
opposing party, their golden opportunity (provided that this is their perception),
i.e., the opportunity to live in a "district-type local system" where a widespread
competitive attitude is mitigated by the presence of relationships based on
interfirm cooperation, good information and reciprocal dependence among
operators 84 .

3.4. Summing It Up

Going back to our starting point, we can say that Marshall, in his early works, was
right to question the compatibility of increasing returns, which in turn engender
cumulative processes, with an economic theory centered on a model of static
equilibrium. We can also say that Marshall was perhaps wrong when, later in life,
he failed to bring his basic intuitions to their ultimate conclusions 85 •
The overall and, obviously, remote goal of the approach to industrial
phenomena here suggested is to look for self-reproducing organizational forms of
the societal production process that are defined, in each case, by a combination of
processes as specified in 3.1. The industrial district is the self-reproducing form
that Italian economic reality has placed before our eyes, but nothing prevents us
from thinking that others may exist.

83Gabi Dei Ottati started to reseatch this issue in Dei Ottati (2001).
84This reminds much some expressions of Kindleberger's analysis.
85 The position of Marshall on the issue in 1906 can be seen in: Raffaelli (1996) n. 1. I
examine more extensively the issue in a forthcoming paper: Becattini (2001).
 From Marshall's to the Italian "Industrial Districts". A Brief Critical Reconstruction 103

In other words, the ultimate unit of analysis for the study of change is not
necessarily the industrial district, but an entity that changes with the overall
change, yet, up to a certain point, remains the same. If we pursue this path of
research to its conclusion, we will end up, I guess, not with a mosaic of markets
for homogeneous and clearly defined commodities, searching for irrelevant, or
impossible, equilibria, between supply and demand which change of "meaning"
with the qualitative change of the commodity, but with areas of convergence
(temporary but relatively stable) between economic and socio-cultural processes.
These are the meso-socioeconomic units that we need.

References

Bagella, M. and Becchetti, L. (2000), The Competitive Advantage oj Italian District -

Theoretical and Empirical Analyses, Berlin, Physica Verlag.
Becattini, G. (1978), "The Development of Light Industry in Tuscany: an Interpretation",
Economic Notes, n. 2-3, pp. 107-123.
Bccattini, G. (1987), Mercato e Jorze locali - if distretto industria Ie, Bologna, II Mulino.
Becattini, G. (1990), "The Marshallian Industrial District as a Socio-Economic Concept",
in: Industrial Districts and Inter-firm Cooperation in Italy, Pyke, F., Becattini, G. and
Sengenberger, W. (eds), Geneva, IlLS.
Becattini, G. (1994), "Nazione economica e nazione politica nel pensiero di Alfred
Marshall", in: Roggi, P. (cd.), Quale mercato per quale Europa, Milano, Franco Angeli.
Becattini, G. (1998), "Dal distretto industriale al processo di distrettualizzazione", in: Le vie
dell 'industrializzaziolle europea, Fontana, G. L. (ed.), Bologna, il Mulino. .
Bccattini, G. (1999), L'industrializzaziolle leggera della Toscana. Ricerca sui campo e
conJronto delle idee, Milano, Franco Angeli.
Bccattini, G. (2000a), "Anomalie marshallianc", Rivista Italiana degli Economisti, No. 1.
English translation: "Marshallian Anomalies", Marshall Studies Bulletin, 2000, n. 7, new
series, http://www.cce.unifi.itJdse/marshall/welcome.htm.
Becattini, G. (2000b), II bruco e la JarJaUa. Prato nel mondo che cambia (/954-1993),
Firenze, Le Monnier and The City of Prato.
Becattini, G. (2001), "Alfred Marshall fra analisi e storia econornica", Rivista di Storia
Ecollomica, Dicembre.
Becattini, G. and Menghinello, S. (1998), "Contributo e ruolo del made in Italy
"distrettuale" nelle esportazioni italiane", Sviluppo locale, vol. V, n. 9.
Becattini, G. and Rullani, E. (1993), "Sviluppo locale e mercato globale", Economia e
Politica industriale.
Becattini, G., Bellandi, M., Dei Ottati, G. and Sforzi, F. (2001), II caleidoscopio della
sviluppo locale. Tra~formazioni economiche nell'ltalia cOlllemporanea, Torino,
Rosenberg & Sellier.
Belfanti, C. M. and Maccabelli, T. (1999), Un Paradigma per i distretti industriali, Brescia,
Grafo (ed.).
Bellandi, M. (1982), "Il distretto industriale in A. Marshall", L'industria, n. 3.
Bellandi M. (1989), "Capacita innovativa diffusa e sisterni locali di imprese", in: Becattini,
G. (ed.), Modelli locali di sviluppo, Bologna, II Mulino, pp. 149-172.
Bellandi, M. and Russo, M. (1994), Distretti industriali e cambiamento economico locale,
Torino, Rosenberg and Sellier.
104 G. Becattini

Bellandi, M. and Trigilia, C. (1991), "Come cambia un distretto industriale. Strategie di

riaggiustamento e tecnologie informatiche nell'industria tessile pratese", Economia e
politica industriale, n.70, pp. 121-152.
Bellandi, M. (2001), "Local Development and Embedded Large Firms", Entrepreneurship
& Regional Development, 13.
Belussi, F. and Gottardi, G. (2000), Evolutionary Patterns of Local Industrial Systems.
Towards a Cognitive Approach to the Industrial District, Ashgate, Aldershot.
Belussi, F., Gottardi, G. and Rullani, E. (eds) (2002, forthcoming), The Net-Evolution of
Local Systems. Knowledge Creation, Collective Learning and Variety of Institutional
Arrangements, Kluwer, Dordrecht.
Brusco, S. (1986), "Small Firms and Industrial Districts: The Experience of Italy",
Economia Internazionale, Vol. XXXIX, n. 2-3-4.
Brusco, S. (1989), Piccole imprese e distretti industria Ii, Torino, Rosenberg and Sellier.
Brusco, S. and Paba, S. (1997), "Per una storia dei distretti industriali italiani dal secondo
dopoguerra agli anni novanta", in: Storia del capitalismo italiano dal dopoguerra ad
oggi, Barca, F. (ed.), Roma, Donzelli.
Cairnes, 1. E. (1874), Some Leading Principles of Political Economy, London.
Ciappei C. and Mazzetti G. (1996), Lo sviluppo produttivo a rete nel distretto tessile
pratese, Firenze, Regione Toscana.
Clapham, 1. H.( 1922a), "Of Empty Economic Boxes", The Economic Journal, sept.
Clapham, 1. H. (1922b), "The Economic Boxes: A Rejoinder", The Economic Journal, dec.
Cliffe Leslie, T. (1888), Essays in Political Economy, London, Longmans, lIed.
Conti, G. and Menghinello, S. (1996), "Territorio e competitivitii: l'importanza dei sistemi
locali per Ie esportazioni italiane di manufatti. Un'analisi per provincie (1985-94)",
Rapporto sui Commercio Estero, 1995, Roma, ICE, pp. 286-303.
Cooke Taylor, W. (1841), Notes of a Tour in the Manufacturing Districts of Lancashire,
London, Duncan and Malcolm.
Cora, G. and Rullani, E. (eds) (1998), Percorsi loeali d'internazionalizzazione. Competenze
e auto-organizzazione nei distretti industriali del Nord-Est, Milano, Franco Angeli.
Cossentino, F., Pyke, F. and Sengenberger, W. (1996), Local and Global Response to
Global Pressure: the Case of Italy and Its Industrial Districts, Geneva, IlLS.
Dardi. M. (1984), II giovane Marshall. Accumulazione e mercalO, Bologna, Il Mulino.
Dei Ottati, G. (1994), "Trust, Interlinking Transactions and Credit in the Industrial
District", Cambridge Journal of Economics, Vol. 18, n.6, dec., pp. 529-546.
Dei Ottati, G. (1995), Tra mercato e comunita: aspetti concettuali e ricerche empiriche sui
distretto industriale, Milano, Franco Angeli.
Dei Ottati, G. (2001), "Concertazione e sviluppo nei distretti industriali", in: Evoluzione del
lavoro, crisi del sindacato e sviluppo del paese, Ninni, A., Silva, F. and Vacca, S. (eds),
Milano, Franco Angeli, pp. 163-195.
Dobb, M. (1923), The entrepreneur myt, paper read in Cambridge Club The Heretics.
Fawcett, H. (1863), Manual of Political Economy, Cambridge, Macmillan.
Fortis, M. (1996), Crescita economica e specializzazioni produttive. Sistemi locali e
imprese del "made in Italy", Milano, Vita e Pensiero.
Fortis, M. (1998), II "made in Italy" - Quando stile e creativita non sono solo moda,
Bologna, Il Mulino.
Garofoli, G. (1983), Industrializzazione diffusa in Lombardia: sviluppo territoriale e
sistemi produttivi locali, 1st ed., Milano, Franco Angeli.
Georgescu Roegen, N. (1966), Analytical Economics: Issues and Problems, Cambridge,
Mass, Harvard University Press.
Gobbo, F. (1989), Distretti e sistemi produttivi alia soglia degli anni 90, Milano, Franco
Angeli.
 From Marshall's to the Italian "Industrial Districts". A Brief Critical Reconstruction 105

Groenewegen. P.(l995). A Soaring Eagle. Alfred Marshall, 1842-1924, Brookfield, Edward

Elgar.
Hearn, W. E. (\863), Pluto logy, or the Theory of the Efforts to Satisfy Human Wants,
Melbourne. G. Robertson and Co.
ILO (\ 985), The New Industrial Organization, coordinated by W. Sengenberger, with the
initial support of M. 1. Piore.
IRPET (\969), "Lo sviluppo economico della Toscana: un'ipotesi di lavoro". II Ponte, No.
II.
ISTAT (1996), Rapporto Annuale. La situazione del paese nel 1995, Roma, IS TAT, pp.
262-271.
ISTAT (\997), 1 sistemi loculi dellavoro 1991, Roma, ISTAT, Argomenti n. 10.
Jevons, S. (1871) The Theory of Political Economy, London, Macmillan.
Kindleberger, C. P. (1989), Economic Laws and Economic History, Cambridge, Cambridge
University Press.
Lange, O. (1952), Price Flexibility and Employment, Chicago, Cowles Commission.
Lazzeretti, L. and Storai, D. (1999), II distretto come comunita di popolazioni
orgunizzative. II caso Prato, Prato, Quaderni dell' IRIS, n. 6.
Le Play Les Ouvriers Europeens.
Loasby, B. 1. (1989), The Mind and the Method of the Economist, Aldershot, Edward Elgar.
Loasby. B. 1. (1998), "Industrial Districts as Knowledge Communities", in: Industry, Space
and Competition. The Contributions of Economists of the Past, Bellet, M. and I'Harmet,
C. (cds), Cheltentham, Edward Elgar.
Marshall, A. (1919), Industry and Trade, A study of illdllStrial technique and business
organization; and of their influences on the conditions of various classes and nations,
London, Macmillan, First Edition.
Marshall, A. (1961), Principles of Economics, Variorum edition overseen by C. Guillebaud,
London, Macmillan.
Marshall, A. and Marshall, M. P. (\ 881), The Economics of Industry, London, Macmillan,
2"d cd.
Marshall, A. Ye Machine, early manuscript published and edited in Raffaelli, T. 1994.
Mill, 1. S. (1848), The Principles of Political Economy, I SI cd.
Nonaka, I. and Takeuchi, H. (1997), The Knowledge Creating Company, Milano, Guerini e
Associati.
Nuti, F. (\992), I distretti dell'industria manifatturiera in Italia, 2 voll., Milano, Franco
Angeli.
Onida, F., Viesti, G. and Falzoni, A. M. (\ 992), I distretti industriali, crisi 0 evoluzione?,
Milano, Egea.
Parri, L. (1997), "Risultati di azione umana ma non di progetto umano. I distretti industriali
per Hayek e la Scuola austriaca", in: Un Paradigma per i distretti industriali, Belfanti, C.
M. and Maccabelli, T. (eds), Brescia, Grafo, pp. 175-190.
Pesciarelli, E. (\ 999), "W.E. Hearn on the industrial organization of society", in: Industry,
Space and Competition, Bellet, M. and I'Harmet, C. (eds), Cheltentham, Edward Elgar.
Pigou, A. C. (1925), "Where to House the London Poor", in: Memorials of Alfred Marshall,
Macmillan, London. pp. 149-150.
Pigou, A. C. (1925), Memorials of Alfred Marshall, London, Macmillan.
Pigou, A. C. (1914), Wealth and Welfare.
Porter, M. (1989), The Competitive Advantage of Nations, New York, Macmillan.
Porter, M. (1999), On Competition, Boston, Harvard Business School Press.
Pyke, F. and Sengenberger, W. (1992), Industrial Districts and Local Economic
Regeneration, Geneve, IlLS.
106 G. Becattini

Pyke, F., Becattini, G. and Scngenberger, W. (1990), Industrial Districts and Inter-firm
Cooperation in Italy, Geneva, IlLS.
Pyke, F., Becattini, G. and Sengenberger, W. (1991), Distretti industriali e cooperazion€
fra imprese in Italia, Firenze, Banca Toscana.
Raffaelli, T. (1994), "Alfred Marshall's Early Philosophical Writings", in: Research in the
History of Economic Thought and Methodology, Samuels, W. (ed.) London, JAI Press.
Raffaelli, T. (1994), "Alfred Marshall's Early Philosophical Writings", in: Research in the
History of Economic Thought and Methodology, Samuels, W. (ed.), London, JAI Press.
forthcoming (Routledge, 2002) book on Marshall
Raffaelli, T. (1996), "La formazione di Keynes nel laboratorio marshalliano", Rivista
ltaliana degli Economisti, n. 1.
Raffaelli, T. (forthcoming), II concetto di distretto industriale nel sistema teorico di Alfred
Marshall, Sviluppo Locale.
Sargant, W. L (1857), Economy of the Labouring Classes, London, Simpkin, Marshall and
Co.
Sforzi, F.(1989), "The Geography of Industrial Districts in Italy", in: Small Firms and
Industrial Districts in Italy, Goodman, E.and Bamford, 1. with Saynor, P. (eds), London,
Routledge, pp. 153-173.
Sforzi, F.(1994), "The Tuscan Model: an Interpretation in Light of Recent Trends", in:
Regional Development in a Modem European Economy: The Case of Tuscany, Leonardi,
Rand Nanetti, R. Y. (eds), London, Pinter.
Sforzi, F. (2000), "Local Development in the Experience of Italian Industrial Districts", in:
Geographies of Diversity. Italian Perspectives, Cori, B. et aL (eds), Roma, CNR, pp.
133-156.
Signorini, L F. (2000), Lo sviluppo locale. Un 'indagine della Banca d'Italia sui distretti
industriali, Roma, Meridiana Libri.
Solinas, G. (1993), "Competenze, grandi imprese e distretti industriali. II caso Magneti
Marelli", Rivista di Storia Economica, n. I, pp. 79-111.
Sraffa, P. (1998), "On the Relations Between Cost and Quantity Produced", in: Italian
Economic Papers, Pasinetti, L L (cd.), voL III, Bologna, II Mulino.
Steedman, J. (1986), 'Trade Interest versus Class Interest", EL'onomia Politica, August.
Tattara, G, (cd.) (2001), II piccolo che nasce dal grande. Le molteplici facce dei distretti
industriali veneti, Milano, Franco Angeli,
Varaldo, R. and Ferrucci, L (eds) (1997), II distretto industriale tra logiche di impresa e
logiche di sistema, Milano, Franco Angeli.
Whitaker, J. (1975), "Some Features of American Industry" in: Early Economic Writings of
Alfred Marshall, London and New York, Macmillan, 2 1lu voL, pp. 355-376
Whitaker, 1. (1975), Early Ecomonic Writings of Alfred Marshall, London, Macmillan, 2
vols.
Young, A. (1928), "Increasing Returns and Economic Progress", Economic Journal,
pp.527-542.
Diversity, Not Specialization:
The Ties That Bind the (New) Industrial District

Charles F. Sabel

Abstract. Whatever they are exactly, industrial districts are also a worldly success
and a conceptual innovation. In Italy, the map of district agglomerations of small
and medium-sized firms specializing in particular branches of light industry is no
longer limited to the triangle marked by Venice, Florence, and Ancona in the
Adriatic Marches. New agglomerations are spreading, as though by contagion,
down the shores of the Adriatic and the Mediterranean, now and again obstructed
in their march south by the detritus of failed, heavy industrial complexes. Quite
apart from this expansion of the districts' homeland, district-like agglomerations
are emerging in some patches of the South of Italy. Looking at the economic map
of Europe, the economic agronomist sees a "hot banana" of thriving districts
curving from London, through Switzerland and the Southwest of Germany into
Northern Italy. In the developing economies districts are identified as a motor of
growth in countries as different as Chile, Brazil, and India.
The advance of the industrial district as a concept has at least kept pace with its
progress in fact. Discussion of economic growth in both the advanced and
developing countries is dominated by what is often called the Washington
consensus - concern for opening economies to free trade, getting exchange rates
and other prices right, and (lately) building the institutions needed to do this. But
insofar as it is not, debate gravitates toward the creation and expansion of
"clusters" - the business name for districts. In the European Union in particular
fostering clusters is often seen as a way of encouraging economic competitiveness
without giving in to US pressure to give markets free reign. In theoretical debates
about economic growth and international trade too agglomerations, often
explicitly identified with districts, have come to playa central role (although. as
we shall see, with surprisingly subversive effects).

Why. then, are the districts prospering and diffusing? What explains their success
in today's volatile and rapidly changing markets'? Surprisingly - or not, you be the
judge - there is no compelling answer to such brusque and elementary questions.
Not that familiarity with the subject has dulled curiosity about its causes. Rather, I
suspect the fruitfulness of the district idea as a research agenda and as a policy
tool has dulled incentives to inquire too closely into the consistency and
generalizability of the sustaining ideas. Concepts that both allow exploration of
economic forms uncognizable in the light of standard theories of the firm and
108 C. F. Sabel

discussion of public engagement in economic development otherwise taboo seem

somehow self justifying if not self evident. Put another way, you don't look a gift
horse in the mouth, especially when you are bestride it.
The result is a growing gap between a venerable and originally plausible
explanation of district success and the sources of competitive flexibility generally,
and a discrepant mass of material regarding developments in the world, research
findings and alternative views of innovation. The familiar explanation focuses on
the specialization of activity and the tacit or lived quality of the knowledge that
such specialization produces. The discrepant material, on the contrary, links the
innovative capacities of agglomerations to diversity and the (partial) formalization
of the knowledge that coordination across difference demands. The districts initial
success was mysterious. How could small, traditional artisan firms outperform
giant, technologically sophisticated corporations in the application of new
technologies in industries as diverse automotive components and textiles? Now
the deep changes wrought by the districts' capacities for transformation make the
grounds of their continuing success mysterious again.
In this essay I want to argue for the discrepant view. More exactly, I want to
argue that under current conditions innovation, and problem solving generally,
depend on disciplined comparisons of alternative solutions, and these in turn
require transforming tacit knowledge into what might be called pidgin
formalizatIOns: accounts sufficiently detailed to be recognizable to those who
know the situations to which they refer first hand, but sufficiently abstracted from
them to accessible to outsiders, from various disciplines.
Such pidgins are more than mere descriptions yet less than fully developed
causal theories. As the work of Galison (1997) shows, they are familiar to the
history of science as the working languages of collaborations between, say,
experimentalist instrument makers or engineers on the one hand and theoreticians
on the other. From these collaborations radically innovative theories sometimes
emerge. For our purposes, though, these formalizations are less interesting for
their generative potential than for the working exchanges they enable. Their
availability, we wiII see, breaks down or at least attenuates the distinction between
insiders and outsiders, intimates and strangers upon which the success of districts,
and related organizational forms, was premised. Conversely, organizations
designed to gener"te such formalizations, and collaborate with others doing the
same, have internal structures and build relations not contemplated in the familiar
district literature or in its analogue in the discussion of large firms. These novel
organizations are neither craft communities that organize themselves on the basis
of common, tacit skills nor hierarchies tied together by tacit routines. Their
distinctive feature, the one from which flows their peculiar openness to change
within and reconnection to the world without, is precisely the ability to question
their routines and the foundations of their common skills without undermining the
utility of either. They represent, we will see in the end, a form of self-organization
that engages the human capacities for self-reflection and deliberation without
supposing that we are endowed with anything like a panoramic understanding of
our situation.
 Diversity, Not Specialization: The Ties That Bind the (New) Industrial District 109

To intercept misunderstanding at the outset let me underscore two important

limits to the argument. First, in emphasizing diversity and formalization rather
than specialization and taciturnity as conditions of problem solving I am not
arguing that things have always been so. The traditional view of the districts is
still a good account - the best we are likely to have - of how districts operated
traditionally (until roughly the early 1990s). The discrepant material, and the
reading given here of it, is less concerned with revising history than taking
account of new circumstances. Second, however great the changes entailed in
accommodating the changed circumstances there is no compelling reason to think
that districts, starting from their traditional constitution, will be unable to make the
adjustment, or will be changed beyond recognition if they do. Indeed, on the
current evidence "new" districts agglomerating firms and other institutions
capable of jointly formulating pidgins as and so that they can collaborate are often
arising from the stepwise transformation of "old" ones. .
In presenting these views I am, some readers will know, speaking out of both
sides of my mouth. I was a partisan of something akin to the "traditional" view of
the districts when it seemed to many outlandish. I'm reasonably sure that the
concept of flexible specialization that Michael Piore and I (1984) coined to
capture the innovative responsiveness of districts apparently bound by craft
tradition is ambiguous enough to paper over the differences explored here (see
also Sabel and Zeitlin 1997). But nothing is served by this kind of evasion. Self-
ret1ection, and the distance on past selves it requires, is by definition not just for
the others.
The conception of innovation and t1exibility founded on tacit specialization finds a
natural expression in the idea of a craft or artisan economy. In such a world
problem solving is associated with particular tools and materials. Through
apprenticeship the use of the materials and instruments characteristic of a
particular trade become second nature. So too apprenticeship teaches that the
exercise of individual prowess draws on, and replenishes a fund of community
knowledge. Mastery comes with further experience, as the artisan ga;ns the
autonomy to apply the familiar techniques to the solution of unfamiliar, even
novel problems.
To connect this form learning by socialization to an economic structure, take the
simplest case, where each artisan in a given trade owns and operates a single lathe
or loom. Together the ensemble of owner-operators command the skills and
machines needed to whatever specialized good the market for their class of
product demands. A broker (who may also be an artisan) solicits customers and
contracts with the appropriate group of skilled producers to deliver the goods.
Since producers and brokers can diversify the risk of any transaction by pursuing
other projects with different partners simultaneously or in sequence, the artisans
and brokers can jointly respond to highly volatile markets while safeguarding or
advancing their separate interests. Given the manifest advantages of cooperation,
moreover, the artisans can easily agree to create institutions - public or private - to
provide themselves with services (bookkeeping, monitoring of technical standards,
quality assurance) that none, given the small scale of operations, can provide for
him or herself (Brusco 1989).
110 C. F. Sabel

Notice that this craft version of the district economy is largely self-governing in
the sense of (almost) proof against opportunism or self-seeking guile, and for
related reasons expansively self-reinforcing. The governance of opportunism is a
limited problem because the artisans' skills are largely complementary.
Homogeneous viewed from the outside, they are highly differentiated in their own
view, with micro-specializations - the use of a particular kind of loom or lathe -
that make most of their colleagues into potential partners rather than rivals. To be
sure, even partners are rivals when it comes to the distribution of the gains to
cooperation; but such rivalry is disciplined in explicit partnerships and in the craft
districts by the recognition of the mutual dependence. Socialization into the
community of craft knowledge, and the sense of dignified competition for the
respect of ones' peers that goes with it, echoes the conclusions of prudent
calculation. So it proves relatively easy to establish institutions to police the wage
cutting and substitution of shoddy materials that, in hard times, tempt strapped
producers to defend their individual livelihoods at the price of jeopardizing the
stability and reputation of the district ensemble.
This craft model of the district is naturally expansive: given that skills are fixed
in a community and a community in a place, the more specialized the district, the
more it attracts similarly specialized resources. Alfred Marshall is famous today
not least for having seen this connection. In craft communities, he noted, looking
to the metalworking agglomeration of the Birmingham of his day, the "mysteries"
of a trade are "in the air". Youngsters learn them before they are fully conscious of
learning at all, let alone investing in their human capital. Once acquired, the skills
become a local resource, attracting outsiders who need in some sense to
incorporate themselves into the community to benefit from what it can do only
where it is. Today we would say there are positive returns to specialization.
Marshall put it this way:
When an industry has thus chosen a location for itself, it is likely to stay
there long: so great are the advantages which people following the same
skilled trade get from near neighbourhood to one another .... A localized
industry gains a great advantage from the fact that it offers a constant
market for skill .... Employers are apt to resort to any place where they are
likely to find a good choice of workers with the special skill which they
require; while men seeking employment naturally go to places where there
are many employers who need such skills as theirs and where therefore it is
likely [they will] find a good market (Marshall cited in Fujita 2000, p.lO.
See also of course Becattini 1990).
The craft model of the small-firm district connects to large-firm settings by two
routes; and it is useful to consider these briefly to see the full reach of the
assumptions of specialization and taciturnity on current discussion of innovation
and flexibility. The more direct, but also more parochial connection is simply that
in some cases large firms are an assembly of craft shops under one corporate or
physical roof. This is notably so in Germany, where large firms did not specialize
in the mass production of standard consumer goods (cigarettes, packaged cereal,
cars). Instead they traditionally manufacture customized producers or investment
 Diversity, Not Specialization: The Ties That Bind the (New) Industrial District I II

goods: for example, the complex machine tools that mass producers use in their
own factories to turn out long runs of a single make or model of product. Workers
are largely autonomous in the same way, and by virtue of a similar (though more
formal) experience of apprenticeship as artisans in their small firms. Because the
firm is large, it can provide itself with many of the support services provided by
the institutional exoskeleton in the craft district. But familiar collective action
problems - especially fear that trained workers will be poached away from firms
where they apprenticed by firms that provide no training at all - require that here,
too, the community character of specialized skill is formally acknowledged and
protected against tragedies of the commons.
The second connection to large firms is via the school of evolutionary or neo-
Schumpeterian economics (for the fans et origo see Schumpeter 1926). It is less
obvious but broader in reach than the first, and shows that the idea of specialized,
tacit knowledge need not be associated with the notion of craft in any familiar
sense at all. One key assumption of this school is that large firms are configured
not as (nearly) self-organizing communities of producers, but rather as hierarchies.
The peak of the hierarchy fixes a master plan, including rules for partitioning that
plan into executable tasks. Successive tiers of superiors then actually decompose
the tasks into finer and finer jobs for their respective subordinates. This is
specialization by task. Because it is disconnected from any systematic exploration
of tools or materials it is disconnected from, even antithetic to specialization by
craft. A second assumption is that the partitioning of the rules by which complex
projects are subdivided and the rules by which the outputs of the partitioned tasks
are aggregated into useful wholes are never comprehensive and unambiguous
enough to cover all contingencies. The written rules therefore have to be
supplemented and corrected by informal, lived experience of how to actually get
things done in the organization. This is tacit knowledge. But it is tacit knowledge
at least as much of process and organization as of substance: of the routines that
define a firm and enable it to execute as a collective tasks that employees could
not manage if they needed to think both about their work and how to link it with
the efforts of others.
Unlike the craft model district, the tacitly routinized large firm faces substantial
governance problems and is not inherently expansive. The governance problem
arises from the fact that the firm's assets, physical and human, are specialized to
work most efficiently with each other as provided in the master plan. Their value
in alternative arrangement - in conjunction that is, with assets not specialized
according to the same plan - is close to zero. Hence the hold up problem: suppose
a power-plant owner and a coal-mine owner agree respectively to build a power
plant and dig a coal mine that are co-located and otherwise specialized to each
other. Whoever invests first is at the mercy of the other, who can withhold the
complementary investment until the dupe agrees to redistribute the benefits of the
venture in favour of the knave. The solution of course is for the one owner to
make both investments (with the one better positioned to use the whole effectively
taking control). This vertical integration formally closes the large firm in on itself
in a way that contrasts with the - formal - openness of the craft districts.
112 C. F. Sabel

With such closure comes another kind of self-limitation. All the crucial aspects
of large-firm operation have to be under exclusive control of a single owner, and
the exercise of control becomes more expensive as the complexity of what is
controlled increases. Firms therefore do not benefit from the positive returns to
agglomeration in the way that "centreless" districts do. Or rather they do no
benefit from such returns as a matter of course. Economies of scale - the larger the
production run, the lower the unit costs of production - favour such expansion; the
costs of extending control to every more complex situations cut against it. (We
are, many readers will recognize, in a Coasian world (Coase 1937) where the
decision to extend the organization, and how, or rely on the market, depends on a
comparison of the relative costs of executing the marginal transaction in each of
these ways).
But beyond these differences the craft district and routine large firm views are
fundamentally similar in the emphasis on tacit knowledge as both a precondition
but also limit to innovation and flexibility. Because craft skill and corporate
routine are taken for granted and second nature, they allow for "spontaneous" or
self organized co-ordination within certain boundaries. For the same reason, they
make it hard to identify those boundaries, and harder still to say when they are
obstacles to problem solving, and hence to be corrected.
The neo-Schumpeterians are especially clear on this point. Given the grip of
routine, large firms can only innovate - break free of past assumptions to aspire to
"autonomous strategic behaviour" - if they can establish "breakout" structures:
"entities separate from current operations which" are used to "incubate new
projects" (Teece 1998, p. 152). This recognition that the large firm can only be
reformed from without is tantamount to an admission that it can not· be
transformed from within. A congruent view is current in network analyses of
economic sociology (Grano vetter 1973). Ties among neighbours - friends who are
themselves likely to be friends-are robustly self-reinforcing. They are the social
substratum of the reciprocity that fosters trust and tacit knowledge. These ties do
the work of coordination, but, again, at the price of closing the networked
organization off from the world. The information about the outside on which
innovation ultimately depends comes from sparse, "weak" ties to distant actors:
the cousin of a distant friend from my home village who tells me where to look for
a job in the city to which we have both migrated. These ties convey formal
knowledge, but neither its tacit complement nor the propensity to trust from which
collaboration grows.
Accordingly, boundary blurring, "breakout" structures aside, innovation for
these schools of thought occurs largely outside the firm, in university or
government research labs or "distant" companies. Different from country to
country, these constellations form what the neo-Schumpeterians call the "national
innovation system". This system is to the large firms what the exoskeleton of
service providers is to craft firms in the districts: the external, innovative
corrective to the routines of self-coordination within the firms themselves.
Explicitly in the case of the neo-Schumpeterian large firm, implicitly in the case of
the craft district, the assumption is that competition selects for those firms or
bundles of routines that make the most effective use internal and external
 Diversity, Not Specialization: The Ties That Bind the (New) Industrial District 113

resources. Economies of large or small firms survive in the long run in part
because their constituent "organisms" are sufficiently diverse in the genetic
material of their tacit knowledge to accommodate even the broadest shifts in the
conditions of competition. But they survive in part tno because the "we" of society
is capable of renovating the "external" institutions of the district or the national
economy as circumstances warrant. If these routine-revising institutions were
themselves prisoner of their own incorrigible routines, only blind luck would save
us from tragic attachments to solutions that no longer serve.
So in the end, and in the large - at the level of society at various scales - some
kind of reflection on and amendment of routine is possible and necessary. But
only outside the economy, in "society"? Or at least the firm? Is it really plausible
that large firms can devise "breakout" structures and small firms district
institutions, yet neither is capable of continuing self-revision in the small of its on-
going activities? Surely the existence of these structures and institutions attests
awareness of the problem of routine entrapment. Is that awareness fully exhausted
in the split solution - routine firm, routine-jiggling frame - these theories take for
granted? Or is this assumption itself an example of a (scientific) routine that
impedes recognition of a novel solution? The evidence presented next strongly
suggests the latter.
It is common, though perhaps not yet commonplace, to observe in the business
and business-school, engineering, and organizational sociology literatures that
firms are today able to adjust their operating routines and organizational
structures, and that disciplined comparisons of alternatives is an indispensable aid
to this. Most of the organizational innovations that permit this were pioneered in
modern times by Japanese firms, but are no longer limited, or indeed most
effectively developed by them. Here, sidestepping any considerations of how these
changes came about, I present an eclectic review of the innovations to highlight
the way the formalization of difference is key to the revision of routine. We start
with the operating units of the economy: first large, then small, district firms. The
distinction can be tenuous given that the latter are often suppliers to and must
mesh with the structures of the former, and that decentralization on the one side
and centralization on the other produces another kind of convergence. But the
discussions we have been canvassing draw this line, and we respect it here if only
to show that it is irrelevant to the innovations of interest. Finally we look at some
recent work on economic growth and geography which, innocent of any concern
for the structure of firms, comes to a complementary finding regarding the
contribution of diverse settings, rather than (relatively) homogenous and
specialized ones, to innovation.
Developments in large firms are reflected in the explosive diffusion of standards
that create and signal the existence of the preconditions for the creation of formal
pidgins or lingua francas for collaboration across firms and areas of expertise. On
these are built in turn disciplines that specify processes for comparing alternative
solutions to design problem or trouble shooting existing production set-ups
(Helper, MacDuffie and Sabel 2000). The most common of the standards is the
ISO 9000 series maintained by the International Bureau of Standards in Geneva. It
requires simply that firms be able to demonstrate that they can say what they do
114 C. F. Sabel

and do what they say with whatever level of resolution is relevant to their
industry. Put another way, the standards require that a firm be able to demonstrate
that it can make its "tacit" routines for, say, registering an order, or designing a
product, or correcting a manufacturing defect explicit to a potential customer.
Certification by a third party under the standard does not, of course, ensure that
any actual customer will find the explication sufficiently detailed and reliable for
its own collaborative needs. But certification is widely regarded in the automobile,
chemical, semi-conductor and computer industries as a precondition of a more
demanding examination of a firm's capacity for making its practices transparent
enough to be discussed with outsiders. And this is surely indicative of shifting
assumptions regarding the "stickiness" or embeddedness of (once) firm-specific
knowledge (von Hippel 1998).
The comparative disciplines, having arisen in practice, not under official aegis,
do not have standard names. In design the relevant practice is often called
concurrent or simultaneous engineering. A new product is defined initially and
provisionally by benchmarking (itself a comparative discipline) a current model
against successful competitors, relevant analogues in other fields, and adjusting
these comparisons to take account of laboratory innovations likely to be ready for
market in the next product cycle. Then the provisional design is decomposed into
modules (engine, transmission, seats, instrument panel, heating, ventilation and air
conditions system, and so on in the case of automotive vehicles) and each module
is then presented to the scrutiny of one or more of the corresponding module
makers. These module makers then benchmark their systems and suggest
refinements in the design, including if necessary changes in the performance
specifications of closely connected modules. Once the lead design group adjusts
the overall plan and selects module makers from those competing, the process is
iterated again, with the chosen module makers decomposing their systems into
sub-modules, and revising their plans in the light of their suppliers' suggestions.
This method both shortens design times and increases the reliability of designs,
while of course increasing the possibilities for incorporating innovations. This is
counterintuitive. It seems reasonable, after all, that successive refinement of a
single design will be either quicker, or more reliable than repeatedly correcting a
vague initial idea in the light of comparisons of variants, or both. Such in any case
was the view in the vertically integrated firm, where the design process was
typically seen as a mirror of the process by which the company partitioned itself
hierarchically. Concurrent engineering succeeds nonetheless in overcoming an
apparently inevitable trade-off because careful comparison of alternatives reveals
the defects and virtues of each variant, allowing for quicker, more reliable, and
more inclusive choices (Actually the comparisons involve a kind of double
diversification: different projects are compared by design teams whose members,
drawn from different specializations, are cognitively diversified. The underlying
mechanism of instigating surprise through investigation of difference remains the
same).
Similar mechanisms are at work in new regimes for error detection and
correction. An example is the five "why": Why is machine a broken? No
preventive maintenance was performed. Why was the maintenance crew derelict?
 Diversity, Not Specialization: The Ties That Rind the (New) Industrial District liS

It is always repairing machine B. Why is machine B always broken? The part it

machines always jams. Why does the jam recur? The part warps from heat stress.
Why does the part overheat? A design flaw.
Thus error-detection and correction, like benchmarking and simultaneous
engineering, discovers unexpected (mis-) connections among the parts of complex
endeavours. The cumulative effect of these results is captured in improvements in
the benchmark standards for various production processes.
Apparently modest, even commonsensical institutions such as these embody a
deep innovation in the cognitive capacities of (large) organizations. For
benchmarking, simultaneous engineering and error-detection methods like the five
"why" are procedures for doing just what the tacit-knowledge view of specialized
action says cannot be done: routinely questioning the suitability of current
routines. Whether in the initial specification of new designs (benchmarking), the
concretisation of these approximations (simultaneous engineering), or in the
course of their practical application (error detection), this disciplined inquiry of
routines occurs at just those times when self-interrogation seems most valuable
but most difficult.
These mechanisms oblige the actors to search for solutions in a circumscribed
space of possibilities (the set of best current or potential designs, the activity
chains that might have caused a particular breakdown) whose exact contours and
contents they could not have anticipated. The outcome of the search is thus likely
to be sufficiently unfamiliar and disconcerting to force re-evaluation of habitual
responses. The new large firm is thus a member of a new class of institutions
defined not by the fixed routines to which they are oblivious, but rather by the
routines they use for interrogating and altering their routines. Think of the new
institutions as pragmatist: they systematically provoke doubt, in the pragmatist
sense of an urgent suspicion that habitual beliefs are poor guides to current
problems. They create the pidgins that make collaboration between insiders and
outsiders, neighbours and distant acquaintances, and specialists or professionals of
radically different kinds possible (for a thoughtful account of the relation between
neo-Schumpeterian theory and the new disciplines lhat dovetails with the
discussion here, see Winter 2000, pp. 56-58).
The new pragmatist standards and disciplines suppose and foster a distinctive
form of industrial organization. In contrast to the tacitly routinized large firm, the
pragmatist corporation is federated, not hierarchical and centralized: decisions of
higher-level entities are crucially shaped by the decisions of their constituent units.
The federation is open, not vertically integrated: components or services crucial to
the final product of one firm can be provided by independent companies, and the
firm's internal specialized producers can provide outsiders with crucial inputs.
These differences repose the question of governance; and the resulting answers
prompt reconsider of, among other things, the distinction between weak and
strong ties. We return these matters after completing the eclectic tour of discrepant
material.
Consider next the coordinate changes in the organization of districts and district
firms. Some of these changes are of course so directly tied to the pragmatist
reorganization of the large firm as to count as part and parcel of the latter, not
116 C. F. Sabel

independent developments. When large firms begin to collaborate more explicitly

and continuously in design and production with their suppliers, district firms, as
privileged suppliers, must respond in kind or lose a key market. It is not, therefore,
surprising to see the new disciplines diffuse within districts. But (hereby probably
revealing an irresponsible ignorance) I am unaware of broad-gauge studies that
trace the higher order effects of these adjustments on the structure of individual
district firms and the relations among them and their large customers.
Similar changers emerging from within the districts - endogenously, if you like -
are more revealing. Here I am guided by the experience of a two-week stay in the
textile district of Prato in 1996. The salient development was the drive to
pragmatist formalization of design and production, and the search for adequate
institutional structures within which to house the emergent processes, in response
to the sheer complexity and pace of product change. Traditionally the spinner,
weaver, and finisher were assigned tasks by and communicated largely through
the broker who, we saw, marketed the project in which they participated. But
when fabrics consist of novel combinations of novel yarns, each under torsion and
woven into a structure that precisely balances torque against torque, costly
missteps are likely in the absence of richer, more direct communication among the
producers themselves. One upshot was the proliferation of explicit handling
instructions: booklets that describe how a yarn or fabric is constructed, and how it
is to be further worked. Another was the spread of working relations among
producers, bypassing the brokers. In some cases the finishers, who transform the
look and feel of the fabric by sophisticated chemical and mechanical treatments,
were using their position at the end of the production process and their expertise in
the preceding steps to become brokers of a new type. They were both organizing
more direct relations among the other collaborators and managing their relations,
and those of the "group" around them to the final customer. But there was no
sense at the time that the changes were coalescing into a new district structure, let
alone a new model district. Indeed the social dislocations that went hand in hand
with even piecemeal adjustment - especially the pending displacement of brokers
and rise of finishers - was undermining the social order underpinning the district's
ability to formulate and decide questions of common concern.
A large body of recent Italian writings on the evolution of districts comes to
findings that converge with those of this vignette. The common ground in these
debates is the observation of large, "lead" firms in districts:
Competition from low wage countries and large firm restructuring are
forcing the industrial districts to compete almost exclusively in quality-
conscious markets. At the same time, changes in demand composition, a
need for higher intrinsic production quality, shorter delivery times and a
more active product commercialisation push firms to specialize in their
core businesses and develop stable relationships with a limited number of
suppliers. Groups are becoming common, and at least in some districts
there is an increase in external hierarchy. The adequacy of the collective
and local governmental institutions so important in the districts' past is
 Diversity, Not Specialization: The Ties That Bind the (New) Industrial District 117

being challenged (see Whitford 2001, p, 27, e-copy of manuscript, and

generally for a comprehensive review),
The role of these firms is to formalize relations with the market and increase
(the self-) control of subcontractors, One side in the discussion - the aziendalisti -
argues that these changes lead inevitably to hierarchy, and to the demise of the
district community, The distrettisti counter the new forms of coordination should
be seen as jointly managed or bilateral governance structures not hierarchies (Dei
Ottati 1996, p, 58), Hence even
a district with coordinating lead firms is still a system and often is not
dependent only on those lead firms. It may still need the reproduction of the
"old" values and informal mechanisms of coordination, and there may be
some collective goods that lead firms and their networks can create only
with difficulty (Whitford 2001, p.23).
The view developed here agrees with the statements of the facts but is at odds
with both interpretations: it suggests that the "new" district is neither a mere
assembly of hierarchical firms and their satellites, nor a system of firms having
their true foundation in a local community. Pragmatist governance is in theory
both non-hierarchical Uoint) and formalizable. My hunch is that a study of
districts that took seriously the possibility of an alternative to both the traditional
district and the traditional large firm would allow us to see quickly enough if it is
worth pursuing the claim that the Third Italy is going pragmatist.
The third body of discrepant evidence, from debates in international trade,
growth theory and economic geography, bears directly on the value of diversity as
against specialization as a source of innovation, but has little to say about the
relation of this dispute to firm organization. At the origin of much of this
discussion was a puzzle about the pattern of post-War international trade. The
theory of comparative advantage predicts that a country richly endowed with
capital relative to its endowment of labour will export capital intensive-goods to
and import labour-intensive goods from a country whose relative endowments are
the reverse, regardless of the absolute costs of production in either country. But in
the post-War period much of the growth in trade was between countries with
similar, not complementary factor endowment: the US exported cars and aircraft
to Europe, and bought back (different kinds of) cars and aircraft. A plausible
explanation for this intra-industry trade was to see it as the result of positive
returns to the mass production of different types of highly specialized products:
The US churned out big cars, the Europeans made Beetles or their equivalent. The
first mover in each case acquired, because of the increasing returns to scale, a cost
advantage that prompted further specialization in the same line. As it became clear
that knowledge is a key to economic growth, and that knowledge begets
knowledge, positive returns and its relation to specialization have become research
topics of broad concern (see especially Helpman and Krugman 1985).
One branch of this discussion has taken a surprising turn that bears on the
argument here. In economic geography Puga and others are finding that diversity,
not specialization promotes innovation. A locale is said to be specialized in this
118 C. F. Sabel

context if, relative to the national distribution of economic activity, it concentrates

in an industry or a broad sub-sector of one - in the way Biella specializes in
woollen textiles. It is diverse or diversified if it is disproportionately specialized in
several different activities simultaneously. Suggestive findings in this research
area include these: in their formative and innovative phase, firms locate in cities
with diverse economies. Once they have "innovated" - settled on a product line or
process - they move to (district-like) locales specialized in the activities they will
continue to pursue (Duranton and Puga 1999). Firms adopting new processes,
such as programmable automation of metal cutting tools, are, other things being
equal, more likely to do so in diverse counties than in counties specialized in
metalworking, and so on (Harrison and Kelley 1996).
By themselves these results are hard to interpret. They could be an artefact of
aggregation. It is hard to imagine, in the absence of any account of the micro-
mechanisms at work, just how the diverse airs of a great metropolis or suburban
county foster innovation. It is equally hard to see why proximity to cooperatively
competitive neighbours (talking incessantly, it might seem, about their ingenious
plans to outdo you) hinders it.
But taken in the context of the foregoing discussion these findings are less
perplexing. If firms are using disciplined comparison of difference to test and
adjust their assumptions and structures, why not seek out environments where
difference is conspicuously provocative') If you are not sure whether your firm is
part of the software industry or the financial services industry, why not
"concurrently engineer" your company in an environment where you prototype
and compare both at the same time? Of course actors that think this way are
unlikely to limit themselves (as the economic geography literature implies they
might) to locational decisions. If comparative examination of assumptions is what
you are after, why not pursue this both within the firm, through introduction of the
new disciplines, for instance, and without, in the choice of environment. Firms
that acted consistently in this way would in the end contribute to just the kind of
transformation of districts that I suggested are already in course: the "young"
firms defines itself by innovative expose to diversity in the big city; once mature,
it relocates to a district, there to avoid entrapment by unworkable assumptions
though the use of the pragmatist disciplines that are, somehow, transforming the
district from old to new.
Given all the possibilities for continuity amidst change, and vice versa, in the
districts, one reaction to all this might be, why fuss? If "old" districts are so
constituted that they can re-constitute themselves as "new" and we understand
their original constitution, why the bright-line distinctions and juxtapositions? The
real distinction, after all, is between districts, old and new, and the traditional,
hierarchical corporation. And no one is claiming that the latter has rehabilitated
itself.
Certainly the economic actors find ingenious ways of blurring the lines
traditional and craft forms. Think of the German firms that increasingly
"apprentice" youngsters to problem-solving teams, rather than master craftsmen,
leaving the philosophically inclined to wonder whether socialization into
pragmatist institutions produces a "tacit" propensity to doubt assumptions, or
 Diversity, Not Specialization: The Ties That Bind the (New) Industrial District 119

something else, Or consider that a numerically controlled machine tool is still a

tool, if you are trained to think of instruments as' such. Indeed Eric Raymond and
others in the Linux world think of programming itself as a craft (Raymond n.d.).
Surely the tenaciousness of the vocabulary suggests some continuity in fact?
Without suggesting that craft and pragmatist institutions are incommensurate
(the existence of pidgins is a strong argument against such intranslatability), I
want to conclude by briefly emphasizing a conceptual difference between them
regarding the social embeddedness of economic activity that returns us to themes
sounded at the outset
In the tacit-knowledge view economic activity is embedded in society in two
ways. First and most obviously, knowledge of substance and process "sticks" to
the social structures of networks, communities, and hierarchies. Because it is
unintelligible apart from the experience of those structures, it is tacit But trust -
the disposition to assume that one's potential partners will not take advantage of
the vulnerabilities collaboration creates - is socially "sticky" too. Put another way,
the disposition to cooperate is coincident with, and therefore limited by, the
particular social structures in which embedded. In yet another formulation
governance is built into a social order that is distinct from economic exchange
itself. The neo-Schumpeterian view of the large firm asserts this in a backhanded
way by making the formal institution of property - the exclusive ownership that
overcomes holdup problems - the precondition for the community of the large
firm. But the thrust is the same.
Pragmatist firms, large and small, dis-embed economic activity in both senses.
The new standards, disciplines and pidgins, we saw "unstick" tacit knowledge,
allowing local ret1ection on the assumptions of practice that can cumulate to large-
scale changes. How is trust or governance dis-embedded in the pragmatist firm?
By the same means. The very same information that permits collaborators to
continuously redefine their joint projects across disciplinary and geographic
boundaries allows them continuously to assess one another's probity and capacity
to meet joint expectations. Governance problems arise in general because of
information asymmetries that allow the knowing part to exploit the ignorant one.
Pragmatist institutions address these problems largely by symmetricizing
information so that the parties are alerted to trouble as it starts (Just how the
pragmatist self-correction of project teams or business units aggregates to or
otherwise shapes high-level corporate decisions about which assets to own and
which strategies to pursue is still an open question. If it is any reassurance, high-
priced talent is probably at work on it in a company near you right now).
Supposing lots, suppose this is right What, if anything, can we say about the
structure of economic activity once it is disembedded? Hierarchies are for many
obvious reasons optimal in a stable work where projects can be reliably partitioned
into narrow tasks. Accidental agglomerations of craft skill are competitive in a
world where hierarchy is too inflexible to respond to changing markets, but
different or even distant skills are very hard to connect to each other. What is the
"optimal" structure in a world where skills are connectable? Or, to use another
locution, what is the optimal network in a world where the distinction between
120 C. F. Sabel

strong and weak ties, or local and long-distance ones, tends towards irrelevance?
(Watts 1999; Dodds, Sabel and Watts forthcoming).
Those are, given the tendencies sketched above, key questions for the near
future. Addressing them will, I think, be or practical importance to economic
actors, and a spur to the renewal of renection of complex self-organization as well
(Dobbs, Watts and Sabel forthcoming).
But looking beyond the beyond, addressing these questions won't be the end of
the story. For as soon as we begin to discern "optimal" organizational relations or
search procedures in the most successful of the disciplined comparisons of
difference, we will want to know what social conditions explain why these
relations emerge first and/or nourish better in some places rather than others. And
the search for the social preconditions of "disembedness" will take us back, by a
circuitous but (hopefully) upward spiralling route to some of the oldest and most
persistent questions about the foundations - if any - of our modernity.

References

Adler, P. S., Goldoftas, B., et al. (1999), "Flexibility vs. Efficiency? A Case Study or
Model Changeovers in the Toyota Production System", Organization Science, 10. pp.
43-68.
Barnard. C. I. (1947), The Functions of the £-I:ecutive, Cambridge, Harvard University
Press.
Beardsell, M. and Henderson, V. (1999), "Spatial Evolution of the Computer Industry in
the USA", European Economic Review, 43, pp. 431-456.
Becattini, G. (1990), "The Marshallian Industrial District as a Socio-Economic Notion", in:
Industrial Districts and Inter-Firm Cooperation in Italy, Pyke, F., Becattini, G. and
Sengenberger, W.(eds), Geneva, International Institute for Labour Studies, pp. 37-51.
Brusco, S. (1989), Piccole imprese e distretti industriali: una raccolta di saggi, Torino,
Rosenberg & Sellier.
Chandler, Alfred D. Jr., Hagstrom, P. et a1. (eds) (1998), The Dynamic Firm: The Role of
Technology, Strategy, Organization and Regions, Oxford, Oxford University Press.
Coase, R. H. (1937), "The Nature of the Firm", Economica, 4, pp. 397-405.
Dei Ottati, G. (1996), "The Remarkable Resilience of the Industrial Districts of Tuscany",
in: Cossentino, F., Pyke, F. and Sengenberger, W.(eds), Local and Regional Response to
Global Pressure: The Case of Italy and Its Industrial Districts, Geneva, International
Institute for Labour Studies, pp. 37-65.
Dodds, P., Sabel, C. F. and Watts, D. (forthcoming), Searching for Searchers:
Organizational Robustness and Innovation in Highly Complex Solution Spaces.
Duranton, G. and Puga, D. (1999), Diversity and Specialisation in Cities: Why, Where and
When Does It Matter?, CEPR (Center for Economic and Policy Research) Discussion
Paper.
Duranton, G. and Puga, D. (2001), Nursery Cities: Urban Diversity, Process Innovation,
and theLife-cycle of Products, CEPR Discussion Paper.
Feldman, M. P. and Audretsch, D. B. (1999), "Innovation in Cities: Science-Based
Diversity, Specialization and Localized Competition", European Economic Review, 43,
pp. 409-429.
 Diversity, Not Specialization: The Ties That 13ind the (New) Industrial District 121

Fujita, M., Krugman, P., et al. (1999), "On the Evolution of Hierarchical Urban Systems",
European Economic Review, 43, pp. 209-25l.
Fujita, M. and Thisse, 1. F. (2000), "The Formation of Economic Agglomerations: Old
Problems and New Perspectives", Economics of Cities, Cambridge, Cambridge
University Press.
Galison, P. (1997), Image and Logic: A Material Culture of Microphysics, Chicago and
London, University of Chicago Press.
Harrison, B., Kelley, M. et al. (1996), "Specialization Versus Diversity in Local
Economies: The Implications for Innovative Private-Sector Behavior", Cityscape 2, pp.
61-93.
Helper, S., MacDuffie, 1. P. and Sabel, C. (2000), "Pragmatic Collaborations: Advancing
Knowledge while Controlling Opportunism", Industrial and Corporate Change, 9, pp.
443-487.
Helpman, E. and Krugman, P. (1985), Market Structure and Foreign Trade: Increasing
Returns, Imperfect Competition, and the World Economy, Cambridge, Mass., MIT Press.
Huriot, 1. M. and Thisse, 1. F. (eds) (2000), Economics of Cities, Cambridge, Cambridge
University Press.
Granovetter, M. (1973), "The Strength of Weak Ties", American Journal of Sociology 78
(May), pp. 1360-1380.
Granovetter, M. (2001), "A Theoretical Agenda for Economic Sociology", forthcoming in:
Economic Sociology at the Millenium, Guillen, M. F., Collins, R., England, P. and
Meyer, M. (eds), New York, Russell Sage Foundation.
Kelley, M. and Helper, S. (1997), "Firm Size and Capabilities, Regional Agglomeration,
and the Adoption of New Technology", Economics of Innovation and New Technology,
forthcoming.
MacDuffie,1. P. (1997), "The Road to Root Cause: Shop-Floor Problem-Solving at Three
Auto Assembly Plants", Management Science, 43, pp. 479-502.
MacDuffie, 1. P. and Helper, S. (1999), "Creating Lean Suppliers: Diffusing Lean
Production Through the Supply Chain", Remade in America: Transplanting and
Transforming Japanese Production Systems, Liker, J., Adler, P. and Fruin, M. (eds),
New York, Oxford University Press.
Monteverde, K. and Teece, D. J. (1982), "Supplier Switching Costs and Vertical Integration
in the Automobile Industry", Bell Journal of Economics, 13 (1), pp. 206-213.
O'Connell, L. G. (2001), Breaking the Bounds of Peripherality: Irish Business in an Era of
Multilevel PartiCipative Internationalisation, Ph.D. thesis, University College Dublin.
Ottaviano, G. I. P. and Puga, D. (1997), Agglomeration in the Global Economy: A Survey of
the "New Economic Geography", CEP Discussion Paper.
Piore, M. J. and Sabel, C. F. (1984), The Second Industrial Divide: Possibilities for
Prosperity, New York, Basic Books.
Puga, D. (2001), European Regional Policies in Light of Recent Location Theories, CEPR.
Raymond, R. (n.d.), http://www.tuxedo.orgl-esr/writings/.
Sabel, C. F. (1994), "Learning by Monitoring: The Institutions of Economic Development",
Handbook of Economic Sociology, Smelser, N. and Swedberg, R. (eds), Princeton, NJ,
Princeton University Press, pp. 137-165.
Sabel, C. F. and Zeitlin, 1. (eds) (1997), Worlds of Possibility, Cambridge, Cambridge
University Press.
Schumpeter, 1. (1926), Theorie der Wirtschlaftlichen Entwicklung: Eine Untersuchung
Ober Unternehmergewinn, Kapital, Kredit, Zins und Den Konjunkturzyklus, Mtinchen
und Leipzig, Duncker und Humblot.
Signorini, L. F. (ed.) (2000), Lo Sviluppo Locale: Un'indagine della Banca d'Italia sui
distretti industriali, Roma, Meridiana Libri.
122 C. F. Sabel

Teece. D. (1998), "Design Issues for Innovative Firms: Bureaucracy, Incentives and
Industrial Structure", The Dynamic Firm: The Role of Technology, Strategy,
Organization and Regions, Oxford, Oxford University Press.
Von Hippel, E. (1988), "Sticky Information and the Locus of Problem Solving:
Implications for Innovation", in: The Dynamic Firm: The Role a/Technology. Strategy.
Organization and Regions, Chandler, A. D. Jr., Hagstrom, P. and Solvell, O. (eds),
Oxford, Oxford University Press.
Weick, K. E. (1995), Sensemaking in Organizations, Thousand Oaks. Sage Publications.
Whitford, 1. (2001), The Decline of a Model? Challenge and Response in the Italian
Industrial Districts, Economy and Society, 30, issue I, February 2001, pp. 38-65.
Winter, S. G. (2000), "Organizing for Continuous Improvement: Evolutionary Theory
Meets the Quality Revolution", in: The Quality Movement & Organization Theory, Cole,
R. E. and Scott, W. R. (eds), Thousand Oaks, CA, Sage Publications, pp. 49-66.
Watts, D. (1999), Small Worlds: The Dynamics 0/ Networks between Order and
Randomness, Princeton, Princeton University Press.
Competitive and Synergic Behaviours
in the Development of Industrial Clusters:
Ecological Modelling and Empirical Evidence

Marco Fortis and Mario A. Maggioni 1

Abstract. Aim of the paper is, on the one hand, to highlight the economic
relevance of industrial districts in Italy in terms of employment, value added and
export and, on the other, to apply an original theoretical framework, derived from
population ecology modelling, to the analysis of the development of industrial
districts in Italy.
In this way it is possible to underline the interplay of agglomeration economies and
diseconomies in the growth process of an industrial cluster, to distinguish three
main phases of its development process and to stress the complex and different
(i.e. synergic, competitive, etc.) interactions which exists between different
industries, within the same area, and between different areas within the same
industry.
The formal model is able to identify and discuss the existence stability conditions
of long run equilibria through a graphical analysis.
The paper continue by describing the structure and evolution of some
representative examples of Italian industrial districts (namely Cusio-Valsesia and
Lumezzane districts of valves and fittings, Varese district of plastic, rubber and
specialised machinery, Castel Goffredo districts of stockings).
A discussion of the main methodological problems connected to the empirical
implementation of the original analytical framework, described above, concludes
the paper.

ISections 1 and 3.1 can be attributed to M. Fortis, Sections 2, 3.2 and 3.3 to M. A.
Maggioni. The authors thanks Erika Uberti for outstanding research assistantships and
fruitful discussions; Simona Beretta, Alberto Bramanti, Bernhard Bohm, Paul Stoneman,
Peter Swann and Massimiliano Riggi for comments and observations on various draft of the
paper.
124 M. Fortis and M. A. Maggioni

Ecology is the study of patterns in nature, of how these patterns came to be,
how they change in space and time. why some are more fragile than others.
Population ecology is concerned with how populations interact with the
environment and how these interactions give rise to the larger patterns of
communities and ecosystems. The environment is more than just sun, air,
earth and water: it includes other organisms which may help or hinder the
survival of a species. Population ecology is also the study of how these
organisms interact (... ) in competition and in co-operation.
s. H. Kingsland (1985), Modelling Nature. Episodes in the History of
Population Ecology.

No population is alone. Population interact with each other in many ways.

These interactions are important. We will see that we cannot understand
many population phenomena without considering the interactions. Even if
we want to ignore population interactions we cannot. But above all, the
interactions are intrinsically interesting because they produce perhaps the
most intricate andfascinating patterns in biology.

1. Roughgarden (1979), Theory of Population Genetic and Evolutionary

Ecology: an Introduction.

Biology is the Mecca of economists.

A. Marshall (1920), Principles of Economics.

1. Introduction

As a matter of fact, unlike the other major advanced countries, Italy has a small
number of large companies. In 1999, according to Mediobanca's list of industrial
and services companies, only four groups posted sales in excess of 20 billion Euro
(IFI-FIAT, ENI, Olivetti and ENEL) , but only one of them (FIAT) operates
primarily in the manufacturing sector, while the others do business above all in the
services industry. The latest Fortune's list of the top world companies includes
only three Italian companies (Assicurazioni Generali, Fiat and ENI).
Most of the Italian manufacturing system consists of small and medium-sized
enterprises, often organised in a number of single-product industrial districts
unequalled in any of the other advanced countries.
As highlighted at the OECD Conference held in Bologna last year, Italy has the
largest number of workers in small and medium-sized industrial enterprises
(enterprises employing less than 250 people) in Europe, with 3.6 million, followed
by Germany (3.3 million), France (2.4 million) and Great Britain (2.2 million). It
should also be stressed that Italy is the advanced country with the highest
percentage ratio of employees of small and medium-sized enterprises to total
number of industrial workers, together with Spain and Japan (Italy 73%; Spain
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 125

72%; Japan 71%), while the other major countries have a higher rate of workers
employed in large companies, with a smaller percentage working in small and
medium-sized enterprises (Germany 40%; United Kingdom 48%; France 53%).
Another peculiarity of the Italian economy is that the sectors specialised in the
so-called "made in Italy" business (fashion, home furniture, food, traditional
mechanics) are the real engine of the Italian manufacturing industry; it has been
calculated that the "made in Italy" business employs more than 3 million people.
The specialised "made in Italy" sectors consist mainly of small and medium-sized
enterprises and of districts. About 200 local manufacturing systems specialised in
particular manufacturing sectors employ over 2 million people, or more than 40%
of the domestic manufacturing workforce and more than 60% of the typical "made
in Italy" sectors.
Moreover, thanks to high-level, quality and innovative productions, the
traditional Italian sectors are a big source of value. For instance, it has been
calculated that the added value of the Italian apparel, footwear and furniture
industry, assuming the same purchase power, is higher than that generated by the
German automotive industry or the American pharmaceuticals industry.
In 1999, the "made in Italy" sectors, where most enterprises are small to medium
in size (fashion, furniture, household goods, Mediterranean food and traditional
mechanics) recorded exports for 235,000 billion lire and a trade surplus of as
much as 142,000 billion. This surplus helped to pay the energy and raw materials
deficit not only of the "made in Italy" business itself but also that of the whole
Italian system, amounting to 63,000 billion, as well as the aggregated deficit of the
other manufacturing sectors (means of transport, information technology,
electronics, chemicals, paper, rubber, and cement), equal to 53,000 billion, leading
ultimately to a total net surplus of 25,000 billion lire in the Italian trade balance. In
2000 the surplus of the "made in Italy" business exceeded 150,000 billion lire and
allowed Italy to record a modest surplus in the trade balance in spite of the
dramatic worsening of the energy balance.
These figures highlight the major role of the "made in Italy" sectors in the Italian
foreign trade. It has also been calculated that about one third of the whole Italian
exports and almost 60% of the "made in Italy" sector's come from district system-
areas.
The pattern of the Italian industrial districts has drawn the attention of politicians
and experts from all over the world.
A few districts in Italy record sales revenues of 4 to 5 billion Euro: Sassuolo in
ceramic tiles, Prato and Biella in textiles. Approximately 30 districts post sales
revenues ranging between 500 and 1,500 million Euro: Castel Goffredo in
women's hosiery, Como in silk products, the areas of Brianza and Alto Livenza
(between Pordenone and Treviso) in furniture, Arzignano (Vicenza) and Santa
Croce (Pisa) in tanning, Fermo (Ascoli Piceno) in footwear, the Cadore (Bell uno)
in eyewear and frames, Montebelluna (Treviso) in sports shoes, Verona and the
Apuan Alps in marble and granite, Vicenza and Arezzo in jewellery, Bologna in
packaging machinery, Milan and Varese in plastics processing machinery, the
Murgia region (Bari and Matera provinces) in couches, Pesaro in kitchen furniture,
126 M. Fortis and M. A. Maggioni

Rimini, ForB and Pesaro in wood working machinery, etc. Many other districts,
though smaller, are leaders in their sectors, with sales revenues of 50 to 200
million Euro: from Rossano Veneto (Vicenza), world leader in saddles for
bicycles, to Bergamo in buttons, from Premana (Lecco) in scissors to Vigevano
(shoe-making machinery). Not to mention the typical food districts such as those
of Parmigiano Reggiano or Grana Padano cheese, Parma and San Daniele hams,
Gorgonzola cheese (Novara), and wines everywhere in Italy.
In their respective sectors, the Italian districts of small and medium-sized
businesses hold world trade shares comparable to, or even larger than those of the
big automotive, chemical and electronics multinational groups: the Emilian
ceramics district and the Castel Goffredo women's hosiery district hold each about
40% of the world exports in their sectors; the Biella and Prato woollen fabrics
districts and the Como silk district hold each world exports shares ranging from
15% to 25% in their respective sectors; similarly, world exports shares around
1570 are held each by the jewelry districts of Arezzo and Vicenza, the Belluno
eye wear district, the Apuane marble district and the Emilia packaging machinery
district; shares around 10% are held by the Emilia wood-working machinery
district, the Puglia couch district, the Alto Livenza furniture district, the Vicenza
tannery district, the Verona marbles district, etc.

2. The Ecological Approach

The introduction has shown the empirical relevance of industrial districts in Italy
measured both in terms of employment and value added. This section is devoted to
the elaboration of a new theoretical modelling framework able to explain, on the
one hand, the location process of firms and, on the other, the development process
of industrial clusters.
The section studies the development path of an industrial cluster as the result of
two distinct but interrelated processes: an internal dynamics (spurred by firm's
location and spin-offs) and an external dynamics (determined by inter-industry and
inter-cluster interactions).

2.1. The Effect of Previous Locations on Clustering Dynamics

If one starts the analysis from this macro-economic perspective, the first issues to
be addressed concern the existence and the extent of spatial clustering processes,
the main questions to be answered being the following: why do firms cluster? And
why do clusters have a finite size?
In the population ecology literature, a generic modelling framework describes
the growth process of a species in an environment as driven by its reproductive
capacity and limited by the available amount of resources and the presence of
other interacting species. From a mathematical viewpoint, the core of this situation
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 127

(;an be represented by a differential equation (or by a system of differential

equations) which describe the changes over time of a variable (birth minus deaths)
as function of: the level of the same variable (the size of the population) at each
moment in time; a ceiling level (which takes into account the limit imposed by the
available amount of resources); and the level of other variables (which represents
the interacting species).
Similarly, in the following sections, firms' decisions to locate into a cluster (and
consequently the entry rate and the development path of the industrial cluster) are
explained in terms (as a function) of the "economic mass" of the cluster, measured
through the number of firms already located there.
However, these models do not merely postulate the existence of a stock-flow
mechanism between the number of incumbent firms in the cluster and the entry
rate, even if one may reasonably assume the number of spin-off firms, which are
generated in each period within the cluster, to be a proportion of the number of
firms already established there. Neither do they assume that the number of firms is
the only relevant variable considered by potential entrants when taking locations
decisions. Instead, the number of already located firms reduces potential entrants'
search costs, signals the existence of geographic and agglomeration benefits,
diffuses information, reduces uncertainty, and causes informational cascades.
What follows is thus a series of models aimed at explaining location decisions
and the development path of an industrial cluster when firms' interactions (both
effective and perceived) are relevant. For the sake of exposition, we first present a
model describing the effects of the location decisions of monoplant firms (i.e.
where firm and establishment coincide), belonging to a given industry, within an
isolated cluster (i.e. when there is only one potential site for location). In the
following sections, the scope of the analysis is enlarged to encompass a number of
interactions between two (or more) clusters and/or industries.

2.2. Locational Benefits and Costs and the Development of an

Industrial Cluster

Firms decide to settle in a cluster on the basis of the expe(;ted 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.
For simplicity it can be assumed that, in an uncertain world - with limited
information regarding local costs and revenues available to the outsiders -
profitability expectations for any particular location will be based solely on the
number of firms already located there (the number of previous locations being the
only observable variable).
128 M. Fortis and M. A. Maggioni

Let us assume, as in Arthur (1988 and 1990), that locational gross benefits
B fq for firm f locating in cluster q are composed of geographical and
agglomeration benefits 2 •
Geographical benefits Gfq depend on the intrinsic features of the site (such as
the quality of local factors of production: capital kq and labour lq; the efficiency
of the local network of specialised suppliers and business service firms Sq; and
the quality of urban and industrial infrastructures u q ). Agglomeration benefits

A fq (n q) are a concave non-monotonic function of the number of incumbents (i.e.

firms already established in cluster q) nq . Thus:

(1)

The assumption of concavity and non monotonicity in Aq implies that, as the

number of firms located in cluster q increases, gross benefits firstly increase
because of agglomeration economies (due to productive specialisation; scientific,
technical and commercial spillovers; reduction in both transport and transaction
costs, increases in the quality of the local pool of skilled labour force and in the
efficiency of the local credit market); then decrease when congestion more than
compensates for agglomeration economies.
Locational costs C fq' symmetrically, include two components: geographical
costs g fq (ret1ecting the cost structure of the cluster in terms of locally prevailing
wage W q and interest rate rq ; average price of business services d q ; and level of
land rent and taxation t q ), and agglomeration costs a q , which are assumed to be
a convex non monotonic function of the number of regional incumbents nq .

2For analytical convenience we split locational benefits in two classes: geographical 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, we do not intend to
state that agglomeration benefits refer only to spillovers of scientific and technological
knowledge and know-how. On the contrary we are convinced that relevant agglomeration
benefits derive also from external economies of scale in the use of local resources. The
same variable (i.e. labour productivity) has a fixed geographical 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.
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 129

The assumption of convexity and non monotonicity in aq implies that, as the

number of firms in cluster q 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, labour, business services, land and public infrastructures) which raises
their prices 3 •
Net locational benefits can now be calculated as the difference between
equations (1) and (2).

Assuming that the geographical benefits and costs do not change overtime, if
we focus the analysis of the location process on the dynamics of the interactions
between the level of available locational benefits, what becomes relevant for
describing firms' location decisions is just the net benefit function N jq in the
incumbents' space. We can therefore summarise the geographic components
H fq with a parameter aq , which vertically shifts the locational net benefits
function, and write the following expression:

(3b)

3 An alternative explanation for the convexity of the locational costs function for firm/runs
as follows: the locational costs function is compose by a "fixed" and a "variable"
component. The fixed part of the costs (geographic costs) decreases as the number of
entrants increase; while the variable part increases (because of competition) as the number
of entrants increase. The combination of these two effects produces an U-shaped (convex)
cost curve as the interaction between tixed 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 paper,
however it is not as theoretically efficient as the other one since both components become
dependent on the number of incumbents.
130 M. Fortis and M. A. Maggioni

It easy to see that the locational net benefits function (3b) is always concave,

since Nfl] is equal to the difference between a concave function B Jq (n q ) and a

convex one CJq (n q) . In other words, each marginal firm, which enters the cluster,
increases the 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 entrant4 .
This formulation recalls some general results, obtained in the industrial location
and urban/regional economics literature (Weber 1929; Isard 1956; Richardson
1978; Papageorgiou 1979; Tauchen and Witte 1983; Miyao and Kanemoto 1987),
which show the existence of an optimal dimension of a given spatial agglomeration
of firms and/or households because of the concavity of the various benefits
functions.
We can therefore state that - if the number of potential entrants is sufficiently
large and there are no relevant entry barriers - as net locational benefits initially
increase, the growth rate of incumbent firms increases; then it decreases and is
finally driven to zero.
If we assume that the number of potential entrants is not constant but it is
changing overtime, than the entry pattern will be determined by the rate of birth of
potential entrants, since a proportion of such firms will locate in the cluster. If we
allow the number of actual entrants to be proportional to the average locational
benefits available in the cluster - as a first approximation (or, alternatively, as a
macroeconomic hypothesis without explicit micro-foundations 5) - and we assume
that the entry rate of firms into the cluster is proportional to the current level of
locational net benefits 6 , then the cluster growth will initially be fast, then will slow
down and finally will stop7

4However, as it is made graphically evident in Figure 4.2, 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.
5 Another part of this story refers to the role played by spin-off firms. If we believe that
average net locational benefits give incumbent firms some extra profits, workers (or, better,
managers) of incumbent firms, which see these extra profits, will decide to start a new firm
to appropriate them. Therefore the number of spin-off firms, which are generated by parents
firms. is positively related to the level of locational benefits in the cluster. generated by the
incumbent firms.
GIt is evident that, for the moment, we are implicitly assuming firms to behave in a non-
strategic manner (i.e. although their entries modify radically the locational benefits
available in the cluster, they do not take into account such a modification) and to be totally
myopic (i.e. their entries are only dependent on the current level of locational benefits and
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 131

If this is the case 8 , one would expect the industrial growth of the cluster to
follow an S-shaped path with a slow start (when locational benefits are still low),
an "explosive" central period of rapid increase (when the average net benefits in
the cluster are highest) and a final part when the cluster gradually reaches and then
settles to its "equilibrium,,9 size as in Figure 1.

number of
incumbents

maximum
dimension
(K)

discovery

time

Fig. 1. The development path of an industrial cluster.

not on some discounted value of the flow of locational benetits they enjoy from the moment
of entry onward). In section 4.3, we will relax some of these hypotheses.
7This is not to say that, from a certain moment onward, entries will not take place. On the
contrary, once the equilibrium level of the cluster is reached, new entries are still possible,
but these happen at the expenses of some incumbents which are driven out of the cluster
(or, more probably, out of business). This continuous series of entries and exits may
determine a continuous oscillation of the cluster's industrial mass around the equilibrium
level.
8 And we collecled some empirical evidence supporting this story (see Maggioni 1993;

Maggioni and Porro 1994; Gambarotto and Maggioni 1998; Maggioni 1999).
9 It should be bore in mind, however, that this notion of equilibrium value can be changed
in the long run by radical innovations, exogenous demand shocks and appropriate regional
policies.
132 M. Fortis and M. A. Maggioni

What is needed now is a formal model of the location process and the
development of clusters, which will allow a rigorous testing of all these
assumptions.

2.3. A Logistic Model

The simplest growth model for an industrial cluster q - 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 lo to the regional population's growth and the number of firms
already in the cluster" (Maggioni 1993, 1999).
If only agglomeration economies and positive 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 the
cluster would be equal to a constant rq . In this case cluster industrial growth
would follow an "explosi ve" exponential path II, formally:

dnq
-dt = rq n (t)
q
(4a)

Alternatively one can solve the equation for nq (t) as a function of the
exogenous initial industrial mass of the cluster nq (0)

(4b)

On the other hand, if 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. A simple dynamic
model, which takes into account these features, is the logistic equation l2 , which
can be written as (Sa)

lOIn terms of changes in the average locational net benefits, due to the interactions of
agglomeration economies and diseconomies.
liThe higher is rq, the faster is the growth process.
12The logistic equation - firstly developed by Verhulst (1845) and Pearl and Reed (1925)
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
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 133

dnq
-=rn(t) (11
1--'1-
(t)'J (5a)
dt 'I 'I K
'I

where r is the incipient (or maximum) rate of increase and Kq

'I
= limnq
t~-
(t) , is
called the cluster "equilibrium" level 13 . Integrating equation (5a) and solving for
11'1 (0) one obtains (5b):

(5b)

Here the individual firm's contribution to regional growth decreases as a linear

r{
function of regional population size and is equal to rq - -I-n (t) (Roughgarden,
K'I 'I

1979).
Plotting nq (t) against time yields an S-shaped curve due to the counteracting
roles played by rq and K q . When the cluster is small (11'1 (t) is near to zero) the
term in brackets in equation (5a) is close to one (hence the logistic equation
approximately describes an exponential growth path); but as nq (t) approaches
Kq , the term in brackets tends to zero, driving the growth rate to zero and
terminating the entry process. Both Kq and rq playa major role in shaping a
logistic growth path: the greater is rq the steeper is the S shaped curve, the larger
Kq the higher the ceiling level of the function (and the equilibrium size of the
cluster).

tirst 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, p.
38).
13 K in the original ecological jargon is called carrying capacity: "a measure of the amount

of renewable resources in the environment in units of the number of organisms these

resources can support" (Roughgarden 1979, p. 305).
134 M. Fortis and M. A. Maggioni

rq is the Incipient rate of growth. In the ecological literature it is often

approximated by 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 industrial growth of a cluster) is
calculated as the difference between total entries (or start-ups) and exits (or
bankruptcies) in the period considered. The same value of rq can therefore
correspond to two very different situations: a steady growing cluster 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. Hence rq is a composite index
that describes the cluster growth "potential" and the probability that firms, once
entered, survive in the cluster.
Kq defines the regional industrial carrying capacity: the maximum number of
profitable firms the cluster can sustain in isolation (i.e. when inter-regional
interaction are not considered). Kq will depend upon:
- the finite quantity of geographical benefits (which is related to the limited
availability of local "resources" such as: labour, capital, land, and
infrastructures);
- the decreasing part of the agglomeration benefits function (which depends on
the strategic interactions between firms: competition, congestion and lobbying
of incumbents).
Kq is therefore determined by the relationship between the amount of
resources (inputs) available in the cluster and the (technical and organizational)
efficiency of incumbents in the use of these resources. T.herefore in the long run
Kq may change as result of the int10w of additional skilled workers, the provision
of new advanced public infrastructure, the diffusion of (technical, organisational,
etc.) innovations.
For a given cluster q and a given population of M q outsider firms14, therefore,

we assume that there is an equilibrium level Kq ~ M q acting as an upper limit to

the cluster's growth. In each period t, the number of entlies therefore depends both
on the actual number of potential entrants Kq - nq (t) (i.e. the number of
outsider firms which can enter the cluster in time t and still make profits) and on
the number of firms already located there nq (t). Kq and nq (t) in fact

14Composed of two main categories: tlrms already established outside the cluster and
potential entrepreneurs inside the cluster looking for the right moment to start their own
business.
 Competiti ve and Synergic Behaviours in the Development of Industrial Clusters 135

determine the level of average locational net benefits available to incumbent firms
in each period of time.
In this first formulation, the number of located firms 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.
Such a formulation of the location process is very simple but can be used to
empirically estimate key parameters of the location path of different clusters's.
These estimated parameters could also be used as dependent variables in cross-
section analyses in order to assess the int1uence of different factors on the level of
the intrinsic growth rate of a cluster or on its maximum dimension.
However this formulation can be criticised on a number of grounds. The main
drawback refers to the "isolation hypothesis": firms' location decisions are
modelled as a dichotomous choice, in the sense that there is only one possible site
for location and the choice variable is just the timing of the entry. The following
sections extend the simple logistic growth path to take into account bilateral
interactions.

2.4. Location Process in a Two-Clusters Framework

Several different interactions may arise when a framework with two clusters (I
and 2) is considered. The first concerns the location decisions of a set of industry-
specific potential entrant firms confronted with the choice of locating in one out of
two clusters. This problem can be formally described through a model of inter-
cluster competition with a given number of potential entrant firms l6 . The entry
decision is again modelled as dependent on the number of firms located in each
cluster, ret1ecting the effects of agglomeration economies and diseconomies, as
described in the previous section.
Formally, the multi-cluster situations l ? can be described through the following
system of equations which directly recalls the mono-cluster development process
of equation (Sa):

15Where the difference may refers to different industries in the same geographical site, or to
different geographical sites in the same industry.
16This peculiar approach to the process of multi-clusters development (where the cluster is

modelled as an agent and firms as passive industrial mass) reveals its advantages when
modelling active policy pursued by different local authorities willing to foster the industrial
development of their respective territories.
17The same analytical framework can be used to describe the interaction between two
industries in the same cluster which compete on a given pool of local limited resources
(land, workers, bank credit etc.).
136 M. Fortis and M. A. Maggioni

(7 a)

where rl and Kl are respectively cluster 1 incipient growth rate and equilibrium
levels and c 12, is the "competition coefficient" which measures the extent to which
cluster 2 competes as an alternative location to cluster 1.
However - for expositional purposes - it is more convenient to replace this
formulation (which underlines the similarities between the two-clusters case and
the model of development of one isolated cluster depicted in 4.Sa) with an
alternative formalisation which is able to model different multiregional interactions
through simple differences in the sign of parameters. We thus replace (7a) by (7b),
where time references are dropped for the sake of clarity:

(7b)

In system (7b), n1 and n 2 are the "economic masses" (number of incumbents)

of each cluster, a l and a 2 are the intrinsic rates of increase of each cluster in
isolation (1j, r2 in system 4.7a), all and a 22 - the parameters on the quadratic
terms which reflect the concavity of the net agglomeration benefits function - i.e.
that reflect the inhibiting effects that a firm's entry has on the growth rate of the
same cluster l8 (intra-cluster competition parameters); a l2 and a 2l show the
inhibiting effects that one firm - locating in a cluster - has on the growth of the
other cluster l9 (inter-clusters competition parameters).

18 ~ and ~ in this formulations are the clusters' maximum size or isolated equilibrium
all an
values (K 1 and K2)·
190ifferent interpretations for these parameters are possible: the simplest relates the values
of inter-regional competition parameters to the fact that - given a certain number of
potential entrants - each tirm that locates in cluster 1 does not locate in cluster 2. Then the
inter-clusters coefficients should be equal, i.e. a l2 = a 21 . Another interpretation explicitly
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 137

According to the relative value of these parameters, three main outcomes 20 are
possible, as shown by phase diagrams in Figure 2.
Phase diagrams are useful tools to describe systems of differential equations. At
any moment in time the state of the system is fully described by the number of
firms located in each cluster (n 1 and n 2 ). Within the phase plane one can further
draw for each cluster the locus of points (i.e. combination of n l and n2 ), which is
called the isocline, where the cluster's industrial mass does not change (i.e. where
dn l dn z
--- = 0 and - - = 0) The multiple equilibria of the system are thus
dt dt'
identified by the intersections of one isocline with either the axes or the other
isocline 21 . The arrows in the phase planes indicate the direction in which the
system, at that point, will move; the colours and shapes of the dots show whether
an equilibrium is either stable or unstable or a saddle point.

relates to the possible differences in the logistic growth path of the two clusters in isolation
(which renect differences in the net benefits function). In this case inter-clusters
competition coefticients can be different.
20Selected from the analysis of 13 possible consistent cases. For a more complete analysis,
see Capelo (1989, pp. 165-169) and Beltrami (1987, pp. 72-73).
2lThe intersection of an isocline with an axis is an equilibrium point since one cluster in
deserted and the other does not change its industrial mass, the intersection of two isoclines
is an equilibrium point since both clusters do not register any changes in the number of
located firms.
138 M. Fortis and M. A. Maggioni

2a 2b
n2

n'I=O
E2

E2 :x

EO
EI
nl
EO EI nl

2c 2d

E2

EO EI nl EO EI nl

Legend:

o unstable cquilibriulll

•o stable equilibrium
saddle point

Fig. 2. Inter-regional competition.

Fig. 2a shows the persistent coexistence of two clusters (both recording a positive
number of locations) as a stable equilibrium when:

al az al a2
->- and -<- (8a)
a l2 a l2 all a 21
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 139

Fig. 2b shows the coexistence of two clusters as an unstable equilibrium, which

is extremely vulnerable to local perturbations and initial biases 22 when:

and (8b)

Fig. 2c and 2d show the total exclusion of one cluster from the location process
that occurs only in the other cluster (irrespective of the initial state) which reaches
its carrying capacity leaving the opponent deserted 23. In particular, only cluster 1
is entered when:

al a2 al az
->- and ->- (8c)
a l2 a 22 all aZI

while only cluster 2 is entered when:

al a2 al al
--<- and -<- (8d)
all all all all

Since the industrial growth of the cluster in isolation follows a logistic path, then
the isoclines are linear and the stability of equilibrium depends on the equilibrium
point being above the line24 joining the maximum dimensions of each cluster in

isolation: (~, oJ and (0,

all
a2
a Z2
J 25.

This presentation of the possible outcomes of firms location processes in a two-

clusters framework would seem to suggest that inter-cluster competition leads -

22More precisely the equilibrium defining the coexistence of two clusters is a saddle point;
any initial state off the stable manifold, or any perturbation that move the system far from
it, leads to the "deprivation" of one out of two clusters.
23This case includes two opposite situations of only cluster I surviving and only cluster 2
surviving.

24Whose equation is nl = (1- _n_2_Y

l allaz2
.!:LJ .
~ all
25 Alternatively, if one expresses the clusters size in "equivalent numbers" (i.e. as fractions
of their maximum dimension), the equilibrium is stable if the total equivalent number at the
joint equilibrium is greater than unity (i.e. greater than the equivalent number at the
isolated cluster's equilibrium).
140 M. Fortis and M. A. Maggioni

almost necessarily - to a situation of locational monopoly, in which only one

cluster is chosen as the preferred site by industry-specific firms while the other
remains deserted. Although the economic history of industrial clusters displays few
cases supporting this result, the coexistence of mUltiple non-empty clusters is a
more common phenomenon. This must therefore be caused by factors still missing
from the above analysis.
One way to explore the problem assumes the intrinsic growth rates of two
clusters to be equal (i.e. 01 =°
2 ). Then the persistent coexistence of both clusters

is achieved when:

and (8e)

These two inequalities state that, to obtain persistent coexistence of both

clusters, an increase in the numbers of incumbents in either cluster must inhibit its
own growth more than it inhibits the growth of the other (e.g. that the degree of
intra-cluster competition and congestion must be larger then the degree of inter-
regional competition). This result is likely to be achieved if the clusters are, at least
partially, "attracting" different types of firms or, in other words, if firms have
significantly heterogeneous locational preferences.
This observation has been acknowledged within the population ecology
literature for a long time under the name of the "Gause principle"26. This principle
states - in the original biological terms - that in a situation where "two species are
resource-limited, ( ... ) only one species can survive on a single resource" (Levin
1970, p. 413). Translated into an economic framework, the Gause "principle of
competitive exclusion" states that - if firms are completely homogeneous in their
locational preferences - then the most likely outcome is the unanimous choice of
one single cluster (the one ensuring at the start the highest level of geographical
benefits). The existence of a plurality of non-empty clusters inhabited by firms
belonging to the same industrial sector can therefore be explained in terms of the
existence of different types of firms that have different locational needs and
preferences 27 .
The degree of competition between two clusters is therefore proportional to the
degree of locational homogeneity of the firms' population. The same concept can
be expressed in a rather different way: the degree of inter-regional competition is

26 Being formally stated by Gause (1934), even though an earlier and slightly different
formulation is ascribed to Volterra (1926).
27 It is very interesting to compare this result to very similar ones obtained independently
and following a totally different approach by Arthur (1988 and 1990).
 Competitive and Synergic Behaviours in the Development of Industrial Clusters [41

Inversely proportional to each cluster's specificitl8 . According to this

observation, each cluster should foster the development of specific regional
characteristics in order to develop a "niche" strategy able to limit the degree of
inter-regional competition. This observation, which has a general value, has a
particular interest for any public authority willing to start a new industrial cluster
in an area where there are already other established clusters. The identification and
the attraction of a specific subset of firms is the crucial element, which determines
the success, or the failure, of the regional development project.

2.5. Co-operative Interactions

The analytical representation, used in equation (7b) to describe the location

decisions of a set of potential entrants between two alternative locations, can be
easily encompassed to describe other significant interactions which may arise
during the development process of clusters and/or industrial clusters. These
different relationships can be modelled simply by changing the signs (or values) of
the inter-regional coefficients a 12 and a 21 in the system of equations (7b) as
shown in table 1.

Table 1. Different types of bilateral interactions.

Types of mteractions Sign and values of interaction coefficients

al2 and a2l

Competition

Mutualism + +

Commensalism + 0

Predation + -

Neutralism o0
Amensalism o-

28Where regional specificity means that each cluster is the ideal location for a different type
of firm.
1~2 M. Fortis and M. A. Maggioni

The regional economics literature records several contributions on predatory

relationships (see for example; Dendrinos and Mullally 1985; Maino 1989;
Nijkamp and Reggiani 1992 and 1998). This section therefore concentrates on co-
operative relations that - despite their empirical relevance - have been neglected by
the main stream of population ecology and, consequently, by the early pioneers of
"ecological" regional economics 29 . These relationships describe a set of co-
operative interactions where the development of a certain set of firms, such as an
industrial cluster in a given nation (or one industry in a given cluster) is not
obtained at the expense of the development of another industry (or cluster) but the
two processes enjoy synergistic effects.
Both ecological and economic systems record empirical evidence of mutualistic
relationships between two populations over a long period of time. There are plenty
of historical cases and empirical evidence of these phenomena: from the
development of the textile districts in the mining clusters of Britain and Belgium in
the 19th century30, to the synergistic development of an industrial and a residential
bordering area 3l , from the development of a subcontractors' belt around a main
industrial core, to the long timed establishment of a cluster of integrated producers
and craftsmen in one specific quarter within a ci ty32.

29There are two different reasons why mutualistic and amensalistic relations have been
ignored. The first reason is an historical one: both Lotka and Volterra - generally
acknowledged as the forerunners of population ecology - limited their analyses to predation
and competition. The second reason is a theoretical one: in its purest formulation, a
mutualistic model docs not have a stable equilibrium and one or both populations undergo
"an unbounded exponential growth in an orgy of mutual benefaction" (May 1976, p. 95). A
growth model without stable equilibrium seemed of no interest to regional and applied
economists. However, more sophisticated models of mutualistic interaction allow for
saturation in the magnitudes of reciprocal benefits and obtain - as general result - a stable
equilibrium with both populations (usually clusters, but they can equally be industries)
developing larger than in isolation. It must be said, furthermore, that this equilibrium is less
stable - in the sense that perturbations are more slowly damped - than in correspondent
systems without mutualistic relations.
30Textiles firms, in these locations utilised the female components of the miners' families as
source of cheap labour force.
3lThe first demanding workers and offering wages. the second demanding goods and
offering labour. This interaction goes well when there are not relevant negative externalities
(such as pollution, noise etc.) spilling from the industrial cluster to the residential one. In
such a case we may well describe the situation as amensalistic.
32For an example of such clustering dynamics, see Storper and Walker (1989, p. 81) which
report an amazing map showing the agglomeration in the London watch and clock making
quarter of Clerkenwell, in 1861.
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 143

In particular, we think that the model analytically described in expression (9)

could well describe the behaviour of two technologically related industries within
the same industrial cluster, each receiving positive benefits from the development
of the other one 33 . We think that the analysis of inter-industry relations is as
important as (if not more than) the analysis of inter-regional competition since one
of the main findings of the literature on industrial clusters, and of the original
empirical evidence presented in section 6.3, is that the physical proximity of firms
belonging to different but interrelated industries i= (1,2) act as an engine
sustaining the cluster's long run growth 34 .
The above remarks can thus be described more formally as follows:

(9)

In this model each industry in isolation would follow a logistic growth pattern,
but the presence of one industry in the same cluster has a positive influence on the
rate of growth of the other one. The system's dynamics - which is described
through a phase diagram in Figure 3 - always displays three trivial unstable
equilibria (Figure 3a): the origin (corresponding to the underdevelopment of both
industries) and each industry's isolated maximum dimension. However, under
certain assumptions concerning the slope of isoclines (Figure 3b), a fourth stable
equilibrium (E 3) emerges which enables both industries to reach a higher level in
the cluster. The coexistence of two industries in the same cluster allows both
industries to grow larger than would have been possible in the isolated case, but
intra-industry competition effects prevent the system from experiencing explosive
growth.

33However it is also theoretically possible that inter-industries relations, within a cluster,

become cont1icting because of competition on non industry-specific inputs such as real
estates, tinancial services, fiscal and accounting consultants etc. The prominence of co-
operative versus competitive interaction must therefore be empirically assessed.
34Swann defines such a process as "convergence between technologies" and shows that
"single technology clusters, while they might grow faster in the formative stages, did not
have the lasting power of diversified clusters" (Swann 1998, p. 64).
144 M. Fortis and M. A. Maggioni

3a 3b
n2 nl
n'I=O

n'2=O

n'I=O
E2
E2

EO EI EO
nl EI nl

Legend:
o unstable equilibrium
• s[able equilibrium

Fig. 3. Inter-industry facultative mutualism.

The conditions for an equilibrium solution in which both industries survive can
be re-expressed as a strict positivity requirement on the co-ordinates of E 3 in the

(111' I1 z ) space, formally:

(aZa ll +alaZI )
...,-'-------'--,- >0 (lOa)
(alla ZZ - alZa ZI )

(a Za l2 +a l a Z2 ) >0
( lOb)
(all a 22 - a 1Z a 21 )

Given that all coefficients are positive, the two conditions reduce as follows:

(lOc)

A stable equilibrium in a mutualistic interaction is therefore achieved when the

product of the two intra-industry competition coefficients exceeds the product of
the inter-industries cooperation coefficients. The same analytical framework may
describe the mutualistic growth of two neighbouring clusters, specialised into
 CompetitIve anti Synergic Behaviours in the Development of Industrial Clusters 145

complementary productions, where the driving forces of the joint development are
vertical and horizontal productive linkages established between clusters.
Other stimulating results are obtained when the symmetry assumption required
by facultative 35 mutualism is relaxed. An interesting case concerns a situation in
which the first cluster in isolation will grow according to a logistic path and its
growth is further fostered by the presence of industrial activity in the neighbouring
one, whilst the second cluster is strictly dependent on the existence of the
industrial activity in the first cluster for its very survival. A typical example of this
kind is the relationship existing between the "core" of an industrial area (where the
large companies are located) and the surrounding area, "periphery", crowded by
several small firms (i.e. sub contractors and suppliers) - specialised in accessory
productions - which are vitally linked to the industrial success of the core 36 .
Similar relationships (which do not imply any spatial differentiation) may well
arise between two different industries in the same cluster. A typical example
(thoroughly discussed in Swann, 1993) is constituted by the relationships between
the manufacturing core of an industrial cluster and the network of specialised
service sectors. Since these specific business services must be "consumed" in the
very place of their production and cannot be exported to other areas, the existence
of a service sector within the cluster is crucially dependent on the existence of a
prosperous manufacturing core.
These interactions can be formally described as follows:

(11 )

Where 111 is the economic mass (number of located firms) of the manufacturing
core cluster and rq is the economic mass of the specialised service sector.
With the aid of a phase diagram (see Figure 4a), it is possible to see that the
system has three equilibria. The first is trivial, unstable and corresponds to the
origin; the second, stable, corresponding to the isolated maximum dimension of
the core manufacturing industry; the third, which is a saddle point, corresponds to

35 Where the term "facultative" refers to the fact that each industry, in isolation, develops
and reaches its equilibrium level. The presence of an industry is therefore no necessary
condition for the development of the other one.
36 The Motor vehicle industry (which is clustered around Detroit in the USA, Coventry in

the UK, Besan<;on in France, and Torino in Italy) displays a core-periphery spatial
configurations in almost every countries.
146 M. Fortis and M. A. Maggioni

a balanced co-operative relationship3?, in which manufacturing and service sectors

coexist with mutual benefits. When the system lies in this unstable mutualistic
equilibrium, small positive perturbations lead the system to an unlimited growth
while small negati ve perturbations may cause the "extinction" of the service sector.
a2 al
This is the only possible outcome when - <- (see Figure 4b).
aZI all

4a 4b

n2 n'2=0
n2

! n'I=O

E3

!
I
EO EI EO
nl E3 El nl

Legend:
o unstable equilibrium
• Slable equilibrium
o saddle poinl

Fig ..... Core-periphery dynamics (inter-industry commensalism).

2.6. Predatory Interactions

The relationships between two industries in the same cluster may also display a
predatory pattern with one industry growing at the expense of the other.
Predatory interactions have been widely studied in the ecological and in the
economic literature (one does not need to recall the names of Alfred Lotka, Vito
Volterra and Richard B. Goodwin) and they may well describe the sort of
interaction existing between a rising and a declining industry in a given area.

37Which exists if and only if az >!:L (see Figure 4.5a).

a ZI all
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 147

The story is as follows: an established industry in a certain area produces, as a

sort of spin-off, the birth of an interrelated industry (i.e. the footwear industry
requires specialised machinery for cutting and sewing leather). If the derived
industry has a higher value added per worker (or, more generally, if firms in this
sectors are more productive and profitable) then it may well happen that the
development of the second industry determines endogenously the crisis of the first
one since it may cause the rise of local wages, rent and interest rate to such a level
which is not sustainable for the original industry.
Analytically one may model a prey-predator system as a mix of systems (7b) and
(9) since its essence lies in the combination of a competitive with a synergetic
relationship.

(12)

Where n1 is the prey (or the older and currently declining industry) and n2 is
the predator (or the younger and rising one).
Predatory interactions typically display oscillatory dynamics. A phase diagram
(see Figure 5) illustrates two possible configurations: case (a) where the
coexistence of both population is a stable equilibrium surrounded by a converging
spiral, case (b) where the coexistence of both population is an unstable equilibrium
surrounded by a diverging spiral.

Sa Sb
n2 n2

n'l=O
/

E2
/ " £2

EO EI EO
nl EI nl
Legend:

0 unstable equilibrium

•
0
stable equilibrium

saddle point

Fig. 5. Intra-cluster structural dynamics (inter-industry predation).

148 M. Fortis and M. A. Maggioni

In the two situations, if the system starts far from the equilibrium, both predator
and pray number oscillate. However, while in the first case the oscillations have a
decreaSIng amplitude and, in the long run, the system reaches its equilibrium; in
the second case the oscillations continuously increase their amplitude and the
system dynamics may well determine the extinction of one or both populations.
Besides the one above mentioned, various type of interactions coexist within the
development dynamics of industrial clusters. The qualitative identification of such
interactions and the quantitative measurement of their effect on the growth rate of
industries are thus left to the empirical analysis of section 3.3. The effects of the
variability of macro-economic conditions on the behaviour of the systems have
been extensively discussed in Gambarotto and Maggioni (1998).

2.7. Multi-Cluster Interactions: Some Methodological Remarks

Previous sections have shown some results that can be obtained by using a series
of ecologIcally derived models to analyse the growth of an industrial cluster (or
industry) in isolation and of a system of two interacting clusters (or industries).
It would seem therefore natural to extend these results to the case in which more
than two clusters are involved. Unfortunately, this extension is not straightforward
for empirical as well as theoretical reasons. The firsts refer to the geometrical
increase of the relevant interaction parameters, while the second concerns the
dynamic behaviour of the system which becomes qualitatively complex.
Two different approaches can be used to overcome this problem. The first
relates to the possibility of constructing a large computational model, assigning
plausible parameters' values and then simulating the dynamic of the system; the
second consists of abandoning the description of individual cluster behaviour -
within a multi-cluster complex - and focusing entirely on some overall aspects of
the resulting structure, such as the relation between the increase in system
complexity and its overall stability'8.
Even though it is difficult to draw some general conclusions, here are some
principles and guideposts for multi-cluster analysis which could be derived from
the original population ecology literature (see Levins 1968; Levin 1970; May
1974; Roughgarden 1979) which may give some indications on possible behaviour
of such a model.
Equilibria may exist only within limited ranges of the interaction and
environmental parameters (such as cluster's incipient growth rates, maximum

38 Inthis section "increased complexity" is, rather loosely, defined in terms of increased
number of cluster, more interactions and therefore more parameters. Stability is described
as a lower level of fluctuations in the economic mass of the clusters, and as the ability of
the system to recover from perturbations.
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 149

dimension, and macro-economic variability). Outside this restricted range,

equations may undergo sustained oscillations and descflbe a collapsing system in
which one or all clusters are completely deserted by firms.
Even when the value of parameters ensures an equilibrium solution, this
equilibrium may be unstable to large perturbations in the "economic mass" of the
cluster. This feature involves the concept of "resilience 39" of the system and
highlights the difference between global and local stability conditions.
Fixed parameter values have been assumed throughout the above analysis; on
the contrary, in the real world, both interaction and environmental parameters may
t1uctuate through time, causing further perturbations to the system.
Studies based on a generalisation for n-agents of equation (7b), in which the
coefficients are assigned randomly, show that as the number of agents involved
increases, the probability to find a "feasible" equilibrium (i.e. where all clusters
register a positive number of located firms) decreases. Biological models also
show that dynamic stability typically decreases with the increase in the number of
populations, or in the number and strength of inter-populations interactions 40
The previous statement should not be interpreted as a general law stating that, in
the real world, complex systems are less stable than simpler one. If a long run
perspective is assumed - and we allow the market to act as a selection device - then
it is likely that, in a relatively stable environment, a complex and fragile system
will persist forever; whilst, in an highly unpredictable and exogenously determined
environment the same system will collapse.
The Gause principle of competitive exclusion holds for multi-cluster systems as
well. In its original ecological formulation it reads as follows: "no stable
equilibrium can be attained in an ecological community in which some £
components are limited by less than £ limiting factors" (Levin 1970, p. 419-20)41.
In economic terms this means that only one cluster can be sustained through a
single-industry development which is based on one single type of firms. In other
words a system of £ clusters is stable only if there are at least £ different Lypes of
firms.

39 Which can be detined as a measure of the magnitude of the populations' perturbations

that the system will tolerate before collapsing into some qualitatively different dynamic
regime.
4o"Thus, as a mathematical generality, increasing complexity makes for dynamic fragility
rather than robustness" (May 1974, p. 160).
41 Some similarities (not simply formal) can be found between Gause principle and the

requirement on the number of objectives not to exceed the number of instruments in the
theory of economic policy, and the relation between number of factors and number of
commodities in international trade theory.
150 M. Fortis and M. A. Maggioni

3. The Empirical Evidence

This section is twofold. On the one hand it is devoted to the description of the
structure and evolution of four Italian industrial districts; on the other hand it
contains an explorative descriptions of the implementation of the econometric
estimation of the relationships described in section 2 and of the main problems
(both methodological and related to the sample size) encountered in such an
empirical analysis. Finally some first empirical results, derived from both
econometric and calibration techniques.

3.1. Structure and Evolution of Three Industrial Districts: A First

Descriptive Analysis

With about 300 firms and over 2,000 billion lire of annual sales revenues, the
women's hosiery district of the Mantova area, with its centre at Castel Goffredo, is
one of the successful industrial districts of the "made in Italy" business. The
district, the absolute leader in its sector in Europe, contains all the leading
companies operating in the business in area of a few square kilometres: the largest
Italian groups, among which is Golden Lady (owner of the Omsa and Sisl trade
marks), Csp International (now listed at the Milano Stock Exchange), Pompea,
Levante, Calzificio Real, a number of small and medium-sized firms, as well as
Filodoro, a company controlled by multinational Sara Lee, that settled in the
district to enjoy the location economies and benefits available in a highly-skilled
area.
The historical origins of this industrial district go back to the first women's
hosiery factory (Noemi) built at Castel Goffredo in the early '30s, when the area's
prevailing activity was still farming. The young were the first to leave the rural
areas and start working in the industry, thus contributing to the development of the
first textile settlement, which, however, had a rather short life. When Noemi closed
down in 1958, the workers, employed until then in the various manufacturing
stages of the hosiery industry, became entrepreneurs and set up small handicraft
shops that would later slowly develop into the first firms of the district; this way,
through a typical budding process, they gave rise to a network of small enterprises,
each specialized in a single manufacturing process. Later, early in the '70s, the
panty hose appeared in the domestic marketplace and was such a success that the
local manufacturers were initially hardly able to meet the demand. The number of
factories rose, also helped by the local allied activities such as the manufacture of
machinery for the textile and stocking-weaving industry in particular (Lonati,
based in Brescia, the first manufacturer of tubular hosiery machines in the world,
was the biggest example). The Castel Goffredo pattern became more and more
synonymous with quality, attracting big foreign dealers and encouraging the local
operators to look outside the domestic boundaries to discover a strong export
vocation that is still a typical feature of the district. One of the success factors of
Castel Goffredo is the close relationship between the local manufacturers and large
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 151

cumpanies such as Du Pont and Nylstar (a fifty-fifty joint-venture between French

Rhodia and Italian Snia) for the supply of nylon, the raw material for women's
hosiery. With Lycra, for example, hosiery has become high-tech: the advantages of
comfort and longer life were immediately appreciated by women, whose purchases
led the sector to new production records. Paradoxically, however, the growing
product quality is causing problems to some companies in the Mantova district: the
increasingly strong panty hoses have been causing a sales decline, with some
companies losing as much as 10-20% in the last few years. In addition, east
European manufacturers are entering the business with a production of lower
quality, which however is shrinking the Italian market shares in countries like
Poland, Romania and Russia. In order to oppose this sales drop, the Castel
Goffredo companies have changed their strategy, diversifying into the underwear
industry, where they make seamless garments using a micro-fibre very similar to
that used for hosiery. It is obviously a very expensive switch that requires capital
investments in machinery, research work in materials and communications, which,
however, has been successful for the district's larger companies, for some of which
this new line of products already accounts for 40% of the total sales revenue.
Even better known than Castel Goffredo are the two industrial districts where
most of the Italian manufacture of brass and bronze valves and fittings is
concentrated: one is located in north-east Piemonte, around Orta Lake, astride the
provinces of Novara and Vercelli (Valsesia) with some branching into Verbania;
the other one, with slightly lower sales revenues and workforce, located around
Lumezzane, in the province of Brescia.
The origin of the Cusio-Valsesia district of valves and fittings manufacture lies
in the long tradition of brass and bronze processing, which dates back to the first
bell casting activities at Valduggia as early as in the 15 th century. The valves and
fittings industry gradually developed around the villages in the low Cusio region,
Valsesia and the low-central Novara region in the last century, especially starting
from the '50s. In the last ten years, in particular, the district consolidated its
leadership in the international market, with a number of leading companies pulling
the whole district, giving a major contribution to its worldwide reputation thanks
to medium-high quality levels, a strong process and product innovation rate and, as
far as sanitary faucets are concerned, a high design level.
According to a balance sheet survey and questionnaires collected by the
Fondazione Comunita e Innovazione and the Research Centre in Economic
Analysis, International Economics and Economic Development (CRANEC) of
Universita Cattolica (Milano), in 1994 the Cusio-Valsesia district of valves and
fittings, without considering the allied industries, posted sales for about 1.850
billion lire of finished products, which, according to estimates based on the
projections of a smaller sample of businesses, rose to 2,200 in 1995, were more or
less unchanged in 1996, rose further to 2,400-2,450 billion a year over the 1997-
99 period and reached about 2,600 billion lire in 2000. Brass valves for sanitary
fixtures accounted for 43% of the sales revenues and chromium-plated sanitary
faucets accounted for the remaining 57%.
152 M. Fortis and M. A. Maggioni

With approximately 1,700 billion lire's worth of sales abroad, the Cusio-
Valsesia district accounts for approximately one third of the Italian exports of the
sector and has a share of about 5% in the world exports and of 9-10% of the EU's
exports of item 84.81, representing the most important valves and fittings
manufacturing area in the world. The shares are even larger if brass and bronze
valves and fittings only are considered, which is the sector the district is
specialized in. The exports are about 60% of the total sales, with some companies
reaching peaks of 80% or more and, in a few cases, almost the total sales. The
main exportation markets are the EU countries (Germany, Great Britain, France
and Spain in this order), the United States and the Middle East, with growing
exports to Eastern Europe, the Far East and South America in the last few years.
According to 1ST AT's intermediate industry census, in 1996 the Novara,
Vercelli and Verbania provinces recorded 431 companies in the val ves and fittings
sector, employing 8,101 workers. More difficult is the assessment of the impact of
the allied industries because many activities (e.g. metal treatment and polishing)
are shared with another industrial district in the area, namely that of cookware and
household goods, concentrated in the Omegna area. It is a fact, however, that,
again in 1996, the moulding, casting, metal treatment and polishing businesses in
the three mentioned provinces employed a workforce of 6,897 distributed over
1,139 local units, a lot of them very small in size, and most of them are more tied
to the valves and fitting business than to that of household goods. It can therefore
be estimated that the Cusio-Valsesia district of valves and fittings involves not less
than 12,000 jobs in direct activities and allied industries, excluding services such
as transport, advertising, etc.
The growth of the valves and fitting sector in the Novara, Vercelli and Verbania
provinces has been very fast: according to the 1ST AT censuses, the jobs were
3,228 in 1971,5,390 in 1981,6,684 in 1996 and, as already said, 8,101 in 1996.
Based on a projection of the figures of the top 44 companies, the direct jobs are
likely to have risen to about 8,800 in 1999-2000. Equally significant was the
employment growth in the casting, moulding, galvanic treatment and polishing of
metals in the three provinces: the jobs rose from 3,146 in 1971 to 6,897 in 1996
and there is no doubt that most of the increase can be ascribed to the growth of the
valves and fitting business in the area.
Through the figures of the 1ST AT census of 1996 we can also assess the district
from the viewpoint of the size of the businesses. Considering only the valves and
fittings sector proper (i.e. without the allied industries), we find: 3 big companies
(with more than 250 workers), 34 medium-sized enterprises (50 to 249 workers),
132 small enterprises (10 to 49 workers) and 262 micro-enterprises (less than 10
workers). Again according to the 1996 ISTAT census, the main manufacturing
centre in the district is San Maurizio d'Opaglio (see Figure 6), with about 2,000
workers employed in the valves and fittings sector and allied industries, followed
by Gozzano (about 1,500 workers plus allied industries), Pogno (1,000 workers),
Valduggia (900 workers) and Pella (500). Other important municipalities in the
district are Invorio, Borgomanero, Briga Novarese, Fontaneto d' Agogna,
Borgosesia, Quarona and Grignasco.
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 153

30
Kms

30 Kms

Fig. 6. Epicenter of the Cusio- V alsesia district of taps and valves based on the % impact of
the local units specialized in the main business and allied industries on the total 19cal units
of each municipality: year 1996.

In addition, if the workforce figures of each municipality in 1991 and 1996 are
compared, a growth appears in the relative weight of the most external
municipalities as a result of new production settlements outside the ring of the
central core of the district's municipalities. As a result of this trend, though the
number of workers has been rising in the 5 central municipalities (Pogno, S.
Maurizio d'Opaglio, Pella, Valduggia and Gozzano), the relative impact of the
workers on the total workforce of the district decreased from 58% in 1991 to 56%
in 1996, while that of the peripheral municipalities rose from 42% to 44%.
The historical factors of success of this district include some elements common
to the two distinct businesses of sanitary faucets and valves. First of all, the
previously mentioned tradition of brass and bronze processing in the area. This
tradition has been developing over the years and has produced a very high specific
know-how in the manufacture of both valves and faucets, training of highly
qualified technicians and workers and setting up a strong network of relationships
inside the district for the supply of components, semi-finished parts and services.
Another strong point of the district as a whole is its proximity to the manufacturing
centres of the raw material, the brass bar, and the availability of a scraps recycling
system. In addition, the leading role of the bigger manufacturers was a major
154 M. Fortis and M. A. Maggioni

factor in building the world reputation of the district and the quality of the output,
which meet the highest international standards. In the sector of sanitary faucets,
design is an important element for success, on which the district concentrated great
efforts to be able to compete with the big foreign manufacturers. Compared to the
valves sector, that of sanitary faucets has seen the proliferation of small businesses,
sometimes only assemblers, whose development was possible thanks to the strong
presence of component manufacturers and providers of services (galvanizing,
polishing).
Let's review now the Lumezzane and Brescia district in Lombardia, which is a
bit smaller than that of Piemonte but equally important in the international market.
Located north of Brescia, along the Trompia and the Sabbia valleys, this industrial
district is mainly concentrated in Lumezzane, the pivot of the district, Sarezzo and
Villa Carcina, with branches both to the north through Marcheno and Gardone Val
Trompia and to the south as far as Concesio and Brescia itself.
As highlighted for the Piemonte district, also here tradition was a decisive factor
in the birth and development of the industrial district. Iron working flourished here
thanks to the abundance of minerals and of water already in ancient times. It was
arms manufacture that made these valleys well known as early as in Roman times.
Later, in the course of the centuries, the manufacture of farming tools, cutlery and
crockery developed and, more recently, that of brass and bronze valves and
fittings. In addition, the old tradition of iron and metal working in the Brescia
province allowed this district to specialize in the upstream steps of the
manufacturing process, i.e. in casting and moulding of the raw material (brass bar),
with the valves sector prevailing over that sanitary faucets.
In order to complete the analysis of the Italian valves and fittings industry, the
Fondazione Comunita e Innovazione and Universita Cattolica (Milano) studied
this second industrial district with the same methodology as used for the other one.
An initial quantitative assessment of the district was obtained through an extensive
survey made in 1995 by means of questionnaires sent to the enterprises and an
analysis of the balance sheets of the bigger companies operating in the sector in the
Lumezzane and Brescia area: the total sales revenues in 1995 amounted to 1,670
billion lire and the exports to over 850 billion. In addition, the intermediate 1ST AT
census of 1996 recorded for item 29.13, "manufacture of valves and fittings", 185
local units and 4,531 workers in the Brescia province. For the following years,
instead, the figures were updated on the basis of a sample of 35 companies, 19 of
which manufacturing valves, 8 faucets and 8 intermediate components. From the
data of the sample it was possible to estimate the total figures of the district both
for 1994, i.e. one year before that of the first survey, which was 1995, and for the
following years until 1999. Starting from 1,350 billion of sales revenues in 1994,
the Lumezzane district recorded a fast growth (+23%) and reached 1,670 billion in
1995; the 1996 figure was slightly lower (1,650 billion), while increases to 1,890
billion (+14%) and 1,950 billion (+3%) were recorded in 1997 and 1998,
respectively. The latter level was maintained in 1999.
The same trend was recorded in exports, which exceeded the 1,000 billion
threshold in the last few years. Like the Piedmontese district, that of Lumezzane is
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 155

a strong exporter: on the whole the ratio of exports to total sales revenues is about
50%, but a large number of enterprises reach much higher percentages, with peaks
of 98%. Considering the sectors, instead, the impact of exports on sales revenues is
higher for sanitary faucets (60% on the average) than for valves (50-55%) and
intermediate components (about 35%). The last-named sector, in particular, is
represented by the enterprises that operate in auxiliary processing such as metal
polishing, colouring or chromium-plating, or manufacture intermediate
components rather than finished products. For all these firms, usually small in size,
the outlet is the local market, which explains the smaller percentage of sales
abroad.
As mentioned before, unlike the other district, that of Lumezzane is more
specialised in valve manufacture. The enterprises operating in this sector are many
more than those manufacturing sanitary faucets, and account for 60% of the sales
revenues of the whole district. Some of the bigger valve manufacturers, in
addition, take care also of the initial step of the manufacturing process, i.e. casting
and moulding of the brass bar; these are capital intensive operations which require
the installation of casting or moulding works, so the companies having such plants
tend to use them also for other companies in the sector, to which they sell raw
product for subsequent processing.
Finally, concerning the allied industries, it is very difficult to estimate the
number of workers employed in allied industries because of the overlapping of the
valve and fitting sector with other mechanical industries present in the area, such
as that of cookware manufacture.
At this point, after the women's hosiery district of Castel Goffredo and the two
valve and fitting districts, we are presenting briet1y the case of the Varese
province, with special reference to the plastics material processing industry and
related machinery. The chemical, plastics and rubber industry is undoubtedly a
major sector in this area of Lombardia; it is a broad and diversified business that
ranges from chemical commodities to pharmaceutical products, from semi-fir.ished
to tinished products. Apart from the drugs industry, that has been developing
especially thanks to multinationals that selected the Varese province for the
existing technical and production background and its favourable location between
Milano and Switzerland, it is the manufacture of rubber and above all of plastics
articles that saw the biggest growth, making it one of the Italian areas with the
highest concentration of plastics conversion companies and a real industrial
district. To give an idea of the extension of this sector, in 1996 (intermediate
1ST AT census figures) the enterprises operating in the "manufacture of rubber and
plastic articles" in the Varese province were 831, with a total workforce of about
11,900 and a research conducted by U ni versita Cattolica of Milano in 1998
estimated that the 200 bigger companies in the district posted sales revenues of
about 3,400 billion lire.
Historically, the conversion of plastic materials and rubber entered the industrial
system of the Varese province immediately after World War II, but the real growth
was recorded in the last two decades, with the breakthrough of plastics in the
manufacture of intermediate and finished product. A major factor allowing the
156 M. Fortis and M. A. Maggioni

birth and consolidation of this industry, besides the diversitication of the expertise
available in the region and the presence of a few big companies operating in the
sector, was the strong concentration in the province (the highest in Italy) of
manufacturers of calenders, i.e. the equipment that - starting from the raw material,
here plastics - makes sheets of the desired thickness for subsequent moulding. The
manufacture of plastics conversion equipment is in line with the long mechanical
tradition of the Varese province. The mechanical industry, that was initially a
support for the textile industry, especially for maintenance of looms of British or
German origin, has been expanding to become one of the main businesses in the
province. It is emphasised that the mechanical industry of this province has been
closely tied to the history of the means of locomotion: from the early Caproni,
Marchetti and Macchi airplanes to the Isotta Fraschini cars, the Cagiva motorbikc:s
and the Agusta helicopters. Now the most widely spread and qualified mechanical
production is tied to the conversion of plastic, a sector, again according to the
1ST AT census of 1996, counting 452 local enterprises with a workforce of over
7,300. With the mechanical industry it is more difficult to quantify the production
value because the mentioned Universita Cattolica research surveyed only a very
small portion of the district's companies, but it estimated overall sales revenue of
about 650 billion lire for the 40 leading equipment manufacturers in 1998.
Therefore, on the whole, the Varese district of plastics and related mechanical
industry consists of about 1000 local units with over 19,000 workers.

3.2. The Data

The data used in the following section concern the level of employment and the
number of local units in different "clusters" from 1951 to 1991 (for some districts
data for 1996 are also included). Data are taken from the "Censimento delle
attivita produttive", the Italian businesses' census which is run each 10 years42.
The term cluster here identifies the intersection of a given industry with a given
territorial area (Provincia). It is therefore clear that in the empirical analysis we are
counting employees and local units which do not strictly belong to the cluster
(which is a subset of the provincial territory) but data at a finer geographical
disaggregation (Comune) are available only from 1991.
Data thus suffer from two main problems: the first, as explained above, concerns
the imprecise identification of the cluster, the second regards the small numerosity
of the available sample (6 observations at most). A similar analysis has been
conducted on US high-tech cluster in Maggioni (1999). There, by using County

42 1996 is the first attempt to build an "intermediate census" to shorten the time lag from 10
to 5 years.
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 157

Business Patterns, it has been possible to get more precise identification, yearly
observations and longer time series (about 40 observations).
The empirical exercise has therefore been forced to use a combination of
econometric analysis and calibration techniques. Some results were intrinsically
fragile and statistically not very significant. Further research on this issue is still
needed. We therefore decided to present only a subset, which yields the more
robust and significant results.

3.3. Econometric Analysis and Calibration Techniques

The analytical formulation of the abovementioned interactions when the number of

interacting entities is equal to 2 is as follows:

(13)

where Ilr and Ils are the "economic masses" (number of plants or level of
employments) in cluster rand s and all parameters are as in equation (7b). In
particular ar and as are the intrinsic rates of increase of each cluster in isolation;
arr and (Iss are the intra-cluster competition parameters which reflect the inhibiting
effects that a firm's entry has on the growth rate of the same cluster (because of
congestion effects); ars and asr (inter-cluster competition parameters) show the
inhibiting effects that one firm, locating in a cluster, has on the growth of the other
region (inter-cluster competition parameters).
In theory, one then would like also to take into account the inter-industry
interactions which develop within the same cluster. If this is the case, then a two-
regions, two sectors can be modelled as follows

dl1~ _ ( i
dt- - ar - iii i j j)
a rr l1 r - a,..,.n, + a r I1r I1r
i

dn: _
dt - (i iii i
a, - assn, - a,.,.llr + as 11.,
j j) 11.,i
(14)
dll; _ ( j
dt j j
- a r - arrll j j
r - arsn,. + arll r llr
ji i)
dn;_(j jj jj ii)j
dt - a, - assn,. - asrn r + asns 11.1'
158 M. Fortis and M. A. Maggioni

where the subscripts r, s refer to areas, and the superscripts i, j refer to

industries.
It appears evident that, even though this formulation is linear, and expresses the
agglomeration diseconomies without exogenously determined parameters, its
applicability is limited by the large number of parameters which have to be
estimated.
We thus proceeded by regressing the growth of the economic mass of a given
cluster (i.e. a couplet industry-provincia) on the more relevant interaction
coefficients (these being either inter-regional or inter-industry) as derived from
theoretical a-priori.
We therefore decided to test two sort of interactions:
- the interactions between two districts in the same industry (looking for signs
of territorial competition) namely: Cusio-Valsesia versus Lumezzane, for the
taps and fitting industry and Vigevano versus Macerata and Ascoli Piceno for
the footwear industry;
- the interactions between two industries in the same cluster (looking for signs
of competitive or synergetic inter-industry relation). In particular we focus on
the relation between the sector producing a final good (these being plastic and
rubber products or footwear) and the sector producing machinery specialised
for that particular type of good. The two chosen area are: Varese for the
plastic and rubber/specialised machinery sector and Vigevano for the
footwear/specialised machinery sectors.
From the regression we got estimates for the coefficient of the system equation
(13). By simple calculations it is then possible calculate the coefficients of the
isocline for each population, i.e. the combination of the economic masses of the
two populations which ensure that the population under examination do not varies
over time.
Analytically one must find the solution of the following equations:

dnl =0 and dn2 =0

dt dt

which have a trivial solution for n l and 112 equal to zero and a non trivial soiution
which identify the following two lines:

n2 = ---
a l
- -all- n l
a 12 a lz
a2 a ZI
n2 = --- - --nl
a 22 a 22

and plot them in the phase plan. The intersection of two isoclines is an
equilibrium point for the system, since in this point both populations do not change
overtime.
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 159

If one then plots in the same space the actual measure of the two populations in
the cluster one may suggest the future evolution of the cluster buy predicting the
direction of variation of the two carrying capacities. Figure 7, 8, 9 10 and II
present the main results.
When the econometric estimates were either fragile or not particularly
significant or made impossible by the small size of the sample, we used an
alternative technique which implies the determination of the coefficients for each
equation in system (12) by substituting the variables' value for dit1erent triplets of
years and moving the triplets down the observations in order to "test" the stability
and robustness of these coet1icients to changes in the years-base.
Figure 7 and 8 represent the synergetic relationships existing between the plastic
and rubber industry and the specialised machinery sector in Varese (where the
measured variable is the number of plants) obtained through both calibration and
econometric estimates.
Both techniques show the existence of synergetic relationships: the development
of one sector fosters the development of the other one. However the two pictures
differ in the dynamical structure. The estimates diagram (Figure 7) shows a saddle
point equilibrium (approximately 350 machinery plants and 650 plastic and rubber
plants) and suggests that while the presence of the machinery industry is vital for
the existence of the plastic and rubber producers (which buy most of their
equipment within the cluster) the relevant market for the machinery industry
(which historically developed after the plastic and rubber one) is the world;
therefore the connections with local producers of plastic and rubber products arc
convenient but no more necessary.
The calibration diagram (Figure 8) suggests a pure symmetric synergetic
interactions in which the carrying capacity of both population grows indefinitely.
Figures 9, 10 and II show the other side of intra-cluster relationships. Figure 9
shows a competitive interaction between footwear industry and the specialised
machinery in Vigevano, when the economic mass is measured by the number of
plants. Figure 10 and II show predator-prey relationships between these two
sectors. The lung run forecasting for this sectors in both cases shows the
development of the machinery sector at the expenses of the footwear industry.
160 M. Fortis and M. A. Maggioni

ijhin
I 400 1 Situation in
1996

/',Machinery 0
~/'"'
I!J ./
200 ~
~
~~ ~'tic=O
_- !!.t

I J,lastic
100 200 300 400 500 600 700 BOO:

Fig. 7. Long run evolution of Plastic and Machinery in Varcsc, number of plants
(estimates).

mac in

1000

Situation in 1996

Fig. 8. Long run evolution of Plastic and Machinery in Varese, number of plants
(calibration).
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 161

tlmachinery 0
I'!.I

I'!.ju(}twear = 0
ill
100
1 Situation in 1996
I" ~---.---....--.-.---.----.--.~, - - . _ .........- .. .

1000 2000 3000 4000

Fig. 9. Long run evolution of Footwear and Machinery in Vigevano, number of plants
(estimates).

o tw.

15 0

l!.machinery =0
l!.t

lfrxxa·etJr =0
Situation in tJt
1991

[0 10 A S ~ ----------
3x10 AS 4x10AS

Fig. 10. Long run evolution of Footwear and Machinery in Vigevano, employees
(calibration).
162 M. Fortis and M. A. Maggioni

!::\:<""jootw
I. 80~
\ !:fo()twel1~o

II 600 &

I! Situation in 1991

----_-]i
, -400. I

I 200 ~~_ _
~ _ _-~ -------~--~---------.-.-- D.ma~inery = 0 machin.
I _.
o YX10'4 4xlO'4 6xlO'4 8:<10'4 10'5 1.2xlO'5 1.4xlO'5

Fig. 11. Long run evolution of Footwear and Machinery in VigeVJ.no, employees
(calibration).

4. Conclusion

The paper has shown the relevance of industrial cluster of SMEs in the economic
structure of Italy. The analysis of such a phenomenon. and of its most peculiar
organisational model, the industrial district, have often be based either on isolated
case studies, which were very difficult to compare, or on cross-sectional
econometric analyses, which obtain some general results at the expenses of much
of the individual cluster specificities.
The use of population ecological models may solv(! some of these problems
being explicitly devoted to the analysis of inter-industry and inter-cluster
relationships. The estimation of such models entails several methodological
difficulties which are made worst by the limited availability of data. However, if
this paper succeeds in getting scholars' attention on these techniques and make
them part of the toolbox of industrial economists and the regional scientists, we'll
have reached our main target.
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 163

References

Arthur, W. B. (1988), "Urban System and Historical Path Dependency", in: Cities and
Their Vital Systems, Ausubel, J. H. and Herman, R. (eds),Washington D.C., National
Academy Press, pp. 85-97.
Arthur, W. B. (1990), "Silicon Valley Locational Clusters: When Do Increasing Returns
Imply Monopoly?", Mathematical Social Sciences, 19, pp. 235-251.
Beltrami, E. (1987), Mathematics for Dynamic Modelling, Boston, Academic Press.
Capelo, A. C. (1989), Modelli matematici in biologia, Padova, Decibel.
Dendrinos, D. S. and Mullally, H. (1985), Urban Evolution. Studies in the Mathematical
Ecology ofOties, Oxford, Oxford University Press.
Fortis, M. (1998), Il "made in /taly" - Quando stile e creativitii non sono solo moda,
Bologna, II Mulino.
Fortis, M., Clerici, A. and Nodari, A. (1999), "II distretto della rubinetteria-valvolarne del
Piemonte Nord Orientale: Cusio e Valsesia", in: Aree distrettuali prealpine - Meccanica,
lessile, gomma e plaslica, Fortis, M. (cd.), Milano, Franco Angeli.
Fortis, M. and Nodari, A. (1999), "II distretto bresciano della rubinelleria-valvolarne:
Lumezzane", Fortis, M. (ed.), Aree distreltuali prealpine - Meccanica, tessi/e, gomma e
plastica, Milano, Franco Angeli.
Fortis, M., Bassetti, G. and Nodari, A. (1999), "II distretto mantovano delle calze
femminili: Castel Goffredo", in: Aree distrettuali prealpine - Meccanica, tessile, gomma
e plastica, Fortis, M. (cd.), Milano, Franco Angeli.
Fortis, M. and Nodari, A. (200 I), The Role Played by the Industrial Districts in the ltallan
Valve, Tap and Fitting Sectors, paper presented at CEIR (Comitato Europeo Industrie
Rubinetterie) General Committee, Venezia, 11-12 June.
Gambarotto, F. and Maggioni, M. A. (1998), Regional Development Strategies in
Changing Environments: an Ecological Approach, Regional Studies, 32, pp. 49-61.
Gause, G. F. (1934), The Struggle for Existence, Baltimore, Williams & Wilkins.
[sard, W. (1956), Location and Space-economy, Cambridge (Mass.), MIT Press.
Kingsland, S. H. (1985), Modelling Nature. Episodes in the History of Population Ecology,
Chicago, The University of Chicago Press.
Levin, S. A. (1970), "Community Equilibria and Stability, and an Extension of the
Competitive Exclusion Principle", American Naturalist, 108, pp. 207-228.
Levins, R. (1968), Evolution in Changing Environments: Some Theoretical Explorations,
Princeton, Princeton University Press.
Lotka, A. 1. (1925), Elements of Physical Biology, Baltimore, Williams & Wilkins.
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, 11-14
November.
Maggioni, M. A. (1999), Clustering Dynamics and the Location of High-Tech Firms, Ph.D
Thesis, Coventry (UK), University of Warwick, July, forthcoming by Springer Verlag,
Heidelberg.
Maggioni, M. A. (1994), "Modelli ecologici per l'analisi della dinamica industriale
regionale", in: Modelli d'analisi e d'intervento per unnuovo regionalismo, Pasquini, E,
Pompili, T. and Secondini, P. (eds), Milano, Franco Angeli, pp. 79-105.
Maggioni, M. A. and Gambarotto, F. (1997), "Sviluppo locale, dinarniche globali e molo
dell'operatore pubblico: un modello ecologico", m: La dinamica dei sistemi produttivi
164 M. Fortis and M. A. Maggioni

territoriali: teorie, tecniche, politicize, Bramanti, A. and Maggioni, M. A. (eds), Milano,

Franco Angeli, pp. 280-306.
Maggioni, M. A. and Porro, G. (1994), Dinamiche di crescita regionale: il ruolo delle
aspettative in modelli di tipo ecologico, Quadernn della rice rca di base "Modelli di
sviluppo e regional competition ", Milano, Universitii. Bocconi, 6.
Maino, R. (llJ89), "Aree interne ed interdipendenze settoriali. Uno schema interpretativo
fondato su modelli dinamici del tipo preda-predatore", in: Aree illterne, tutela del
territorio e valorizzazione delle risorse, Becchi Collidii., A., Ciciotti, E. and Mela, A.
(eds), Milano, Franco Angeli, pp. 81-110.
Marshall, A. (1920), Principles of Economics, London, McMillan.
May, R. M. (1974), Stability and Complexity in Model Ecosystems, Princeton, Princeton
University Press.
May, R. M. (1976), Theoretical Ecology, Principles and Applications, Philadelphia,
Saunders.
Metcalfe, 1. S. (1994), "Competition, Fisher's Principle and Increasing Returns in the
Selection Process", Journal of Evolutionary Economics, 4 (4), pp. 327-346.
Miyao, T. and Kanemoto, Y. (1987), Urban Dynamics and Urban Externalities, London,
Harwood Academic Publishers.
Nijkamp, P. and Reggiani, A. (1992), Interaction, Evoilltion and Chaos in Space, Berlin,
Springer-Verlag.
Nijkamp, P. and Reggiani, A. (1998), The Economics of Complex Spatial Systems,
Amsterdam, Elsevier.
Papageorgiou, Y. Y. (1979), "Agglomerations", Regional Science and Urban Economics,
9, pp. 41-49.
Pearl, R. and Reed, L. J. (1925), Skew-Growth Curves, Proceeding of the National
Academy of Natural Sciences of the USA, II, pp. 16-22.
Quadrio Curzio, A. and Fortis, M. (eds) (2000), II made in Italy oltre it 2000 - innovazione
e cOlllunita locali, Bologna, II Mulino - coli ana promossa dalla Fondazione Comunita e
Innovazione.
Quadrio Curzio, A., Fortis, M. and Maggioni, M. A. (eds) (1996), I distretti economici
delle prealpi e delle alpi centrali, Sondrio, Credito Valtellinese - Collana socio-
economica.
Richardson, H. W. (1978), Regional and Urban Economics, Harmondsworth, Penguin
Books.
Roughgarden, J. (1979), Theory of Population Genetics and Evolutionary Ecology: an
Introduction, New York-London, Macmillan.
Storper, M. and Walker, R. (1984), "The Spatial Division of Labour: Labour and the
Location of Industries", in: SlInbeltlSnowbelt: Urban Development and Regional
Restructllring, Sawers, L. and Tabb, W. K. (eds), Oxford, Oxford University Press, pp.
19-47.
Storper, M. and Walker, R. (1989), The Capitalist Imperative. Territory, Technology and
Industrial Growth, New York, Basil Blackwell.
Swann, G. M. P. (1993), Can High Technology Services Prosper if High Technology
Manufacturing Doesn't?, Centre for Business Strategy Working Paper Series, London,
London Business School, 143.
Swann, G. M. P. (1998), "Towards a Model of Clustering in High-Technology Industries",
in: The Dynamics of Industrial Clustering, Swann, G. M. P., Prevezer, M. and Stout, D.
(eds), Oxford, Oxford University Press, pp. 52-76.
Tauchen, H. and Witte, A. D. (1983), An Equilibrium Model of Office Location and
Contact Patterns, Environment alld Planning A, 15, pp. 1311-1326.
 Competitive and Synergic Behaviours in the Development of Industrial Clusters 165

Verhulst, P. F. (;845) Recherches mathematiqlle sllr fa foi d'accroissement de fa

population, Nouveaux Memoircs de I' Academie Royale des Sciences et Belles-Lettres de
Bruxelles, 18 (2), pp. 3-38.
Volterra, V. (1926), Variaziolli e jlulluazioni del fllllllero di illdividui in specie di animali
COil vivel1li, Roma. Memorie Accadcmia dei Lincei. 2, pp. 31-113.
Weber, A. (1929), Theory of the Locatioll of Illdustry, Chicago, Chicago University Press.
On the Ubiquitous Nature of Agglomeration
Economies and Their Diverse Determinants:
Some Notes

Giulio Bottazzi, Giovanni Dosi and Giorgio Fagiolol

Abstract. This highly preliminary work attempts to study the multiple drivers of
agglomeration phenomena in contemporary economies and proposes a tentative
taxonomy where the conditions of knowledge accumulation, often specific-to-
specific locations and specific sectors, playa paramount role. We discuss the
achievements and limitations of current theorizing on spatial location of economic
activities, and we propose a simple model, which is estimated on Italian data,
highlighting the rich intersectoral diversity of agglomeration forces, together with,
in a few cases, also lack of them.

1. Introduction

The embedded ness of economic processes into underlying spatial dimensions

ought to be sufficiently straightforward not to require much further elaborations.
Indeed, spatial dimensions include both literally geographic aspects - related to the
physical locations of agents - and more metaphorical metrics - regarding e.g.
technological and institutional "distances"; mechanisms of inclusion/exclusion
between networks, organizations and, of course, nations; degrees of information
and knowledge sharing, etc. Having said that, it is equally easy to acknowledge
that the economic discipline is far from offering anything resembling robust
accounts of spatial localization of economic activities, or, even less so, their
underlying generating dynamics. Needless to say, there is no possible claim of any
systematic answer here to such tangled questions. Much more modestly, we shall
add a few further question marks, together with some hints on hopefully novel
interpretative conjectures (Indeed, in what follows, we shall somewhat indulge on
our naivete as newcomers to the field!). If space - however defined - matters, it is
also because particular "places" in it, persistently affect (i) identities, capabilities

lSupport to the research by the Italian Ministery of University and Research (MURST,
Project No. 2-13-2-E4099GD) is gratefully acknowledged. Roberto Monducci from the
Italian Statistical Office (ISTAT) has been, as usual, exceptionally supportive. Marco
Lippi, Carolina Castaldi, Deborah Tappi and several participants to the conference
"Complexity and Industrial Clusters: Dynamics and Models in Theory and Practice",
Milan, June 2001, organized by the Fondazione Comunita e Innovazione, have offered
precious comments.
168 G. Bottazzi, G. Dosi and G. Fagiolo

and behaviours of individual agents; (ii) interaction patters; and, ultimately, (iii)
individual and collective performances. In turn, this means that sheer geography.
together with institutional and technological specificities, ought to be studied in
their long-term effects upon economic structures and relative efficiencies. Even
when going along with many our fellow economists in (the admittedly despicable
practice of) blackboxing institutional diversities across nations and regions, one
should nonetheless be able to track intersectoral differences in the agglomeration
drivers exerted by technological factors. This is the first point that we shall
address below (Section 2). Second, an assessment - albeit quite telegraphic - of the
state-of-the-art of diverse strands of "economic geography" might help in t1agging
out achievements, standing shortcomings and challenges ahead (Section 3).
Theories on agglomeration (and dispersion) forces urgently demand stronger links
with empirical predictions. This is what we begin to explore in Section 4, while in
Section 5 we put our techniques to work on three - quite diverse - industrial
sectors on Italian data (i.e. primary metals, transport equipment and furniture).
Broader conjectures will be put forth in the conclusions (Section 6) of a report -
we want to emphasize - which is very much preliminary and "work-in-progress".

2. Empirical Evidence on Agglomeration Phenomena: On

Some Facts and Puzzles

A survey of the enormous empirical literature on agglomeration economics in

general and industrial districts in particular is well beyond the scope of these
notes 2• Here let we just mention four sets of empirical regularities.

2.1. Discrete Types of Agglomeration Structures

The first robust piece of evidence concerns the variety of agglomeration

phenomena yielding equally diverse "types" of local structures, including the
following broad classes.

1. Horizontally Diversified Agglomerations. They comprise a good deal of

the "made in Italy" districts, presenting remarkable ever-changing product
varieties generally produced by a multiplicity of small and medium firms
(e.g. clothing, textiles, jewelleries, tiles, etc.).
2. Agglomerations of Vertically Disintegrated Activities. Again, largely
overlapping with the former, they include a quite few "made in Italy"
districts whereby activities previously vertically integrated within
individual firms undergo a sort of "Smithian" process of division of labour
cum branching out of different firms. In some analogy with the old "pin

20n the Italian evidence about industrial districts, see, among the others, Signorini (2000),
Brusco and Paba (1997) and Onida et al. (1992).
 On the Ubiquitous Nature of the Agglomeration Economies and Their Diverse Determinants 169

story" of Adam Smith, division of labour and spatial agglomeration rest

upon: (a) economies of specialization; (b) input-output links; and (c) user-
producers exchanges of knowledge (see shoe making and textile/clothing
among others).

3. Hierarchical Spatially Localized Relations. They generally involve an

"oligopolistic core" and subcontracting networks (although not necessarily
mechanisms of technological dominance and rent-extraction by such a
"core"). In Italy, transport equipment, white goods, etc. are good cases to
the point.

4. Agglomeration Phenomena Based on Knowledge Complementarities - at

least partly fuelled by "exogenous science"- the "Silicon Valley" in the
U.S. being the most famous example. Incidentally, note that in Italy this
type of agglomeration is almost non-existent.
5. Agglomerations as Sheer Outcomes of Path-Dependence - for example due
to spatial inertia in the birth and death of firms - without however any
particular advantage of agglomeration itself3 .
Different types of agglomeration clearly hint at possibly different drivers of
agglomeration itself and their different sectorial specificities.

2.2. Intersectorial Differences in the Importance of Agglomeration

Economies

A second, related, set of empirical regularities concerns the intersectorial

differences in the revealed importance of spatial agglomerations. Figure 1
summarizes the statistical evidence on the contribution of districts production to
exports in Italy at quite high degrees of sectorial disaggregation 4 , highlighting a
characteristic skewed distribution. Complementary evidence on the contribution of
individual districts to the total Italian exports of the sectors in which they are
specialized (Figure 2) confirms the idea of a significant divide between a group of
"districts activities" and the rest of industrial production for which agglomeration
economies appear to be much less relevant 5 .

3This is indeed the thesis of Klepper (2000) concerning the role of the Detroit area in
automobiles.
4While the contribution to exports rather than production is admittedly less than perfect, it
allows - in Italy - those higher levels of disaggregation often corresponding to "industrial
districts" .
STable A 1 (see Appendix A) provides the complete list of 4-digit sectors accounted in
Figure 1 with the respective contributions by districts to the total Italian exports.
 _--
170 G. Bottazzi. G. Dosi and G. Pagiolo

. . . . .....
... I'

~I
'\. I
, I

~I
1,&01

I
l,E.{)2
-I

l,E~+-----------------------------------------------------~
l,DE+OD 1 ,CE+ 01 l,CE+O<
loc(Ra1K)

Fig. 1. Log of Districts' contribution to Total Italian exports by product category vs. Log of
Rank (Ateco 91 Classification), Year: 1996,
Source: ISTAT, Censimento Intermedio de\1'Industria e dei Servizi. See also App. A, Table
AI.
1,~OBr-------------------------------------------------------,

'- ",
' .. ...........

1,E+01

1,E+OD+-----------------~------------------__- - - - - - - - - - - -____~
1,CE-OB 1.= Icc(Ralk)
1,CE..Q1

Fig. 2. Log of Percentage Contribution to exports of specialized districts by ATECO micro-

sector vs, Log of Rank (Total Italian Exports by the micro-sector =100). (Ateco 91
Classification).
Source: ISTA T, Censimento Intermedio dell' Industria e dei Servizi,
 On the Ubiquitous Nature of the Agglomeration Economies and Their Di verse Determinants 171

2.3. The Importance of Agglomeration Maps into Diverse Sectorial

Patterns of Innovation

Third, the foregoing intersectorial differences in agglomeration economies

interestingly map into taxonomic differences in the sectorial patterns of
innovation, as proxied by Pavitt's categorization (Pavitt 1984t In particular, as
shown in Figure 3, agglomeration economies appear particularly relevant in
"scale-intensive sectors" - hinting at forms of hierarchical agglomeration
discussed above - and in "supplier-dominated sectors" - which tend to include
most of the so-called "made in Italy" activities? Conversely, they appear the least
relevant in "science-based" sectors.

"70 .. 0 ":,
..- . - - - - - - - . - . - - - . - - - - - -•.-.--.---.-.---.... -...--.----'-',

60" (I ( 1 j - - - : = l E : : : : : - - - - - - - - - - - - - - - - - - - - - '

10,0

Seale Supplier Specialized Science

Intensive Dominated Supplier Based
(3D) (SS) (SI) (SB)

Fig. 3. Mean of Percentage Contribution to exports according to Pavitt's Taxonomy.

2.4. Some Form of "Life-cycles" in Agglomeration Phenomena?

Widespread evidence supports some sort of metastability in agglomeration

phenomena, in the sense that in quite a few circumstances they persist on time-
scales of order of magnitude greater than those of the processes supporting them,
even though they tend to vanish on much longer time-scales. So, for example,
Brusco and Paba (1997) report on several industrial districts in Italy (especially
but not only in the South) which existed after World War II but disappeared

6The Italian ATECO 91 classi fication of sectors into Pavitt's taxonomy is available from
the authors upon request.
7"Supplier-dominated" is an unfortunate and somewhat misleading name for a set of
industries which might well be characterized by a lot of product differentiation (e.g. related
to fashion) and organizational innovations, but at the same time acquire most of their
technological innovations via intermediate and capital inputs produced elsewhere. Textiles,
clothing, furniture, lOys, etc. are good examples of such a set of industries.
172 G. Bottazzi. G. Dosi and G. Fagiolo

thereafter. At the international level, sectors like steel, automobiles, tires and
many others appeared quite spatially concentrated near the time of their birth, but
became geographically much more footloose. In fact, it could well be that
geographical stickiness falls whenever: (a) specific technological paradigms
become fully established; and (b) international oligopolistic firms emerge,
incorporating the core knowledge associated with production and innovations.
There are two general conjectures which stem from the foregoing evidence,
namely:
1. The drivers of agglomeration economies are often nested in the nature of
sector-specific patterns of knowledge accumulation.
2. Relatedly, cross-sectional differences in agglomeration forces ought to be
at least partly explained on the grounds of underlying differences in such
processes of technological and organizational learning. They affect
among others the relative importance of phenomena such as localized
knowledge spillovers; inter- vs. intra-organizational learning; knowledge
complementarities fuelled by localized labour-mobility; innovative
explorations undertaken through spin-oft's and, more generally, the birth
of new firms.
To what extent are such knowledge-related drivers accounted for in the current
literature? This is what we shall briet1y discuss in the next section.

3. Space, Geography and Agglomeration: Some

Telegraphic Comments on the State-of-the-Art

In a nutshell, one might identify four main questions that scholars concerned about
the "spatial dimension" of economic interactions have been all trying to address,
albeit from different perspectives, for more than a century, namely: (i) What is the
role of mere "chance" in the observed spatial concentration of economic
activities? (ii) Relatedly, why and when could one observe persistent spatial
patterns that cannot be explained by resorting to pre-existing heterogeneity in
agents and locations (i.e. by some kind of "comparative advantage theory" alone)?
(iii) Could one neatly identify agglomeration (centripetal) and dispersion
(centrifugal) forces lying at the heart of the processes generating sustained spatial
concentration (and possibly its destabilization)? And, (iv) how and when emerging
spatial structures of production and innovation tend to become self-sustained over
time? (And, conversely, what make them wither away?). As well known,
pioneering works such as Von Thunen (1826) and Marshall (1890), tried to
investigate the main economic forces driving geographical differentiation and
agglomeration. For instance, Von Thunen's simple analysis of land use - by
stressing the importance of space constraints in decentralized economies -
addressed the relationships between micro and macro geographical outcomes.
Even more importantly, Marshall's discussion of his famous "localization

8Some related remarks by one of us may already be found in Dosi (1982).

 On the Ubiquitous Nature of the Agglomeration Economies and Their Di verse Determinants 173

externalities" (or "external economies") triad 9 became a cornerstone in the theory

of economic agglomeration. From then on, however, diverse trajectories of
exploration emerged. A first large class of models hinges upon the basic idea that
many different spatial agglomeration patterns (from concentration of economic
activities in few locations to hierarchical structures) can be explained as the
solution of a static, well defined, trade-offs between identifiable agglomeration
and dispersion forces. This intuition, rooted once again in Von Thunen's work,
has become the core of the analyses provided by "central-place" theory developed
by Christaller (1933) and Losch (1940) - and more generally by "regional science"
models of Isard (1956) and by Henderson's (1974) treatments of urban systems.
More recently, it has inspired static models with non-market externalities such as
Papageorgiou and Smith (1983) and Fujita (1988, 1989). For instance, central
place theory stresses the importance of economies of scale in the process of
agglomeration of any economic activity and transportation costs in a community
of farmers, while Henderson (1974), represents the economy as a system of cities
and formalizes as an "inverted U" the relation between individual gains (due to
some form of Marshallian localization externalities) and individual losses (due to
some form of crowding effect) caused by concentrating activities in a single
location. However, in order to explain hierarchical features of central places (an
their supposed optimality) central place theory resorts to quite counterintuitive
ideas as "nested hexagonal patterns" (! !), while urban theory relies on the
existence of exogenous institutions such as city corporations. More elegantly,
Papageorgiou and Smith (1983) and Fujita (1988, 1989) envisage agglomeration
patterns as the outcome of the trade-off between some form of locally positive
informational spillovers arising among agents endowed by heterogeneous
information (e.g. social or technological spillovers) and congestion effects arising
in spatial systems when concentration is too high (use of limited land, commuting
costs, wage rate and land rent, etc.). A second class of models that has become
prominent in the last few years, now known under the perhaps misleading heading
of "New Geographical Economics" 10, acknowledges instead some form of
increasing returns to scale (or indivisibilities) at the level of individual agents as
both the incentive triggering agglomeration and the force able to sustain
concentration (once the latter has emerged). The basic challenge of this stream of
research - derived mainly from the theory rather than from any empirical puzzle -
was to provide a satisfactory treatment of increasing returns to scale and
monopolistic competition in a static equilibrium framework cum fully rational
agents. By bridging monopolistic competition models a la Dixit and Stiglitz
(1977) and Samuelson "iceberg-like" trade costs, such models have been able to

9That is: (i) backward/forward linkages associated to the trade-off between market-size and
market-access; (ii) informational spillovers and (iii) advantages of thick markets for
specialized local providers of inputs.
IOSee among others Krugman (1991 a, 1991 b, 1993), Krugman and Venables (1995a,
1995b, 1996), Venables (1996, 1998), Ottaviano and Thisse (1998), Fujita, Krugman and
Venables (1999), Puga and Venables (1996). See also Ottaviano and Puga (1998), Fujita
and Thisse (1996).
174 G. Bottazzi, G. Dosi and G. Fagiolo

account for agglomeration patterns and inter-Iocational specialization by positing

a self-reinforcement process - stemming from some form of market externality -
which finds its counterpart in dispersion forces caused by competitive pressures
implied by either agglomeration itself or immobility of some factors (e.g. labour).
Although the wide proliferation of models sharing this common framework does
not make easy to provide a taxonomy of both assumptions and results, the basic
argument 11 hinges upon a circular causation between: (i) the decision of a firm to
concentrate in a given area; and (ii) the positive net increase in profits enjoyed by
firms deciding to follow it thereafter. Indeed, localized market-size effects in
presence of imperfect competition usually offset the decrease in profits due to
fiercer competitIOn (unlike what would have happened under perfect
competition) 12. In a two-region, two-industries economy, this can account for the
emergence of core-periphery patterns with agglomerations sustainable as stable
equilibria and all (horizontally-differentiated) industrial goods produced in one
region. However, multiple equilibria arise and small asymmetries in initial
parameters can be amplified to give rise to very different spatial patterns. In
particular, crucial to equilibrium selection are the roles played by the degree of
immobility of production factors, transportation costs, degree of differentiation of
the horizontally differentiated industrial good and the importance of
indivisibilities in production. The most important contribution made by "Ne'W
Geographical Economics" models is perhaps having provided closed-form
solutions to the problem of describing in a common framework monopolistic
competition, transportation costs and increasing returns at the microeconomic
level. In fact, this same formal machinery (of variants thereof13) has been recently
applied to diverse open issues in spatial economics such as industrial clustering in
open economies (and its relationships with economic growth)14, the emergence of
the evolution of urban systems and cities formation l5 , industrial specialization in
an array of imperfectly competitive sectors l6 . However, any hope for analytical
tractability and clearly interpretable closed-form results strongly clashes with
many attempts to generalize the basic framework (see the case of many :ocations
discussed in Fujita and Thisse 1996; see also the discussion in Martin 1999). Even

II See in particular Krugman (1991 a, 1991 b).

12An alternative but complementary argument (see Venables 1996) stresses cost-linkages
instead of demand-linkages as the engine of agglomeration. Agglomeration in a given
region with the associated fiercer competition does not only lower revenues, but also
decreases costs due to the existence of input-output links between firms. Hence, the effect
on net profit can once again be positive.
13The basic model sketched above has been extended to allow for endogenous wage
determination in a general equilibrium setting (Krugman and Venables, 1995b; Pug a and
Venables, 1996); and for initial comparative advantages (Krugman 1993).
14See for instance Fujita, Krugman and Venables (1999), Chap. 4, and Englmann and Waltz
( 1995).
ISSee Fujita, Krugman and Mori (1999), and Fujita, Krugman and Venables (1999), Chap.
3.
16See Krugman and Venables (1996) and Venables (1998).
 On the Ubiquitous Nature of the Agglomeration Economies and Their Diverse Determinants 175

though these models have been dealing only with one element of the Marshall's
proposed triad 17 , thus neglecting issues as informational spillovers and the
advantages of thick markets for specialized local providers of inputs, they need to
rely on highly disputable oversimplifications in order to retain elegance and
analytical tractability, such as the special Dixit-Stiglitz framework itself and
Samuelson's scheme of transportation costs (i.e. no description of a separated
transportation sector), to name a few ls . In our view, the commitment to
mainstream formalism of an "equilibrium cum fully rational agents" framework
(and to analytical solutions) has strongly limited the explicative power of "New
Geographical Economics,,19. In fact, notwithstanding verbal acknowledgment of
the very dynamic nature of any spatial agglomeration processes (Fujita et al.
1999), the treatment of dynamics in their models is admittedly unsatisfactory. So,
instead of following their own mainstream prescriptions and endogeneizing
dynamics as the outcome of intertemporal maximization problems by economic
agents, Fujita et al. (1999) discuss issues of stability and selection among multiple
equiiibria in an essentially static framework. Finally, the list of empirically
testable implications that such models are able to provide is intrinsically quite
poor and always of an indirect nature (Martin 1999, p. 70). Even more
dramatically, "New Geographical Economics" formalizations do not lend
themselves to generate testable implications about industry distributions across
locations, so that there is very little hope that any predictions about e.g. rank-size
relationships could be ever taken to the data employing such models. Despite
"New Geographical Economics" has often been claimed to be the most prominent
attempt to provide a unified framework for spatial economics (as well as the sole
robust answer to the resurgence of interest in agglomeration issues recently
prompted by real-world concerns such as European market integration), there
actually exist at least three other broad streams of research that in the last decades
have been trying, in partly complementary ways, to open up the black box of
spatial issues in economics with quite eclectic theoretical spectacles. First, many
scholars have been providing rich and qualitative analyses, mostly (but not
entirely) empirically focused, of urban/regional development and industrial
agglomeration phenomena. In particular, by thoroughly analysing the role of
externalities and technological spillovers, as well as the importance of social,
cultural and institutional forces in shaping the rise and the decline of industrial
districts, these authors have been providing a huge amount of illuminating insights

17That is the backward/forward linkages associated to the trade-off between market-size and
market-access.
18Notice also that these models do not allow for any treatment of uncertainty. On the
contrary. static models based on spatial competition (i.e. the so-called "shipping" and
"shopping" models) introduced some uncertainty in consumers' perceptions of payoffs
from locational choices. See e.g. Anderson, de Palma and Thisse (1992).
19 A similar point about "New Geographical Economics" has been also made by Martin
(1999). However, we do not agree with his analysis of Polya-Urns type models (Arthur
1994). as the latter class of formalizations clearly avoid any commitment to concepts like
full rationality and general equilibrium. See also below.
176 G. Bottazzi. G. Dosi and G. Fagiolo

about specific examples together with some attempts to offer interpretative

frameworks able to grasp the larger picture 2o . Second, a long stream of literature
on multinational investment - from the pioneering works by Vernon (1966), all the
way to the recent contributions by Cantwell and colleagues 21 - is rich of insights
on the interaction between technologies, corporate strategies and locational
features. It is indeed surprising that geography-centred investigations have largely
neglected such complementary contributions. Third, building on the seminal work
by Brian W. Arthur and Paul David, a more theoretically grounded literature has
been attempting to analyse the nature of economies/diseconomies of
agglomeration in a truly dynamic framework in which persistent spatio-temporal
patterns are conceived as emerging out of direct interactions among very stylised,
boundedly-rational, heterogeneous economic agents. By acknowledging the
history (or "path") dependent nature of the observed uneven spatial distribution of
economic activities, the basic argument stresses the importance of dynamic
increasing returns implied by some form of agglomeration
economies/diseconomies (Arthur 1994, Chs. 4 and 6) and/or local network
externalities (David et al. 1998; Cowan and Cowan 1998). Even more importantly,
by recasting the analysis of agglomeration phenomena in a truly dynamical
setting, one is able to appreciate the subtleties of the trade-offs 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. Without entering into the mathematical details 22 , the basic argument
envisages a discrete-time economy with a finite set of regions (say R) and an
enumerable population of firms deciding where to locate. To keep things as simple
as possible, assume that firms i = 1, 2, enter sequentially the decision stage 23 .
Firm i entering at time t has an idiosyncratic perception of the gain associated to
the choice of locating in region r = 1, .. , R, equal to the sum of an intrinsic, time-
independent, attractiveness term q{ (e.g. the ex-ante geographical benefit) and
some function g of the number of firms YiT(t) that have so far decided to locate
there (i.e. a measure of agglomeration economies - if g' > 0 - or, respectively,
diseconomies - if g' < 0). Assume that firms choose to locate in the region
associated to the best perceived gain. If the intrinsic attractiveness terms are
randomly drawn from some given distribution F, then one can easily work out the
function mapping current regional shares into the probability that each region will
be chosen next. This extremely simple framework (and extensions thereot) can

20We refer here to a huge body of literature covering both "economic geography" studies
(see Lee and Willis 1997 for a survey; and the references in Martin 1999). and in particular.
the Italian studies on industrial districts, see Antonelli (1990. 1994), Brusco (1989). Sforzi
(1989), Becattini (1990). Brusco and Paba (1997). Signorini (2000), Tattara (2001).
21See e.g. Cantwell (1989) and Cantwell and Iammarino (1998).
22See however Arthur (1994), Dosi, Ermoliev and Kaniovski (1994) and Dosi and
Kaniovski (1994) for more detailed discussions.
23The assumption of sequential one-time decisions is not actually crucial. See e.g. David et
al. (1998) for an example in which firms are allowed to revise their current choice from
time to time.
 On the Ubiquitous Nature of the Agglomeration Economies and Their Di verse Determinants 177

indeed provide a rather wide array of predictions, in particular concerning the

ability of economies/diseconomies of agglomeration to shape the long-run
concentration patterns. Together, it is also able to account for early, small, mainly
non-predictable, events as they interact with more systematic forces in conveying
observable structures. Despite their highly stylised nature (firms are indeed
conceived as very naIve entities; microeconomic foundations are rather poor; etc.),
this class of models begins to open-up the black box of spatial phenomena in
economics by focusing on inherently dynamical decentralized systems populated
by simple interacting agents. To summarize: multiple strands of theoretical and
historical literature do highlight the renewed richness of the investigation of
spatial phenomena in economics. However, it is also fair to say that major
persistent shortcomings of the theory concern, at the very least: (i) thorough and
relatively general accounts of the interaction patterns between forms of knowledge
accumulation and types of agglomeration phenomena, and (ii) the ability of
yielding robust empirical predictions concerning agglomeration patterns
conditional on underlying technological and organizational characteristics of
diverse industrial activities. Let us now turn to the latter issue and suggest a
formal machinery able to detect the different revealed strength of agglomeration in
different sectors.

4. A Stochastic Dynamical Model of Plants Location

As mentioned in the foregoing section, a class of models aimed at empirical

predictions on the grounds of explicit dynamics of firm location is the one
presented in Arthur (1994, Chs. 4 and 6). There are however some drawbacks in
such a methodology. First, the prediction of this type of model generally concerns
the asymptotic state of the system, i.e. its state when an infinite number of firms
has chosen its location. The comparison of different asymptotic outcomes does
represent a valuable theoretical method to compare the effect and the relative
strength of the different determinants of the location dynamics (agglomeration
economies, scale economies, etc.). Nevertheless, it is obvious that its predictive
power concerning actual empirical distribution should be taken with some caution
since one is often facing empirical processes involving a relatively small number
of firms (so that "infinity" might well be a misleading approximation). Moreover,
the mathematical tools based on Polya-like processes are particularly well-suited
to the description of sequential entry settings where each decision is primarily
influenced by the choices of earlier entrants: due to its infinite memory, this
machinery does in fact provide an elegant formulation of "history-dependant"
processes. However, it is less suited to all circumstances where individual
decisions are much less irreversible and/or the stochastic component of the
process does not tend to zero (as in Polya dynamics), due to persistent entry and
mortality. Here, in order to account for these drawbacks, we shall explore a
distinct (Markovian) framework wherein, first, we will consider a finite number of
firms and locations and second, we will describe industry dynamics as belonging
178 G. Bottazzi, G. Dosi and G. Fagiol0

to an invariant dynamical process, That is, we shall try to capture the idea that the
actual distribution of plants on the territory keeps changing with the passing of
time, even if generated by a stationary probability distribution, The probabilistic
character of our model, when it comes to firms location choices, can be thought as
taking in account the plausible existence of different "unobservable" constraints
that shape the locational incentives of different plants (or firms), Notice that the
ongoing displacement of plants (firms) can be thought of as the actual change of
location of a given production activity, as well as the death of a plant/firm in a
given location and the birth of a similar one in a different place. In order to
simplify the treatment, we consider the number of firms constant along the system
evolution. (Indeed, one can think to the number of firms as an "average" over the
period of observation 24 ). Suppose to have N firms distributed over M distinct
locations. Consider the occupation number vector

n = (n] ,... , nM ) (1)

which provides the number of firms belonging to each location. If, in a "heroic"
simplification, one assumes that all firms are identical, the vector n completely
specifies the state of the system. The dynamical evolution of such a system can be
described as a finite Markov chain. Let P{n'l n) be generic element of the
transition matrix, i.e. the probability that if the state of the system is n at time t, its
state at time t + 1 would be n'. This probability does not depend explicitly on time
and its specification completely defines our model. In order to capture the effect
produced on the distribution of plants by the presence of agglomeration
economies, the probability that a firm moves to a given location should depend on
the number of firms already located there. Moreover, it is straightforward to
introduce some degree of heterogeneity among locations: this can be done by
allowing for intrinsic "geographical benefits", in general heterogeneous across
locations (in analogy with Arthur 1994). Finally, for the sake of simplicity, we
assume that just one firm changes location at a given time 25 . At each time step, a
firm is chosen at random with equal probability liN and exits the industry. Then a
randomly chosen, new firm enters a location with a probability proportional to the
sum of the number of firms already there and a "geographical benefit" term. The
transition element thus becomes:

24Notice that if one is on the contrary interested in the actual time evolution of the
plant/firm distributions, the exact specification of the entry/exit dynamics becomes
mandatory.
25Notice that this assumption has no effect whatsoever on the form of the invariant
distribution of the process.
 On the Ubiquitous Nature of the Agglomeration Economies and Their Diverse Determinants 179

k :f: i
(2)
k =i

where a = (al, ... , aM) is the array of intrinsic "benefits" for the M locations, A =
I.~=I ak and 6., = (0, ... , 1, ... ,0) is the unitary vector with i-th component equal
to 1. Some considerations are in order. First, notice that the intrinsic benefit ak is
proportional to the probability of choosing location k when the latter is empt/6 .
Second, we have chosen the simplest linear relationship between the probability of
choosing a location and the number of firms already there27 • In fact, a more
realistic account of location-specific returns to agglomeration should allow for the
dependence of the a terms upon the number of plants/firms located there and/or
for threshold effects. However, as we shall see, even the foregoing simpler
approximation does not fare too badly with the data 28 . For positive a's, each
location has a positive probability of receiving the entering firm: hence, any
possible state of the system is reachable, in a suitable number of steps, with finite
probability starting from any other state and the Markov chain defined by (2) is
irreducible. If pen, t) is the probability to find the system in state n at time t, its
evolution reads:

pen, t + 1) = I. P(nln) p(n', t). (3)

n'

The invariant distribution 1[ (n, t) IS thus obtained by imposing the detailed

balance condition:

P(nln )1[(n ') = P(n'ln)1[(n) (4)

The explicit expression for the invariant distribution, known as the M-dimensional
Polya distribution, can be easily obtained and reads:

2~is model is known as Ehrefest-Brillouin model and has been introduced in Garibaldi
and Penco (2000) as a generalization of the famous Ehrefest model of statistical physics. A
similar simplified version has been introduced in Kirman (1993).
27This model overlaps with the Arthur's one with linear returns function, see Arthur (1994).
28Moreover, negative values for the a's can be in principle considered, in order to describe
the presence of agglomeration diseconomies characterizing the distribution of firms over
the different locations. Nonetheless, the purely random exit dynamics constitutes, as such, a
limit to the actual concentration of plants/firms in a given site, since more populated sites
are also more likely to yield dying ones.
180 G. Bottazzi, G. Dosi and G. Fagiolo

( ) r(N + l)f(A) ITIV! r(a + nJ

nna=-'----'-----'----.:...
l
(5)
, r(A+N) rca;)f(n +1)
1;1 l

The values of the parameters ak determine the nature of the distribution: for lower
values of the a parameters, the effect of agglomeration becomes more relevant In
the limit ak ~ + 00 and aJak ~ 1 for any i and k, "agglomeration economies"
disappear and the expression in (5) reduces to a multinomial distribution, while for
ak = 1, \j k it becomes what is known in statistical physics as the Bose-Einstein
distribution.

5. Some Empirical Evidence

The complete multivariate distribution in (5) does provide a complete probabilistic

description of our model. However, in order to obtain a quantity which can be
more easily compared with empirical data, it is better to consider the marginal
probability distribution of the occupancy number of a given site. The problem is to
compute the probability that a site with "intrinsic value" a would end up, after the
placement of N firms, with exactly h among them. By summing over all the
residual degrees of freedom, one gets (see Bottazzi 2001):

p(h; a, A, N) =(N] rea + h) rCA - a + N - h) rcA) (6)

h rca) rCA - a) rcA + N)

In the following we shall use this expression, whose parameter will be set by a
fitting procedure, in order to compare the prediction of our model with the
empirical observations. We shall use a database from the Italian Census of
Manufacturers for the year 1996 containing business units (BUs) belonging to ]l/I
= 784 "local system of labour mobility" (LSLM, for a definition see Sforzi 2000)
and to L = 25 different sectors for a total of 591,110 local units (plants). Here, we
present some experiments over three sectors - primary metals, transport equipment
and furniture - which one should expect to display different degrees of
agglomeration economies and different drivers of the latter (see also the
taxonomic discussions of Section 2). Let ni,l be the number of BUs in LSLM i
operating in sector l. Moreover we denote with n.,1 the total number of BUs
operating in sector l and with ni,. the total number of BUs belonging to i-th
LSLM. As already mentioned, instead of considering average quantities
measuring the "strength of agglomeration" of a given sector (as done, for instance,
in Sforzi, 1991) we shall analyse the complete "occupancy distribution" of the
BUs in the various LSLM, i.e. we compute the observed frequency (obs.(h;l) with
which a LSLM hosting exactly h BUs active in sector l appears in our data:
 On the Ubiquitous Nature of the Agglomeration Economies and Their Diverse Determinants 181

1 M
/ob, (h;l) = M L(\"h (7)
1=1

where 8 is the Kronecker (index) function, and we compare this expression with
the theoretical prediction of (6), once having of course specified the parameters a
characterizing the theoretical distribution. As a first benchmark, one could
consider all the LSLM as equal (i.e. with the same "intrinsic appeal") and obtain a
theoretical expression directly from (6) putting ai = and A = M f3 This f3 .
model would depend on a single parameter f3 which measures the "strength" of
the agglomeration effect - with a low f3
meaning high agglomeration economies.
However, tests of this model yield quite bad agreement with data, and the
theoretical description constantly underestimates the observed distribution tails.
The reason for this becomes apparent if one plots, for a given sector l, the number
of BUs ILi,l of a LSLM against the total number of BUs in all the other sectors,
except the one under consideration (that is ILi,. - ILi,I). Under the previous
assumption of a priori equiprobability, no dependence should appear between the
two variabks, since BUs belonging to different sectors should choose their
locations independently. On the contrary, a strong positive correlation appears
which contradicts the purported identity between the various LSLM's: we plot the
result of this analysis in Fig. 4, for the three chosen sectors. The parameters fitted
from a log-linear regression are reported in Table 1.

Table 1. The "agglomeration" parameter f3 and the slope a of the linear regression by
sector and by LSLM.

Primary Metals Transport Equipment Furniture

f3 10.00 0.32 0.50

a 0.91±0.01 0.70± 0.04 0.70± 0.01

As an alternative, let us assign to each location an "intrinsic attractiveness" which

is proportional to the number of BUs which are located there and belong to all
sectors but the one under analysis:

n -n
a = f3 I,.I,/ (8)
1,1 'N-n.,1

where l is the sector under analysis and the f3, coefficient captures, as above, the
intensity of agglomeration economies. This procedure meant to capture also
182 G. Bottazzi, G. Dosi and G. Fagiolo

"geographical" effects that make a location intrinsically preferable compared to

others, in terms of better industrial infrastructures, sheer overall size, etc .. It is
likely that these advantages are location-specific and apply to all sectors under
consideration. After controlling for "horizontal" locational effects, one can derive
a measure of relative advantages between locations. Following this idea, the
predicted "weighted" frequencies become:

(9)

~"vf
where p is the distribution in (6) and A,= L.J,=1 ai,' . As can be seen in Figs. 5
through 7, the accordance of the theoretical prediction with data is quite high. The
values of the f3's used for the theoretical curves are reported in Table 1. Out of
the three sectors chosen, the first, primary metals, does show an almost total lack
of agglomeration effects, while the other two, transport equipment and furniture,
seems to display rather high agglomeration economies. Notice, however, that the
nature of such an agglomeration is actually very different for the latter two
sectors: transport equipment displays a scattered locational patterns made-up of
relatively few firms (possibly hinting at the "hierarchical agglomerations"
mentioned in Section 2), while furniture highlights more "district-like" patterns,
 On the Ubiquitous Nature of the Agglomeration Economies and Their Diverse Determinants 183

11 11
t¥'-.-,. .
+
10 10 :!J)
++ -.j.
9 9 +
:j:I:"'+A
8 8 ++. + ,.01 1 4 ,,"

7
itt
-L
7
6
5 6
4 5
3 4
2 3
01234 567 8 9 (I 1 2 3 4 5 6

11
10
9
8
7
6
5 data ?Oi:-lt 3 +
4 li:1ear fit -----
...,3
...
(I 1 2 3 4 567 8 9

Fig. 4. Total number of plants in a LSLM (Local System of Labour Mobility) vs. the number
of plants in that location pertaining to a specific sector. a) primary metal, b) transport
equipment, c) furniture. All the variables arc in log scale.
Source: ISTAT, Censimento Intermedio dell' Industria e dei Servizi, 1996.
184 O. Bottazzi. O. Dusi and O. Fagiolo

O. C\.:l
data +
+ The or . "'ght.
0.035

(1.03
+
0.025

0.02

0.015

0.01

0.005

(I
1 10 100 1000
Fig. 5. Frequencies and number of tirms by LSLM (Local System of Labour Mobility). The
primary metal sector.
Source: 1STAT, Censi mento Intermedio dell' Industria e dei Servizi. 1996.
 On the Ubiquitous Nature of the Agglomeration Economies and Their Diverse Determinants 18S

0.14
data +
TheQr. Wght.
\,
--~---

0.12 ,
\
0.1
\
0.03 \

0.06
,

0.04
\+
\\+
\
0.02 " +
-t-x+

1 10 100 1000

Fig. 6: Frequencies and number of firms by LSLM (Local System of Labour Mobility). The
transport equipment sector.
Source: 1STAT, Censimento Intermedio dell' Industria e dei Servizi, 1996.
186 G. Bottazzi, G. Dosi and G. Fagiolo

0,08
data +
, The or . !·'ght.
0.07
,,
,,
0.06 , +
0.05
't\
';.;
+
0.04
-...x, ++
0.03 \<., +
"x,

0.02
'''-t

0.01

0
1 10 100 1000
Fig. 7. Frequencies and number of tirms by LSLM (Local System of Labour Mobility). The
furniture sector.
Source: ISTAT, Censimento Intermedio dell'Industria e dei Servizi, 1996.

6. Conclusions

We mentioned in the Introduction the highly preliminary nature of these notes. In

this spirit, we have tried to, first, flag out some taxonomies of agglomeration
structures and agglomeration drivers; second, suggest some links between the
observed patterns and the underlying dynamics of knowledge accumulation; third,
identify some related achievements and limitations of current theorizing; and,
finally, fourth, develop a formal machinery able to statistically characterize the
revealed intensities of agglomeration forces in different sectors. Indeed one could
consider such a contribution as a small step toward bridging "spatial" analyses on
the one hand, and investigations of knowledge-driven patterns of industrial
change, on the other.
 On the Ubiquitous Nature of the Agglomeration Economies and Their Diverse Determinants 187

Appendix A. Background Evidence on the Distribution of Exports by Sector and

by District.

Table A 1. Rank of Italian Districts contribution to export by product category (Ateco 91

Classification). Year: 1996.
Source' ISTAT Censlmento Intermedio dell' Industria e dei Servizi
Rank Product category Contr.
1 Sports goods 0,935
2 Tanning and dressing of leather 0,842
3 Ceramic tiles and flags 0,820
4 Repair of watches, clocks and jewellery 0,746
5 Musical instruments 0,720
6 Other agricultural and forestry machinery 0,715
7 Knitted and crocheted fabrics 0,711
8 Footwear 0,690
9 Nonwovens and articles made from nonwovens, except apparel 0,669
10 Other furniture 0,658
11 Other textiles 0,655
12 Tanks, reservoirs and containers of metal 0,654
13 Veneer sheets;_plywood, laminboard, particle board, fibre board 0,648
14 Knitted and crocheted underwear 0,643
15 Other textiles products 0,620
16 Nonwovens and articles made from nonwovens, except apparel 0,618
17 Wea~ons and ammunition 0,598
18 Cutlery 0,586
19 Man. Of other basic iron, steel and ferro-alloys 0,570
20 Other products of wood 0,566
21 Meat and meat-based products 0,562
22 Textiles manufacturing (clothes excl.) 0,561
23 Steel tubes 0,549
24 Nonwovens and articles made from nonwovens, except apparel 0,544
25 Sawmilling and planing of wood, impregnation of wood 0,535
26 Refractory ceramic products 0,531
27 Other manufacturing products 0,530
28 Other fabricated metal products 0,530
29 Lighting equipment and electric lamps 0,527
30 Household and sanitary goods and of toilet requisites 0,520
31 Other transport equipment 0,517
32 Machine - tools 0,515
33 Accumulators, primary cells and primary batteries 0,515
34 Dressing and dyeing of fur; articles of fur 0,511
35 Other special purpose machinery. 0,505
36 Cement 0,497
37 Optical instruments and photographic equipment 0,480
38 Cutting, shaping and finishing of stone 0,460
39 Electric motors, generators and transformers 0,460
40 Wooden containers 0,458
41 Games and toys 0,457
42 Other plastic psoducts 0,442
188 G. Bottazzi, G. Dosi and G. Fagiolo

Table Al (continued)
43 Pesticides and other agro-chemical products 0,432
45 Other articles of concrete, plaster and cement 0,428
46 Other non-metallic mineral products 0,423
47 Production and distribution of electricity 0,418
48 Electricity distribution and control apparatus O~~
49 Other general purpose machinery 0,417
50 Insulated wire and cable 0,406
51 Bricks, tiles and construction products. in baked clay_ 0,398
52 Other electrical equipment 0,391
53 Beverages 0,385
54 Non-electric domestic appliances 0,383
55 Manufacture and processing of other glass including technical 0,382
Iglassware
56 Bicycles 0.373
57 Pharmaceutical preparations 0,353
58 Processing and preserving of fish and tish products 0.353
59 Paints, varnishes and similar coatings, printing ink and mastics 0.352
60 Watches and clocks 0.341
61 Paper and paperboard 0,322
62 Parts and accessories for motor vehicles and their engines 0.320
63 Precious metals production 0.317
64 Other food products 0,317
65 Bodies (coachwork) for motor vehicles; trailers and semi-trailers 0,312
66 Soap and detergents, cleaning and polishing preparations 0,302
67 Builders' carpentry and joinery 0,294
68 Man-made fibres 0,292
69 Printing 0,287
70 Basic iron and steel and of ferro-alloys 0,276
71 Instr. and appliances for measur., checking, testing, navigating, 0,275
exc.industrial process.
72 Milk and cheese products '0,258
73 Luggage, handbags and the like, saddlery and harness 0,257
74 Other rubber products 0,240
75 Processing and preserving of fruit and vegetables 0,238
76 Other organic basic chemicals 0,220
77 Printing of newspapers 0,218
78 Television and radio receivers, sound or video recording or reproducing 0,202
apparatus
79 Prepared feeds for farm animals 0,191
80 Other chemical products 0,188
81 Tobacco industry 0,188
82 Motor vehicles 0,170
83 Railway and tramway locomotives and rolling stock 0,168
84 Steam generators (central heating exel.) 0,160
85 Crude oils and fats 0,155
86 Starch and related products 0,136
87 Renting of office machinery and equipment including computers 0,132
88 Basic pharmaceutical Ploduets 0,130
 On the Ubiquitous Nature of the Agglomeration Economies and Their Diverse Determinants 189

Table A 1. (continued)
89 Aircraft and spacecraft 0,128
90 Nuclear fuels treatment 0,078
91 Building and repairing of ships 0,062
92 Electronic valves and tubes and other electronic components 0,050
93 Television and radio transmitters and apparatus for line telephony and 0,047
line telegr.
94 Refined petroleum products 0,008
95 Coke and coal related products 0,001

References

Anderson, S. P., dePalme, A. and Thisse, 1. F. (1992), Discrete Choice Theory of Product
Differentiation, Cambridge, MIT Press.
Antonelli, C. (1990), "Induced Adoption and Externalities in the Regional Diffusion of
Information Technology", Regional Studies, 24, pp. 31-40.
Antonelli, C. (1994), "Technology Districts, Localized Spillovers, and Productivity
Growth: The Italian Evidence on Technological Externalities in Core Regions",
International Review of Applied Economics, 12, pp. 18-30.
Arthur, W. B. (1994), Increasing Returns and path-dependency in Economics, Ann Arbor,
Uni versity of Michigan Press.
Barca, F. (ed.) (1997), Storia del Capitalismo Italiano dal dopoguerra ad oggi, Roma.
Donzelli.
Becattini, G. (1990), "TIle Marshallian Industrial District as a Socio-economic Notion", In:
Industrial Districts and Inter-Firm Cooperation, Pyke, F., Becattini, G. and
Sengenberger, W. (cds), Geneva, International Institute for Labour Studies.
Boltazzi, G. (2001), "A Stochastic Model of Firms Location", Laboratory of Economics
and Management (LEM) Working Paper, S. Anna School of Advanced Studies,
Forthcoming.
Brugnoli, A. and Fachin, S. (2000), "Concentrazione e Diffusione: la Geografia Industriale
Italiana, 1981-1991 ", Economia e Politica Industriale, 27, pp. 103-128.
Brusco, S. (1989), Piccole Imprese e Distretti Indl/striali, Torino, Rosenberg and Sellier.
Brusco, S. and Paba, S. (1997), "Per una storia dei distretti industriali italiani dal secondo
dopoguerra agli anni novanta", in: Storia del Capitalismo Italiano dal dopoguerra ad
oggi, Barca, F. (ed.), 1997.
Cantwell, J. A. (1989), Technological Innovations and Multinational Corporations.
Oxford, Basic Books.
Cantwell, J. A. and Iammarino, S. (1998), "MNCs, Technological Innovation and Regional
Systems in the EU: Some Evidence in the Italian Case", International Journal of the
Economics of Business,S, pp. 383-408.
Christaller, W. (1933), Central Places in Southern Germany, Jena, Fischer.
Cowan, R. and Cowan, W. (1998), "On Clustering in the Location of R&D: Statics and
Dynamics", Economics of Innovation and New Technology, 6, pp. 201-229.
David, P. A., Foray, D. and Daile, J. M. (1998), "Marshallian Externalities and the
Emergence of Spatial Stability and Technological Enclaves", Economics of Innovation
and New Technology, 6, pp. 147-182.
Dixit, A. K. and Stiglitz, J. E. (1977), "Monopolistic Competition and Optimum Product
Diversity", American Economic Review, 67, pp. 297-308.
190 G. Bottazzi. G. Dosi and G. Fagiolo

Dosi. G. (1982), "Technological Paradigms and Technological Trajectories", Research

Policy, II, pp. 147-162.
Dosi. G. and Kaniovski, Y. M. (1994), "On "badly behaved" dynamics", Journal of
Evolutionary Economics, 4, pp. 93-123.
Dosi, G., Ermoliev, Y. M. and Kaniovski, Y. M. (1994), "Generalized urn Schemes and
Technological Dynamics", Journal of Mathematical Economics, 23, pp. 1-19.
Ekelund, R. B. and Hebert, R. F. (1993), "Cycles in Development of Spatial Economics",
in: Does Economic Space Matter?, Ohta, H. and Thisse, J. F. (eds), New York, St.
Martin's Pub!.
Englmann, F. C. and Waltz, U. (1995), "Industrial Centers and Regional Growth in the
Presence of Local Inputs", lournal of Regional Science, 35, pp. 3-27.
Fabiani, S. and Pellegrini, G. (1997), Education Infrastructure, Geography and Growth: an
Empirical Analysis of the Development of Italian Provinces, Roma, Banca d'Italia, Temi
di Discussione, n. 323.
Feller, W. (1968), An Introduction to Probability Theory and Its Applications, New York,
Wiley & Sons.
Fujita, M. (1988), "A Monopolistic Competition Model of Spatial Agglomeration",
Regional Science and Urban Economics, 18, pp. 87-124.
Fujita, M. (1989), Urban Economic Theory. Land Use and City Size, Cambridge, MA,
Cambridge University Press.
Fujita, M. and Thisse, J. F. (1996), "Economics of Agglomeration", Journal of the
Japanese and International Economies, 10, pp. 339-378.
Fujita, M., Krugman, P. and Mori, T. (1999), "On the Evolution of Hierarchical Urban
Systems", European Economic Review, 10, pp. 339-378.
Fujita, M., Krugman, P. and Venables, A. r. (1999), The Spatial Economy: Cities, Regions,
and International Trade, Cambridge, Massachussets, The MIT Press.
Gabszewicz, J. J. and Thisse, 1. F. (1986), "Spatial Competition and the Location of Firms",
in: Location Theory, Fundamentals of Pure and Applied Economics, Gabszewicz, 1.,
Thisse, 1. F., Fujita, M. and Schweizer, U. (eds), London, Harwood Academic
Publishers.
Garibaldi, U. and Penco, M. A. (2000), "Ehrenfest urn Model Generalized: an Exact
Approach for Market Participation Models", Statistica Applicata, 12, pp. 249-272.
Guiso, L. and Schivardi, F. (2000), Information Spillovers and Factor Adjustment, Banca
d' Italia, Temi di Discussione, n. 368.
Henderson, 1. V. (1974), "The Sizes and Types of Cities", American Economic Review, 64,
pp. 640-656.
Isard, W. (1956), Location and Space-Economy, Cambridge, Massachussets, The MIT
Press.
Kirman, A. P. (1993), "Ants, Rationality and Recruitment", The Quarterly Journal of
Economics, 108, pp. 137-156.
Klepper, S. (2000), Entry by Spinoff, Pittsburgh, Carnegie-Mellon University, Mimeo.
Krugman, P. (l991a), "Increasing Returns and Economic Geography", Journal of Political
Economy, 99, pp. 483-499.
Krugman, P. (1991b), Geography and Trade, Cambridge, Massachussets, The MIT Press.
Krugman, P. (1993), "First Nature, Second Nature, and Metropolitan Location". Journal of
Regional Science, 33, pp. 129-144.
Krugman, P. (1996), The Self-organizing Economy, Cambridge, Blackwell.
Krugman, P. and Venables, A. 1. (1995a), "Globalization and the Inequality of Nations",
Quarterly Journal of Economics, 110, pp. 857-880.
Krugman, P. and Venables, A. J. (l995b), The Seamless World: A Spatial Model of
International Specialization, CEPR, London, Discussion Paper, No. 1230.
 On the Ubiquitous Nature of the Agglomeration Economies and Their Diverse Determinant, 191

Krugman. P. and Venahles. A. 1. (1996). "Integration. Specialization and Adjustment",

European Economic Review, 40, pp. 959-968.
Lee, R. and Willis, J. (1997), Geographies of Economies. London, Arnold.
Losch, A. (1940), The Economics of Location, Jena. Fischer.
Marshall, A. (1890). Principles of Economics, London, Macmillan.
Martin, R. (1999), "The New "Geographical Tum" in Economics: Some Critical
Reflections", Cambridge Journal of Economics, 23, pp. 65-91.
Onida, F., Viesti, G. and Falzoni, A. M. (eds) (1992), I distretti industriali: crisi 0
evolu::ione ?, Milano, Egea.
Ottaviano, G. I. P. and Puga, D. (1998), "Agglomeration in the Global Economy: A Survey
of the "New Economic Geography'''', World Economy, 21, pp. 707-731.
Ottaviano, G. I. P. and Thisse, 1. F. (1998), Agglomeration and Trade Revisited, Louvain,
CEPR and CORE (Center for Operations Research and Econometrics) Discussion Paper,
1903.
Papageorgiou. Y. Y. and Smith, T. R. (1983), "Agglomeration and Local Instability of
Spatially Uniform Steady-States", Econometrica, 51, pp. 1109-1119.
Pavitt, K. (1984), "Sectoral Patterns of Technical Change: Towards a Taxonomy and a
Theory", Research Policy, 13, pp. 343-373.
Piore, M. J. and Sabel, C. (1984), The Second Industrial Devide, New York, Basic Books.
Puga. D. and Venables, A. 1. (1996), "The Spread of Industry: Spatial Agglomeration in
Economic Development", Journal of the Japanese and International Economies, 10, pp.
440-464.
Sforzi, F. (1989), "The Geography of Industrial Districts in Italy", in: Small Firms and
Industrial Districts in Italy, Goodman, E. and Bamford, J. (eds), London, Routledge.
Sforzi, F. (1991), "The Marshallian Industrial Districts in the Italian Economy", in:
Induslrial District and Inter-firm Cooperation in Italy, Pyke, F., Becattini, G. and
Sengenberger, W. (eds), Geneva, International Institute for Labour Studies.
Sforzi, F. (2000), "Local Development in the Experience of Italian Industrial Districts", in:
Geographies of Diversi ties. An Italian Perspective. Various Authors, Roma, CNR-IGU
(International Geographical Union).
Signorini, L. F. (ed.) (2000), Lo Sviluppo Locale, Roma, MeridianalDonzelli.
Tattara. B. (cd.) (2001), II Piccolo che Nasce dal Grande, Milano, Franco Angeli.
Venables, A. J. (1996), "Equilibrium Locations of Vertically Linked Industries".
International Economic Review, 37, pp. 341-359.
Venables, A. J. (1998), "Geography and Specialization: Industrial Belts on a Circular
Plain", in: Regional Integration, Baldwin, R., Cohen, D., Sapir, A. and Venables, A. J.
(eds), Cambridge, Massachussets, Cambridge University Press.
Vernon, R. (1966), "International Investment and International Trade in the Product Cycle",
Quarterly Journal of Economics, 80, 190-207.
Von ThUnen, J. H. (1826), Det Isolierte Staat in Beziehung allf Landwirtschaft lind
Nationalokonomie, Hamburg, Perthes.
III. Success Cases Around the World
The Entrepreneurial Spark: Individual Agents
and the Formation of Innovative Clusters

Maryann P. Feldman and Johanna Francis l

Abstract. This paper outlines the development of a high-tech industrial cluster in

the U.S. Capitol region through the efforts of entrepreneurs who adapted to both
constructive crises and new opportunities. We examine the initial spark of
entrepreneurship and how it influences the formation of high technology clusters.
The perspective taken is that entrepreneurs are a critical element in the formation
of clusters and their actions are important in the analysis of clusters as complex
adaptive systems. In addition, we consider how this perspective and the role of
entrepreneurship may influence economic development policy.

1. Introduction

Once established, industrial clusters benefit from virtuous, self-reinforcing

processes that support and sustain them (Anderson et al. 1988; Arthur 1994). Yet,
we have a limited understanding of how these clusters emerge, take hold and
transform a regional economy (Kargon, Leslie and Schoenberger 1992). Much of
the economic development discourse is informed by attempts to replicate the
characteristics associated with a fully functioning regional system in what may be
considered a mechanistic economic development machine - line up inputs such as
venture capital, social networks and support services and economic development
will follow. However, these factors appear to lag rather than lead cluster formation
(Feldman forthcoming). More importantly, this perspective ignores the importance
of entrepreneurs as economic change agents, able to create or attract the necessary
resources and institutions to support their ventures, as well as the rich historical
and regional context in which they operate. Although the literature has largely
ignored the individual change agent involved in the development of regional
economies (see Appold 2000 for a review), this is at odds with our understanding
of the importance of such agents (Kay 2000), the co-evolution of technology and

lprepared for the conference on Complexity and Industrial Clusters - Dynamics and
Models in Theory and Practice organized by the Fondazione Comunitu e Innovazione
under the aegis of the Accademia Nazionale dei Lincei, held in Milan, Italy on June 19 and
20, 200!. We are indebted to the individuals who agreed to interviews and generously
shared their time and expertise in identifying salient issues as well as providing and
validating information.
196 M. P. Feldman and 1. Francis

institutions (Nelson 1993) and way in which entrepreneurs actively interact with
their local environments (Saxenian 1994).
This paper examines the impact of the initial spark of entrepreneurship and how
it influences the formation of high technology clusters. The perspective taken is
that entrepreneurs are a critical element in the formation of clusters and their
actions are important to the analysis of clusters as complex adaptive systems. In
addition, we consider how this perspective and the role of entrepreneurship may
influence development policy. Historical studies document the genesis and
development of successful clusters, however this line of inquiry is not well
integrated into conceptualisations of regional change. Moreover, emulation does
not necessarily lead to success. Silicon Valley's early successes in the electronic
industry, spawned replicates across the country, many of which never took hold
(Leslie and Kargon 1994). Although critical components of the Silicon Valley
success are readily identified, it is not merely a matter of replicating them in new
locations. It is not only the components themselves that matter, but the way in
which they are combined to give rise to a self-reinforcing innovative system
(Kargon, Leslie and Schoenberger 1992). Thus, our perspective, in contrast,
focuses on the role of change agents and relies on concepts from the developing
literature on complex systems. In effect, we propose that entrepreneurs are the key
that unlocks untapped potential in the local environment and sparks economic
growth.
We also focus on individual decisions to form a company and examine how
these decisions are influenced by regional context. The specific case considered is
the development of the U.S. Capitol region 2 . By any number of measures, this
region, which was previously dominated by government employment, lacked the
attributes conventional wisdom associates with an entrepreneurial environment.
Yet, the Capitol Region has established technological leadership based on
entrepreneurial activity in biotechnology and the Internee - two new industries
that have seeded and established themselves in the past twenty years. We focus on
the evolution of this region, specifically on the development of an industrial
cluster as a series of three phases. In the initial phase, early entrepreneurial efforts
emphasized serving existing government agencies and most entrepreneurs were
from the private sector. The second phase is dominated by increased
entrepreneurial activity as an adaptation to a series of changes in the external
environment. These efforts were a response to creative crisis due to federal
downsizing as well as to new opportunities precipitated by changes in the legal
framework - the local effect of changes in the policy environment. During this
stage, entrepreneurs defined resources to promote and protect their interests. In

2The U.S. Capitol Region is considered here as the Consolidated Metropolitan Statistical
Area (CMSA) which includes Washington D.C., Northem Virginia and the Maryland
Suburbs including Baltimore City and its environs. Two counties in West Virginia were
added in 1990.
3Within the region, there is evidence of geographic differentiation. Biotechnology is
primarily concentrated in the Maryland suburbs in Gaithersburg and along the 1-270
Corridor. The Internet companies are concentrated in the Northem Virginia suburbs.
 The Entrepreneurial Spark: Individual Agents and the Formation of Innovative Clusters 197

this way, the inciependent actions of entrepreneurs are catalytic components of a

self-organizing system. The final stage is the establishment of a critical mass of
resources, some developed within the region and other resources, such as venture
capital, moving to the area. In this phase, the full virtuous cycle of self-reinforcing
activity is in play. The critical mass of start-up activity has spawned the necessary
infrastructure to sustain it, which has in turn attracted more activity to the region.
By considering the early entrepreneurial efforts through which biotechnology
and ICT took hold in the region, the approach taken is appreciative history-
friendly theorizing (Malerba et al. 1999; Teubal and Andersen 2000). The
emphasis is on the role of individual entrepreneurs as agents of change who make
decisions to start companies, shape local environments and institutions and
develop the resources and relationships that further their interests. It is argued in
this paper that viewing entrepreneurs as agents of change is critical to
understanding not only the entrepreneurship event but also the creation of a
positive local environment. For external validity, we compare and contrast our
findings with the genesis of Silicon Valley (Leslie and Kargon 1997; Saxenian
1994).
The next section of the paper offers an interpretive history of the development
of entrepreneurship in the U.S. Capitol Region. We then turn to a consideration of
the characteristics of complexity highlighted in the literature using examples from
the U.S. Capitol Region and Silicon Valley. The intention is to provide
prescriptive information for those regions trying to spark entrepreneurship and an
economic transition with an emphasis on the uniqueness of each region and the
ability of creative individuals to capitalize on that unique environment.

2. Entrepreneurship Comes to Washington:

An Interpretative History

The region around the U.S. Capitol has recently emerged as a hotbed of
entrepreneurial activity, wealth creation and economic growth. This recognition
reflects a transformation of the region from an economy dominated by public
sector employment that was neither innovative nor supportive of private sector
activity (Feldman forthcoming). In an attempt to understand this transformation,
we have studied the genesis of entrepreneurial efforts. Our analysis is an
interpretive summary based on interviews with entrepreneurs in biotechnology
and ICT. The interviews gathered information on: where entrepreneurs were
employed prior to starting their companies, what the motivation was for starting
their companies, what resources they used in developing their companies and the
technologies and the subsequent spin-off activity they generated.
There are some methodological issues to mention. First, it is important to note
that this is a retrospective study. We are limited to identifying firms in existence
now or that were at one time prominent enough to leave a record. While we are
able to trace these firms back to their founding, we have no knowledge of similar
firms that were started but may have failed, been acquired or merged into other
198 M. P. Feldman and J. Francis

firms prior to cur stud/. This approach does allow us to consider the roots of
successful entrepreneurship and the ways in which entrepreneurial activity took
hold, but it cannot address the failure of enterprises that died without leaving a
record. While each of these companies has its own unique and compelling
founding story, our objective is to discern trends and patterns.
The Capitol region has the third largest concentration of biotech companies in
the United States (PricewaterhouseCoopers 1998). Leading companies in the
region include Human Genome Sciences (HGS) and Celera Genomics
Corporation, two key actors in the international effort to map the human genome.
In addition, another local company, Medlmmune, is currently the world's eighth
largest dedicated biotech company with six FDA approved products on the
market. In total, there are approximately 300 Small and Medium Sized (SME)
biotech firms in the region as of 2001. The establishment of the industry was
chosen to be the mid-1970s, as Stanley Cohen and Herbat Boyer invented their
genetic engineering techniques in 1973. These techniques gave rise to the modern
commercial biotech industry.
The earliest entrepreneurs in the Capitol region started firms during the time
when many prominent national firms such as Amgen and Genentech formed.
These early entrepreneurs were previously employed at large supplier firms to the
National Institutes of Health (NIH). The presence of the NIH, as the U.S. agency
whose mandate is to oversee health and medical research, is a defining
characteristic of the Capitol region. It employs a large number of researchers at the
agencies' home campus in Bethesda, Maryland. The NIH has been a spawning
ground for new company start-ups, particularly in the last ten to fifteen years.
Other government institutes and agencies, such as the Walter Reed Army Institute
for Research (WRAIR) and the U.S. Food and Drug Administration have also
been a significant source of biotechnology entrepreneurs. Although the region's
universities have generated new companies, their involvement is more recent and
was not apparent at the earliest stages. While the initial entrepreneurs came from
large corporation and government institutions, new start-up firms became
particularly fruitful in generating second, third and fourth generation start-ups.
The Information and Communications Technologies (lCT) industry also has a
strong presence in the Capitol Region with a concentration in Northern Virginia.
According to some sources, the region may be regarded as the birthplace of the
Internet5 . Prominent companies in the region include MCI, AOL, NexTel, and
PSINet. Over 400 SMEs may be considered as ICT firms located in the area 6 •

4Such a study would involve access to a source of historical data on firms such as tax or
employment records to discern when the firms came into existence and when they ceased to
exist. This approach would be limited because the smallest and most typical form of start-
up, the sole proprietorship, might not be captured.
5For example, the Virginia Economic Development Partnership uses this slogan
(http://yesvirginia.org/wva-be.html).
6There is no accepted definition of the leT industry and estimates of the number of entities
in the region vary widely and appear to be influenced by media-hype. This is the authors'
conservative estimate.
 The Entrepreneurial Spark: Individual Agents and the Formation of Innovative Clusters 199

Companies in the region supply half the total worldwide Internet backbone
(PricewaterhouseCoopers 1998).
The modern computer networking technologies that are the backbone of the
Internet and ICT emerged in the early 1970s from the U.S. Department of Defense
Advanced Research Projects Agency (variously called ARPA and DARPA) (see
Kahn and Cerf 1999 for more detail). Individuals leaving the Department of
Defense (DOD) and the allied services formed the earliest start-ups. In addition,
individuals from private industry, both within the region and from outside, figure
prominently. Area universities are notably absent from this list.
While biotech and ICT are very different industries, the objective here is to
discern patterns in the origins of the companies and to explore the temporal
development of entrepreneurial activity. Several notable patterns emerge. First,
entrepreneurs hail from a variety of different prior organizations. Government
agencies served an important incubator function in both industries; however, they
were not the sole source of entrepreneurial talent. There is evidence of great
diversity in entrepreneur's backgrounds. Second, the earliest start-ups were
service firms, not originally involved in the types of R&D activities that generate
new industries. Firms such as Bethesda Research Labs and AMS were not
launched as product development firms although they evolved in that direction
over time. Thus, the industry had rather humble beginnings - not the type of start-
up that would attract much attention from investors, the media or local economic
development officials. Third, entrepreneurship picks up momentum. Over time,
generations of new firms spun-off from the earliest start-ups and entrepreneurs
who cashed in from one new venture created other new companies.
Two basic features in this interpretive history stand out: although the Capitol
region did not have the generally regarded "prerequisites" for high technology
development, a confluence of events created an opportunity for entrepreneurial
individuals to create start-ups. Second is that the organizations and entrepreneurial
ventures co-evolved. Both the biotech and ICT industries had humble beginnings
in the region and initially there was a low level of activity. However, as
entrepreneurship caught hold, the industry as we observe it today began to emerge
and the necessary components to sustain it grew up in the region.
The genesis of entrepreneurship was very different in the Capitol region than
Silicon Valley. Studies of the evolution of Silicon Valley such as Kargon, Leslie
and Schoenberger (1992), emphasize the early input of Frederick Terman, who is
viewed as the founder of the Silicon Valley phenomena. Terman had the goal of
transforming Stanford, circa 1946, into the MIT of the west. As the dean of
engineering at Stanford, he had the desire to foster scientific innovation as part of
the academic process but also to tie entrepreneurial effort and academic
accomplishment together by taking promising young students to visit high
technology start-ups run by Stanford graduates. These young students had a first
hand glimpse at research in application and were able to talk to the Stanford
graduates who had patented their academic research and moved it into an
entrepreneurial enterprise. Terman noted that many early successful young
entrepreneurs had little formal education, and suggested that someone with formal
engineering education and some business knowledge might do even better
200 M. P. Feldman and 1. Francis

(Packard 1965). Further, in 1954, Terman created the honours cooperative

program, giving local companies the opportunity to send their engineers back to
school to gain advanced degrees. He also supported the creation of the Stanford
Industrial Park (Kargon, Leslie and Schoenberger 1992).
In contrast to the Capitol region, the connection between academe and industry
in Silicon Valley was firmly entrenched early on and entrepreneurship was
encouraged and supported. Looking back on the genesis of Silicon Valley, one
administrator commended that it had been no accident, "A clearly visible impact
of Joint Services Electronics Program activity at university centres has been the
evolution of industrial activity, usually not far from those centres, entrepreneured,
manned and carried forward by students (and faculty) who have gone into those
industries" (Shostak 1987). Although the right conditions are often discussed as
the secret to the success of Silicon Valley, the early contributions of a pioneer
such as Frederick Terman cannot be ignored. He did not create the conditions for
development that we recognize in mature clusters, but rather he knew how to
extract the maximum ad vantage from the existing regional resources and
institutional context within which he operated and taught others to do the same
(Kargon, Leslie and Schoenberger 1992). Through his initial foundation and
promotion of Stanford as a major engineering school, he put into play the process
that would develop the "golden triangle" of research, industry and federal
government into a vibrant and self-reinforcing cluster. Although this golden
triangle was widely emulated, not many of those attempts were successful. Some
research universities have a greater predilection for theory and are not interested in
industry application, per se. Leslie and Kargon, (1994) document an attempt by
Bell Labs to develop close industry ties with Princeton, an attempt that largely
failed. "Princeton was the wrong sort of university partner" (Leslie and Kargon
1994). Although the New Jersey corridor did not turn into another Silicon Valley
as envisioned in the 1960s, it did evolve into a cluster of research centres and one
of the densest concentrations of electronics talent in the U.S. Similar patterns are
observed in the Capitol region. Initially, one of the major research universities,
Johns Hopkins, was not interested in industry-university collaboration and had a
climate suited to theoretical discovery but not inclined to practical application
(Desrochers and Feldman 2000).

3. Emergence of a Cluster

Lacking the strong cooperative environment and culture of university-industry

technology transfer that characterizes the history of Silicon Valley, the Capitol
region developed as a creative response to outside factors. In this section, we
discuss these factors and how they coalesce to form a complex adaptive system of
innovation.
Between 1970 and 1990, the Capitol region was affected by a series of
exogenous shocks to its employment structure. Some of these shocks were: the
downsizing of the federal government, the initiation of federal outsourcing for
 The Entrepreneurial Spark: Individual Agents and the Formation of Innovative Clusters 201

services, especially computer and telecommunication services that could be

adapted to the commercial sector, changes in intellectual property that allowed
access to high opportunity technologies and the favourable treatment of small
firms with regard to securing government contracts or financing. Each shock
provided a further impetus for firm formation (see Feldman forthcoming).
Beginning during the Carter administration in the 1970s, there was a
pronounced downsizing in federal employment that continued through the Reagan
presidency. As a result, federal employment became less secure, compensation
levels were frozen and future prospects deteriorated. Many of the affected
individuals were victims of location inertia - they had strong personal ties to the
region. However, simultaneously an opportunity for entrepreneurship emerged.
The Civil Service Reform Act of 1978 contained an initiative to outsource the
production of goods and services to the private sector. Thus, individuals in the
prime of their careers found entrepreneurship a viable employment option. The
threshold for entrepreneurial risk-taking had in effect been lowered by government
downsizing and outsourcing. When the federal "cushion" became less
comfortable, the incentives to leave were higher and there were ready markets to
tap. Federal procurement spending in the metropolitan Washington area grew by
114.3% from 1983 to 1997, creating enormous opportunities for private sector
firms while nationally, federal procurement spending increased by 3.1 % during
this same period (Haynes, Fuller and Qiangsheng 1997, p. 149). The Reagan
administration defence was materially different from other defence initiatives as it
focused on the technical and software attributes of armaments systems such as
electronics, design and systems management rather than armaments production?
While this initiative stimulated economic growth throughout the United States, the
National Capitol Region was one of the major beneficiaries (Stough, Campbell
and Haynes 1998)8.
The changes in employment structure and incentives were simultaneous with
new opportunities for the commercial exploitation of intellectual property rights
that accrued from publicly funded research. In 1980, as a response to declining
American competitiveness, a new era in the transfer of publicly funded intellectual
property to industrial firms began with the passage of the Stevenson-Wydler
Technology Innovation Act, and the Bayh-Dole University and Small Business
Patent Act. Thus, the large number of federal and university labs in the Capitol
Region were allowed to license their innovations to private firms. Most companies
appear to have started with personal funds rather than venture capital, a finding
consistent with the literature (Bhide 1999; Blanchflower, Oswald, and Stutzer
2000). The Small Business Innovation Development Act of 1982 established the
Small Business Innovation Research (SBIR) Program. Under this Act, all federal
agencies with an annual R&D budget greater than $100 million are required to set
aside a percentage of R&D funds for small business. The Act thus greatly
increased the funding available to technologically oriented small business. Lerner

?This affected ICT but also biotechnology. Consider the firm Martek that is a spin-off from
the defence contractor Martin Marietta and was funded by DOD.
8 For examples see Feldman (200 I).
202 M. P. Feldman and 1. Francis

(1996) estimates that the SBIR program has provided over $6 billion to small
high-technology firms between 1983 and 1995.
In conclusion, entrepreneurship in the region was a response to exogenous
factors: underemployed skilled labour brought about by changes in federal
employment policy coupled with new opportunities for the private sector to
contract with the federal government and commercialise new technologies. The
two industries considered here responded to different pressures. The advent of
entrepreneurship was reactive and adaptive. While both sectors benefited from
great opportunity for commercial products, biotechnology was more influenced by
CRADAs and opportunities for licensing and joint product development while
ICT benefited more from outsourcing opportunities. In both cases, locational
inertia kept the entrepreneurs in the area. Over time, the region has developed the
supporting conditions the literature associates with entrepreneurial environments.
The next section considers the ways in which these factors developed.

4. Regional Entrepreneurship as a Complex System

Although we know a great deal about mature clusters (such as Silicon Valley, CA
and Research Triangle Park, NC), we understand very little about how such
clusters develop and why they occur in certain areas and not others (Leslie and
Kargon 1994). By focusing on the entrepreneur as the spark or agent of change,
we provide a rationale for the emergence of clusters and an understanding of the
dynamic that directs their development. The entrepreneur can be viewed as an
agent who takes existing resources and re-defines, re-combines and re-deploys
them to create new products, services, and organizations (Russell 1996). Through
the process of creating new companies, entrepreneurs spark regional industrial
transformation, a transformation that exhibits path dependence, adaptivity, and
self-organization.
Interestingly, these observations about the role of the entrepreneur and
innovation in industrial transformation were already present in Schumpeter
(1942). Schumpeter placed entrepreneurship and its connection with dynamic
uncertainty in the centre of economic reasoning, by tracing all disrupting
economic change to innovations and identifying the innovator with the
entrepreneur (Blaug 1985). Entrepreneurs, in Schumpeter's view, can be
considered the "spark-plugs" or "flash-points" of creative destruction (Schumpeter
1942). When an environment provides a mix of resources and opportunities,
entrepreneurial individuals will move to capitalize on those opportunities. Their
input creates the initial changes in the regional environment that begin to build on
each other. Bygrave and Hofer (1991, p. 19) suggest, "The essence of the
entrepreneurial process is a fundamental discontinuity in the industry involved".
The entrepreneur is also active within a regional context. Leslie and Kargon
(1994) propose that technology clusters are better thought of as ecologies of
mutually dependent institutions. The entrepreneur then, operates in and activates
his eco-system to his purpose. It is in the creative feedback response of the
 The Entrepreneurial Spark: Individual Agents and the Formation of Innovative Clusters 203

entrepreneur to his environment, that the character of the cluster is determined. If

the incentives and the environment are right, these initial changes cascade into a
fully functioning cluster able to adapt to adverse shocks. Looking at clusters such
as Silicon Valley, the Texas conurbation, Research Triangle Park, and the Capitol
region, we see that entrepreneurship responded to each unique environment,
creating clusters each with their own signature characteristics (Leslie and Kargon
1994; Feldman forthcoming).
Before 1980, there was very little ICT or biotech activity in the Capitol region.
Although some of the necessary components for those industries were in place -
resources for example - there was not sufficient incentive for entrepreneurial
activity. Unlike Stanford's open appreciation and interaction with industry in the
Santa Clara valley, the major research universities in the Capitol region were more
conservative and theoretically oriented (Desrochers and Feldman 2000). Without
entrepreneurial incentives or an open culture of technology transfer, there was no
way to ignite the development of biotech or ICT clusters. If the entrepreneurial
event were a linear system - where you line up the inputs and produce a given and
certain output - then the environment and initial conditions would not matter.
Nevertheless, entrepreneurship, when considered as part of a larger complex
adaptive system of innovation, is context sensitive and history or region
dependent.
Feedback effects from the interactions of the agents and their environment
characterize the way in which the cluster evolves. These interactions, in the case
of the Capitol region, produced companies that specialized in vaccines and
genomics, as well as in Internet technology. The conditions that created the high-
technology Silicon Valley cluster generated relative specialization in other areas
such as software applications, hard drives and computer chip technology (Kargon,
Leslie and Schoenberger 1992). The fact that the development process matters for
the character of the mature industry suggests a non-linear dynamic applies and that
history determines the direction of the industry trajectory. From the initial spark of
entrepreneurial activity, to the adaptation of the entrepreneur to environmental
factors, to the emergence and self-organization of the industry around the initial
start-ups, we see a complex interdependent mix of components that develops in a
non-determinate way. Looking at activity circa 1980 or even 1985 cannot give us
much information about what the mature cluster will look like. For example, when
we consider the Capitol region in 1970, there was no indication biotech or ICT
industries were germinating. Instead, we see very little entrepreneurial activity,
virtually no venture capital, a large well-paid government sector and a relatively
hostile academic environment.
There were some positive pre-existing conditions in the region such as a long
tradition of science, fed by the NIH, Department of Defense, and other
government institutions as well as universities, a highly educated work-force and a
large number of medical institutions. But in order to activate those resources, the
set of exogenous events described above were needed to change incentives for
entrepreneurial activity and create an opportunity for individuals who were
enterprising and adaptive. The first entrepreneurs, who re-combined the regions
high-skilled labour and innovative activity in new ways to create the first
204 M. P. Feldman and 1. Francis

biotechnology and ICT companies, seized this opportunity. These early

entrepreneurs did not have an environment rich in social and venture capital, nor a
network of organizations to support them, nor regional development offices to
facilitate their ventures, so they adapted and pieced together their companies from
the building blocks that did exist. Entrepreneurship, in this sense was a break
point, creating a bifurcation or catastrophe point that set in motion factors that
would transform the regional economy.
Entrepreneurs do not know initially which direction their company will go, how
to set up its structure or even what products or services will ultimately be offered
and successful (Zaltman et al. 1973). Rather, the process of entrepreneurship is a
classic trial and error or learning-by-doing process. In this sense, the learning
process and the adaptation to new events and to the existing environment are
important determinants in the final development. The evolution of the new firms
depends itself on the start-up process, the regional (and national/international
where relevant) demand as well as the available resources in the region. The
cluster and the characteristics of the cluster therefore emerge over time from the
individual activities of the entrepreneurs and the organizations and institutions that
co-evolve to support them. For example, in the Capitol region, several large
military institutions work on vaccines. This relative expertise and the ability to
patent innovations generated by the military labs, created a niche for start-ups in
the region. Additionally, the Capitol region is home to one of the largest
concentrations of medical expertise, hospitals, clinics and medical networks,
facilitating the intensive FDA required testing of new biotech drugs.
Further, the cluster self-organizes around the entrepreneurial actIvItIes.
Entrepreneurial activity is creative and innovative and therefore, the specific needs
of the entrepreneur cannot be predicted a priori, but develop over time as start-ups
are created and the needs are recognized. We see in the Capitol region that once a
critical mass of start-ups in biotechnology and ICT was in place, other
organizations were attracted. The organization of the cluster and the
entrepreneurial ventures evolved simultaneously and even symbiotically. Having
the experience and example of the initial start-ups, the industry became self-
sustaining: entrepreneurs attracted physical and human capital to the area, public
and private networks built up to support and facilitate the ventures, relevant
infrastructure was created through public and private initiatives and services grew
up to feed these companies.
One striking fact that emerges is that the history of each cluster is unique,
suggesting that the dynamic of cluster development is path or history dependent.
In Silicon Valley, for example, we see the strength of the aeronautical and
electronics industries being championed into a high-technology conurbation by a
small group of people with a vision for the development of the region. While in
the Capitol region, we see that changes in government structure and science and
innovation policy sparked a frenzy of activity that drew resources to itself as it
grew.
Below, the emergence of high technology clusters in biotechnology and ICT is
grouped into several phases. The first phase is the emergent phase, where the
spark of entrepreneurial innovation is ignited by a confluence of exogenous
 The Entrepreneurial Spark: Individual Agents and the Formation of Innovative Clusters 205

events. The second phase can be characterized as the self-organization of the

cluster and the deepening of the self-reinforcing feedbacks among entrepreneurs,
enterprises, institutions and resources. The third phase is the maturation of the
industry into the well functioning and rich innovative and entrepreneurial system
we observe today.

5. Self-Organization

In the start-up phase, entrepreneurs began by pursuing commercial projects that

did not require high levels of investment and which were unlikely to generate the
type of large profits that would interest venture capitalists. They started with
government contracting, producing rather mundane bread and butter products,
such as medical test kits and reagents for biotechnology or services such as
computer system integrations and maintenance work in ICT. In addition, the
growing number of related firms in the region provided opportunities for
subcontracting, work and asset sharing, thus making it easier for the start-up firms
to bootstrap and steadily grow without large doses of new capital.
Venture capital seeks opportunity and when there are potentially profitable
investment opportunities in a region, venture capital is attracted. Being able to
monitor and mentor the entrepreneurial firms in which they invest makes close
geographic proximity valuable for venture capitalists (Gompers and Lerner 1999).
There are now approximately six venture capital firms headquartered in the region
and firms located elsewhere have opened branches. In March 1999, Silicon Valley
Bank opened a branch office in Northern Virginia with the comment, "We are
trying to be ahead of the curve. As far as new start-ups, this is a real hotbed"
(Montgomery and Bacon 1999).
Over time, as start-up companies grew and went public, or were bought out by
other companies, the dynamics of the region changed. Most notably, local
entrepreneurs who made large fortunes engaged in institution building to support
their activities and to encourage further entrepreneurship. Also important was the
emergence of networks of supportive social capital that began as membership
organizations to promote networking. These activities were primarily private
sector initiatives, financed with private funds. By collaborating with state and
local government programs, these initiatives resulted in cross-fertilization and a
common mission to promote the development of industry in the Capitol region.
Older style quasi-public organizations such as the Washington Board of Trade
and the Greater Baltimore Committee have broadened their agendas and spun-off
new organizations directed at technology intensive industry. Other government-
financed programs found greater success by collaborating with the privately
organized networks (Guidera 1996).
206 M. P. Feldman and 1. Francis

6. Critical Mass/System Maturation

There is evidence that the region has achieved a point where it is a self-sustaining
system. Many entrepreneurs sold their companies and started private incubators to
nurture others. These founders were motivated to share their expertise and to build
the region. In addition, at least three angel networks have formed in the last five
years. The Private Investors Network (PIN), the Capital Investors Club and the
Washington Dinner Club are organizations of experienced entrepreneurs who
actively invest in new companies and offer management advice.
Universities in the region have responded to the increased entrepreneurial
activity by offering new programs and building branch operations closer to
commercial activity. For example, Johns Hopkins University offers a Masters
degree in Biotechnology in Silver Spring, Maryland about 50 miles away from the
main Baltimore campus. Virginia Tech University opened a branch campus in
Northern Virginia about 250 miles from their main campus. The draw has been the
number of workers seeking additional training, the opportunities for industry-
funded research and the interaction with industry. Notably, the local universities
have benefited from the philanthropy of local entrepreneurs. For example, George
Mason University began in Fairfax, Virginia in 1950 as a commuter school. It has
grown into Virginia's second-largest university with 18 doctorate programs and a
focus on technology. Donors have given the university millions of dollars to
endow 43 professor chairs, allowing the university to recruit high-profile
professors (O'Harrow and Lipton 1996). All of the universities in the area have
responded with incubators and other programs to encourage entrepreneurship.
Interestingly too, dedicated university involvement in high-technology
enterprise or the promotion of academic-industry transfer occurred only much
later in the Capitol region's development. In Silicon Valley, the opposite story
could be told; the moulding of Stanford as a producer of entrepreneurs and as an
institution that was sensitive to the needs of industry figured prominently in the
success of the region and occurred in the initial phase (Kargon, Leslie and
Schoenberger 1992).
In the Capitol region, legislative programs also followed the initial phase and
addressed the needs of the newly established industry. A group of Virginia
businesspeople organized a broad campaign to advocate state tax increases for
education in order to address a noted shortage of technology workers and provide
greater infrastructure funds (Baker 1995). This initiative attempted to build
infrastructure to support the development of local industry. Rather than seeking
specific requests for their own business, business leaders were promoting a
broader, collectively responsible social agenda (Feldmann 1997). In 2001, the
state of Maryland passed twelve legislative acts focusing on providing a
supportive environment for technology-based economic development. These cover
the full gamut of infrastructure development, training programs and tax incentives.
One more point to consider besides the characteristics of a mature cluster
mentioned here, is that an industrial region reaches full maturity when it is stable
and able to adapt to adverse shocks (Saxenian 1994). Although the history of high
 The Entrepreneurial Spark: Individual Agents and the Formation of Innovative Clusters 207

technology in the U.S. is highlighted by brilliantly successful clusters, it is also

marked with cluster failures - clusters that were not able to adapt to shocks and
whose entrepreneurs and start-up activity folded or went elsewhere when the
environment became negative (Leslie and Kargon 1994). The ability to re-
configure resources in response to change is an important indicator of maturity
(Andersen and Teubal 1999).
In sum, these factors when added together represent a whole that is larger than
its parts and system which is self-sustaining and reinforcing. The mature state of
the Capitol region looks not much different from Silicon Valley, in terms of
venture capital, major research university involvement, university-technology
transfer practices, and the types of social capital and entrepreneurial efforts
observed. Nevertheless, the genesis of each region is much different - two rich
self-sustaining high-technology clusters which appear similar on the outside, but
evolved through entirely different paths.

7. Reflective Conclusions and Appreciative Theorizing

The economic success of Silicon Valley in terms of individual wealth creation,

corporate profits and job creation, has been so impressive it has pushed
government officials in locations across the nation to try to imitate or replicate its
success. Government policies aimed at replicating the conditions that exist in the
region today are based on the belief that their local areas may also capture the
benefits of new high technology firm formation and the attendant economic
growth. Much of the prevailing conventional wisdom is drawn from a snapshot of
the advanced stage of Silicon Valley's development, that is, on the workings of a
fully functioning innovative system. Looking at a successful region in its full
maturity, however, may not provide prescriptive information about how such
regions do develop. Conditions that we associate with an entrepreneurial
environment are the result of a functioning entrepreneurship and do not illuminate
the early efforts by which such entrepreneurship first took hold and the cluster
developed.
A critical question is how regions change and develop into areas with higher
growth potential. Is replication of a mature entrepreneurial environment sufficient
to foster entrepreneurship? Notable failures in such places as Dallas and Boston
route 128 (Leslie and Kargon 1994) suggest it is not. Saxenian (1994) analyses
Silicon Valley from the perspective of how it adapted to restructuring in the
semiconductor and computer industry and establishes the importance of social
relationships in defining the capacity of the region to evolve, adapt to shocks and
accommodate new demands. In this paper, we have examined how one region,
initially lacking an entrepreneurial tradition, accomplished the transformation to a
fully functioning rich regional system. Such a transformation entails a
fundamental shift or phase change from an inert innovative system to an active
system. Certainly, the Capitol Region was the site of large government research
infrastructure, classified as a State-anchored region using Markusen's (1996)
208 M. P. Feldman and 1. Francis

typology. In this regard, the concentrations of resources and highly skilled labour
plus access to sophisticated, demanding technology users were pre-existing
conditions in the region. The transformation to private sector entrepreneurial
growth did not appear to represent movement along a technological trajectory
(Kenney and von Burg 1999) but instead was a sustained effort at capacity
building that involved human agency, adaptation and evolution. Not only this, a
critical point was reached in the development of the region when it jumped from
virtually no high technology start-up activity to intense activity with start-ups per
year numbering in the hundreds.
In the development of an industrial system of innovation, there are many
individual complex stories and personal motivations. It has been said that
government employees and contractors could never become successful private
businesses - the incentives were very different. Government workers, the logic
went, were too removed from the pressures of the market. Government contracts
followed a practice of placing a low bid in order to get the job and then making a
profit by demonstrating a need for change orders in the absence of competition.
This is a very different philosophy from trying to do a job right the first time at the
lowest cost. However, the earliest entrepreneurs in biotechnology and ICT were
government contractors and employees who proved this logic wrong. What is
critical is that the region did provide opportunity for individuals. They began
working for the government but then realized that they could adapt their products
for dual-use commercial markets. Therefore, they moved to commercial markets
and with this development came innovation. Eventually they succeeded. In
essence, this was a phase change from latent to active entrepreneurship.
The entrepreneurial event in the Capitol Region was a response to and
adaptation to changes that were exogenous to the regional system. In this regard,
Federal policies such as downsizing created slack and surplus resources that could
find new and more productive uses. Thus, the gales of Schumpeter's creative
destruction were unleashed. Policies that created a supply of potential
entrepreneurs would not have been sufficient. A complementary set of
government policies aimed at creating demand for ICT and biotechnology services
through government procurement facilitated the transition. Other exogenous
conditions were the policies that provided mechanisms or tools to enable
companies to access resources. These affected the supply of new ideas by creating
access to intellectual property from government investment.
Both biotechnology and ICT are high opportunity technologies facing growing
product demand and therefore are attractive to investors. Firms working in these
technologies face favourable market conditions. The degree to which this is
exogenous may be debated. Good entrepreneurs may create their own opportunity
and thus define the industry. The idea that technology development is endogenous
to cluster development warrants more investigation. Through the actions of key
individual change agents, the configuration of the cluster and the technological
trajectory of the industry may be jointly determined. It does suggest that
companies, industries and regions benefit from the same factors and decisions -
their evolution may be intricately interwoven.
 The Entrepreneurial Spark: Individual Agents and the Formation of Innovative Clusters 209

Currently, a myriad of economic development policies attempt to encourage

entrepreneurship. We have shown that rather than being actively promoted and
encouraged by economic development policies, these activities had much more
humble and pedestrian beginnings. The conditions that we associate with
entrepreneurship developed over time. In the early stage of these new
technologies, the way in which they would develop was unclear and it would have
been difficult to anticipate the types of specific assistance that entrepreneurs
needed. Individual entrepreneurs were in the best position to move the technology,
the industry and the region forward. This is not to say that there is no role for local
government policy in promoting entrepreneurship. Although no early examples
presented themselves in the Capitol region, we have not directly examined that
question. It is interesting to note that the Silicon Valley success is viewed as an
outgrowth of the intense technology transfer and interaction between industry and
universities in that region. Local government policies did and do playa role, but
these tend to be implemented and effective in the later stages of cluster
development. Are there general lessons to be learned from the development of the
Capitol region or is this case unique and is every case unique? Certainly, this
region benefited from high average household income and higher than average
education levels, giving it very different resource endowments from other
underdeveloped regions lacking an entrepreneurial culture. The general lesson is
that entrepreneurs adapt and when they are successful, they build the types of
resources that support their activities.
A distinction should be drawn between the conditions that support innovation
and the conditions that support entrepreneurship. The two concepts are related:
entrepreneurship facilitates the realization of innovation, as firms are formed to
commercialise and advance new ideas. Conducive external environments and
resources make innovation activity easier but may not be sufficient to induce new
firm formation, which is where the concepts diverge. The critical condition for
entrepreneurial enterprise is opportunity. Even if the regional conditions do not
match those of successful clusters, the incentive to developing locational
opportunities leads entrepreneurs to start firms.
Once established, industrial clusters become virtuous, self-reinforcing cycles.
Legal and tax frameworks, research institutions and social relationships are
definitely areas for public policy intervention in terms of creating a supportive and
positive environment for innovation and entrepreneurship. Specifically relevant
are the conditions that affect the decisions of individuals to become entrepreneurs,
and the ways in which an entrepreneurial culture develops and takes hold. Yet, our
understanding of regional economic systems may be enhanced by a consideration
of entrepreneurs as economic agents who actively interact with their local
environments, adapt to new situations, crises or opportunities using location-
specific assets, and finally, build and augment local institutions. Certainly, this is
not the last word on this topic. It is my hope that this historically informed
appreciative theorizing will inspire others to take a more detailed look. It is only
through an appreciation of the nuances of cluster development that we may begin
to adequately inform policy.
210 M. P. Feldman and 1. Francis

References

Abetti, P. A. (1992), "Planning and Building the Infrastructure for Technological

Entrepreneurship", International Journal of Technology Management, 7, pp. 129-139.
Andersen, E. and Teubal, M. (1999), High Tech Cluster Creation and Cluster Re-
configuration - A Systems and Policy Perspective, Working Paper.
Anderson, P. W., Arrow, K. J. and Pines, D. (1988), The Economy as an Evolving Complex
System, Reading, MA, Addison Wesley Publishing.
Appold, S. A. (2000), "The Control of High-Skill Labor and Entrepreneurship in the Early
US Semiconductor Industry", Environment and Planning, A 32 (12), pp. 2133-60.
Arthur, B. (1994), Increasing Returns and Path Dependence in the Economy, Ann Arbor,
University of Michigan Press.
Bahrami, H, and Evans S. (1995), "Flexible Re-cycling and High-technology
Entrepreneurship", California Management Review, 37, pp. 62-89.
Bhide, A.V. (1999), The Origin and Evolution of New Businesses, Oxford, Oxford
University Press.
Baker, P. (1995), "Giving Allen the Business on Colleges: Executives Say Extra $ 200
Million a Year Is Needed for Education", The Washington Post, July 25, p. AI.
Blanchflower, D. G., Oswald, A. and Stutzer, A. (2000), "Latent Entrepreneurship across
Nations", Manuscript - August 27.
B1aug, M. (1985), Economic Theory in Retrospect, Cambridge, Fourth Edition, Cambridge
University Press.
Bygrave, W. and Hofer, C. (1991), "Theorizing about Entrepreneurship",
Entrepreneurship: Theory and Practice, 16 (2).
Desrochers, P. and Feldman, M. (2000), The Practical Application of "Truth for Its Own
Sake ": A Short History of University-Industry Technology Transfer at the Johns Hopkins
University, Working Paper.
Eisenberg, R. (1996), "Public Research and Private Development: Patents and Technology
Transfer in Government-Sponsored Research", Virginia Law Review 82, pp. 1663-1727.
Feldman, M. P. (forthcoming), "The Entrepreneurial Event Revisited: Firm Formation in a
Regional Context", Corporate and Industrial Change.
Feldmann, L. (1997), "In Virginia, Businesses Argue For Taxes", The Christian Science
Monitor, September, 9, p. 4.
Gompers, P. A. and Lerner, J. (1999), The Venture Capital Cycle, Cambridge, MA, MIT
University Press.
Guidera, M. (1996), "Start-ups Finding it Easier to Get Paired with "Angels"", University
of Maryland's Dingman Center joins Investor Network in Matchmaking Effort, The
Baltimore Sun, August 21, p. 1C.
Haynes, K., Fuller, S. S. and Qiangsheng, L. (1997), "The Northern Virginia Economy: The
Changing Role of Federal Spending", Research in Urban Economics, II, pp. 145-161.
Kahn, R. E. and Cerf, V. G. (1999), "What is the Internet (And What Makes It Work)",
Paper preparedfor the Internet Policy Institute, December.
Kargon, R., Leslie, S., and Shoenberger, E. (1992), "Far Beyond Big Science: Science
Regions and the Organization of Research and Development", in: Big Science: The
Growth of Large-Scale Research, Galison, P. and Hevly, B. (eds), Stanford, Stanford
University Press.
Kay, N. (2000), "Searching for the Firm: The Role of Decisions in the Economics of
Organization", Industrial and Corporate Change, 9, pp. 683-707.
 The Entrepreneurial Spark: Individual Agents and the Formation of Innovative Clusters 211

Kenny, M. and von Burg, U. (1999), "Technology, Entrepreneurship and Path Dependence:
Industrial Clustering in Silicon Valley and Route 12S", Industrial and Corporate
Change,S, pp. 67-103.
Lazerson, M. H. and Lorenzoni, G. (1999), "The Firms that Feed Industrial Districts: A
Return to the Italian Source", Corporate and Industrial Change, S, pp. 235-266.
Lerner, J. (1996), The Government as Venture Capitalist: The Long-Run Effects of the SBIR
Program, NBER (National Bureau of Economic Research) Working Paper No. W5753.
Leslie, S. and Kargon, R. (1997), "Recreating Silicon Valley", Business History Review.
Leslie, S. and Kargon, R. (1994), "Electronics and the Geography of Innovation in Post-
War America", History and Technology 11, pp. 217-231.
Link, A. N. (1995), A Generosity of Spirit: The Early History of the Research Triangle
Park, Research Triangle Park, The Research Triangle Foundation of North Carolina.
Malerba, F., Nelson, R., Orsenigo, L. and Winter, S. (1999), "History-Friendly Models of
Industry Evolution: The Computer Industry", Industrial and Corporate Change, 8, pp. 3-
40.
Markusen, A. (1996), "Sticky Places in Slippery Space: A Typology of Industrial
Districts", Economic Geography, 72, pp. 293-313.
Marshall, A. (1890), Principles of Economics, London, Macmillan.
Montgomery, D. and Bacon, J. A. (1999), "How Green is Our Valley"? Virginia Business,
March, (http://www.virginiabusiness.comlvbmag/yr 1999/march99/howgreen.html).
Nelson, R. and Winter, S. (1982), An Evolutionary Model of Ecunomic Change, Cambridge
MA, Harvard University Press.
Nelson, R. (1993), National Innovation Systems: a Comparative Analysis, Oxford, Oxford
U ni versity Press.
0' Harrow, R., Jr. and Lipton, E. (1996), "George Mason U. Embroiled in Debate about Its
Mission", Washington Post, 14, p. AI.
Packard, D. (1965), Stanford Engineering News, 17 (22) July.
Porter, M. (1990), The Competitive Advantage of Nations, New York, Free Press.
PricewaterhouseCoopers (1998), Toward a New Economy: Merging Heritage with Vision in
the Greater Washington Region, Washington, D.C., Potomac Knowledge Ways Project
(http://know ledgeway .org/voice/newecon/ho mepage.h tml).
Prevezer, M. (1997), "The Dynamics of Industrial Clustering in Biotechnology", Small
Business Economics, 9, pp. 255-271.
Roberts, E. B. (1991), Entrepreneurs in High Technology, New York, Oxford University
Press.
Russell. R. D. (1996), "The Emergence of Entrepreneurship in Eastern Europe: A Self-
organizing Perspective", International Journal of Commerce and Management, 6, pp. 21-
32.
Saxenian, A (1994), Regional Advantage, Cambridge, MA, Harvard University Press.
Schumpeter, J. (1942), Capitalism, Socialism, and Democracy, New York, Harper
Torchbooks, 3rd edition, 1976
Shostak, A. (1987), "History of the Joint Services Electronics Program", in: Proceedings of
the Fortieth Anniversary of the Joint Services Electronics Program, D. Robb and A.
Shostak (eds), Washington, DC, Battelle Laboratories, pp. 15-46.
Stough, R. R. (2000), The New Generation Technology Economy: Comparative Regional
Analysis and the Case of the u.s. National Capital Region, paper presented at the
American Association for the Advancement of Science meeting in Washington, D.C.,
February 2000.
Stough, R. R., Campbell H., and Haynes, K. E. (1998), "Small Business Entrepreneurship
in the High Technology Services Sector: An Assessment of Edge Cities of the U.S.
National Capital Region", Small Business Economics, V 10, pp. 61-74.
212 M. P. Feldman and J. Francis

Teubal, M. and Andersen. E. (2000). "Enterprise Restructuring and Embeddedness: A

Policy and Systems Perspective", Industrial and Corporate Change. 9, pp. 87-111.
Zaltman et al. (1973), Metatheory in Consumer Research.
From Exogenous to Endogenous Local
Development: The Cases of the Toulouse
and Sophia Antipolis Technopoles

Christian Longhi

Abstract. Regional clusters and local organisation of the industry have been
proved to be basic elements of economic growth and innovation. The
understanding of the processes underlying local development is thus pivotal. The
paper attempts to contribute to this aim through the analysis of Toulouse, in the
French southwest region Midi-Pyrenees, and Sophia Antipolis, in the southeast
French Riviera. Both be considered as the outcomes of the French national system
of innovation and the process of decentralisation initiated in the late sixties, and
are today well known centres of high technology activity. But, as the numerous
failed experiences of top-down reinforcement suggest, the factors the two areas
have been endowed cannot explain as such the emergence and development of
local innovative milieux. Considering clusters as evolving open complex systems,
the paper analyses the reconfigurations of the clusters implied by changes in
internal and external conditions that have led to endogenous innovative processes
characteristic of technopoles.

1. Introduction

Since the seventies, regional clusters are object of considerable interest. The
recognition of their distinctive role on the economic and technological scene
became evident with the rise of the industrial districts in the Third Italy. The
districts succeeded to perform better than large firms within the same international
competitive markets because of the great flexibility allowed by the institutional
arrangements set up locally (Becattini 1990). They have represented a kind of
localized achievement of the division of labour where the innovative capabilities
and competitive efficiency have arisen from organisational designs established by
the firms, arranged in networks of specialized sub-contractors in vertically
disintegrated processes of production (Brusco 1982) Another important
development has been the emergence of high tech clusters based on the knowledge
base developed in the universities, as happened in the Silicon Valley or
Cambridge, UK (Saxenian 1994; Garnsey 1998).
Despite very different features regarding industries or resource bases, these
clusters have in common local endogenous processes of development triggered by
innovative SMEs. These models have promoted regional clusters and local
organisation of the industry as basic elements of the economic growth. As clearly
214 C. Longhi

stated by Porter (1990), growth can be measured at the level of nations, but
nations in some sense neither produce nor exchange. Their competitive advantage
arises from the strength of regions, cities, districts, which are the locus of
production. Likewise, innovation, the capacity of economies to change and adapt
to their environments, are deeply rooted in clusters, and strongly related to the
organisation of the industry implemented locally.
The understanding of the processes underlying local development is thus
pivotal. The paper attempts to contribute to this aim through the analysis of cases
very different from the paradigmatic Italian districts or university-based clusters.
They are instead the result of industrial and regional development policies, of top
down processes implemented in the French economy, that have resulted in the
emergence of high tech clusters or technopoles, defined here as regions developed
around several interrelated knowledge elements, including but not limited to
science parks or research or technology centres (Luger 2001). Indeed, Toulouse, in
the southwest region Midi-Pyrenees, and Sophia Antipolis, in the southeast French
Riviera, between Nice and Cannes, can both be considered as the outcomes of the
French national system of innovation and the process of decentralisation initiated
in France in the late sixties, but their evolution was in a sense unpredictable.
Obviously, the regional clusters are not naturally given, they are constructed and
instituted outcomes of economic processes (Metcalfe 2000). As the numerous
failed experiences of top-down reinforcement of local attributes suggest, the
factors the two areas have been endowed cannot explain as such the emergence
and development of local innovative milieux. The focus has not to be on the
factors as such, but on the interactions among them, the feedback effects and
cumulative processes they give rise, and the process of development they entail
(Garnsey and Longhi 1997). The paper traces these processes in the two cases. A
suitable analytical framework has to be adopted to grasp these dynamics and build
the history of the innovative milieux.
A technopole, as well as territory in general, can be identified as a spatial cluster
of activities, a geographically bounded concentration of interdependent firms and
institutions open to its environment, in short an evolving, open complex system.
The "system" is a way of conceptualising the set of related elements with a
designated boundary which represents economically the territory. Regarding
innovation, Andersen and Teubal (1999) identify the business sector (the
backbone of the process), the supporting structures (non-business organisations
and formal institutions like research centres, universities, business association,
formal policy institutions), links and institutions (i.e. user-supplier networks,
research-industry relations, labour markets), culture and social structure as basic
components. The "system" provides a basis for finding connections between
various aspects of the milieu and for tracing its dynamic. At last, territories are
enacted systems, individual agents, key actors, or institutions are the basic
determinants of their trajectories; as such, despite underlying common processes,
their trajectories are unique, unpredictable, and non-replicable (Garnsey 1998).
The relevance of the complexity theory to grasp territorial dynamics has become
evident from the recent developments of the literature. The nature of the local
development appears as an emergent property of the system, i.e. coherent and
 From Exogenous to Endogenous Local Development 215

unexpected macro behaviour of the whole which is not attribute of the parts, but
results from their micro interactions, and which can go with important
discontinuities. This analytical framework will allow make clear the development
of the local systems, their evolutions following the initial public impetus,
adaptations from internal interactions and feedbacks, and from reactions to their
environments. The internal as well as external interactions of the local systems are
essentials to the definition of the dynamical paths; changes, "small events" (Arthur
1989) regarding these interactions can result in crisis or re-configuration of the
systems.
We turn now to the cases of local industrial development. Conceptualising their
development in terms of the genesis and growth of complex adaptive systems
provides an heuristic device to "organize" their histories, the different
reconfigurations that have led to the emergence of technopoles, as well as to
understand their fundamental specific features. Diversity is indeed the basic
characteristic of local development; both cases involved mechanisms of
reinforcement triggered by public policies of regional development, by stimulus to
local activity from leading institutions, by strong public private
interdependences, but the combination of these mechanisms of reinforcement
differed, and have resulted in self-reinforcing expansion although through highly
distinct paths.

2. The Cases

2.1. The Toulouse Technopole

Since thirty years, Toulouse has grown as a major European centre of aeronautics
and spatial activities. One distinctive feature of the area is the role of the
interdependences between industry, research and training, and services in the
development of the local resource and knowledge base, which define it as a
technopole. Nevertheless, the contemporaneous system of innovation has not
found its source locally, but results from a state-led exogenous process of
accumulation of related activitie~. Its development has been closely related to the
construction of the European aerospace industry, in which Toulouse has played a
pivotal role. The name of Toulouse is today closely associated with Airbus or
Ariane, and symbolized in a thematic park, the "Cite de l'Espace", welcoming
numerous visitors. Nevertheless, the development has not followed an unique
path. It is obviously gone with ups and downs regarding the cycles of its dominant
industries. But above all, the institutional and organizational arrangements
qualifying its regime of growth or innovation have deeply evolved. The paper
characterizes the different steps of this process, the internal and external
circumstances that have induced changes in the local system and led to a
technopolitan milieu.
216 C. Longhi

2.1.1. Building of a Productive Capacity

The history of aeronautics in Toulouse begins with World War I, and the location
of factories in areas sheltered from the front. Since then, the region has benefited
from a strong image effect built on aeronautics, and the further public choices of
investment in the area have never been questioned. Key agents explain this
development. Latecoere, following a request of the government, created his
factory in 1917 to produce fighters, and developed the "Aeropostale" after the
war, opening new lines with his planes. The human adventure embodied in the
postal airline, with men like Mermoz or Saint-Excupery, gave a strong identity to
Toulouse. Dewoitine created his firm as a spin-off of Latecoere in 1920; he
developed fighters and employed up to 7,000 people before World War II. The
local tradition of artisan industries helped the firms to find the human resources to
handle the heterogeneous elements making up a plane. Subcontracting appeared
with the industry. As reported by Allinne (1996), Latecoere or Dewoitine has no
difficulties to find small subcontracting firms which adapted their activities and
endorsed the responsibility of design and development of whole parts of the
planes, prefiguring the contemporaneous industrial organization. The
specialisation of the subcontractors resulted in a diversified industrial context of
interdependent SMEs, whose development has been fostered by the Defence
budget.
After the war the state reorganized the industry, merging the numerous
manufacturing firms weakened by the war, which failed to adapt to the new
competitive context. The nationalization process resulted in the creation of "Sud
Aviation", including the Toulouse factories, and "Nord Aviation", the first one
specialized in commercial aircrafts and the second one in military. In 1970, these
two firms together with SEREB (Societe d'Etudes et de Recherches des Engins
Balistiques) were merged to form the SNIAS or Aerospatiale, in order to reinforce
the capacities of R&D and design, and reach a size compatible with the
international standards. The new entity covers the different fields of aerospace
activities regarding civil or defence applications. The nationalisations led to the
programs Caravelle and Concorde, whose assembling was entrusted to the Sud
Aviation factories in Toulouse. This state-led process has resulted according the
terms of Muller (1988) in an "arsenal logic", i.e. a highly centralized industrial
process driven by the political sphere, the decisions on the programs being made
by the ministries of Defence, Air, Finance, and the head of the Air Force.
According this logic, the state controlled the definition of the whole characteristics
of the programs, the financing of R&D and production and was the main buyer of
the products; the budgets allocated were generally exceeded without
consequences. Two main basic goals are entrusted to the national firms in charge
of the realisation of the programs (Muller 1988; Talbot 2000). First, to reach the
technological frontier, without consideration of commercial aspects or airlines
needs. Caravelle has been an innovative medium-haul aircraft sold at 278 units, 80
to Air France, but Concorde is remained a prototype. Second, to be the vectors of
the regional development policy planned by the government. Aerospatiale played
this role in Toulouse, animating a network of subcontractors. The region gained
 From Exogenous to Endogenous Local Development 217

increased productive capacities, but along fordist characteristics; the

subcontractors were no more involved in the conception of the elements to be
manufactured, and supported the peaks of the market. Only up to 20 % of the
subcontracting was localized in the Toulouse area, the bulk of the market, as well
as the R&D and design were still concentrated in Paris. The local system of
production of this period can be depicted as an hub and spoke (Markusen 1996),
the hub itself being driven form the State. This trend was to reverse with the
definition of the European program, and the transition towards a "logic" oriented
by a market approach.

2.1.2. Take-off of the Technopo/e

In the late sixties, different public policy decisions have oriented the whole socio-
economic development of Toulouse towards a technopole, through the
decentralisation from Paris of the main research and education institutions related
to aerospace activities, and the induced impetus given to the existing local
resources. The emergence of a "collective agent", the CNES (Centre National des
Etudes Spatiales) in the local system had a pivotal role to generate positive
feedbacks from this accumulation of new resources.
With the development of spatial activities, the French government created the
CNES in 1961, a public establishment in charge to define and carryon the spatial
programs in collaboration with research, industry and defence. From 1968, the
CNES has decentralized in its Space Centre of Toulouse most of its R&D
activities; the centre has grown today up to 2,500 researchers. The CNES has
developed and coordinated the Ariane program for ESA (European Space
Agency), as well as most of the public and defence applications of satellite
activities in France and in the European programs. The expertise of the CNES
merges different scientific and industrial fields, and it develops thematic R&D on
radio communication, observation, orbital sciences and infrastructures, space
transport, or generic research on design and development of systems. The Space
Centre has been the main catalyst of the development of space activities in
Toulouse, and has induced the location in the area of the main actors in the field,
public or private. In line with its decentralization, the CNES has prioritised its
local relationships and developed synergies between the different actors to feed
the needs in pure and applied research as well as industrial development needed
by its programs, whose main feature is the innovative content. It has implemented
and co-ordinated networks inducing processes of collective learning and the
emergence of a local innovative milieu in spatial activities, which cross-different
scientific and technological fields. The CNES has also created independent firms
based on home technologies to foster the innovative process in the local complex.
The unique feature of Toulouse is that along with the CNES, most of the higher
education and research institutions linked to aerospace activities have been
decentralized from Paris, making a coherent and interdependent knowledge base.
Indeed, the decision to locate in Toulouse the main university level schools
("Grandes Ecoles") of engineers has been conditioned by the simultaneous
location of the related institutes of research, because there are closely interwoven
218 C. Longhi

institutions. They are both required for the quality of the diplomas delivered,
engineers, but also PhD, and for the setting of close links with industry, which is
part of the mission of these institutions. This process is gone with the
reinforcement of the existing institutes of University and the CNRS, which had
already a strong potential in computer sciences, electronics, robotics, materials.
Among these main institutions, one can mention ENSEEIHT (Ecole Nationale
Superieure d'Electrotechnique, d'Electronique, d'Informatique et d'Hydraulique
de Toulouse), SUPAERO (Ecole Nationale Superieure de \' Aeronautique et de
l'Espace), ENSICA (Ecole Normale Superieure d'Ingenieurs de Constructions
Aeronautiques), ENAC (Ecole Nationale de l'Aviation Civile), ONERA-"Centre
de Toulouse" (Office National d'Etudes et de Recherches Aerospatiales), LAAS
(Laboratoire d' Architecture des Systemes, CNRS), LEEI (Laboratoire
d'Electrotechnique et d'Electronique Industrielle), CERT (Centre des Etudes sur
les Rayonnements), IRIT (Institut de Recherche en Informatique de Toulouse). In
addition, the "Ecole Nationale de Meteorologie", as well as French National
Meteorology offices have been located in Toulouse in a specific area, the
"Meteopole". They are backed by the spatial activities or the CNES, using data
collected by the satellites Meteosat or Spot, and have developed an important
computing expertise needed for forecasts I.
This education and research complex strongly oriented towards industries or
applications has met the tradition of the University Paul Sabatier in Toulouse, as
well as the socia-economic and cultural tradition of the city. Indeed, in 1885
already the local government was the main financial support of the new Faculty of
Science, and subsidized a course on electricity to meet the industrialization needs
of that time; the course has been dedicated from 1902 to the training for local
entrepreneurs. Or it provided the premises and the wages of the professors of the
"Institut d'Electrotechnique" created in 1907 (Allinne 1996). This tradition shared
by the main institutional local bodies has allowed the easily integration of such an
amount of new external resources in the area, which have worked as the
reinforcement of an existing socio-economic process of development.
The result of the whole process has been the early emergence of a local labour
market of high qualification, allowing the "diffusion of embodied and tacit
knowledge and technological know-how" (Camagni 1991) and local collective
learning. Indeed, the students are vectors of creation of close relationships
between education, associate research institutions and firms, through training
periods or recruitments, or their involvement through PhD in the problem-solving
activities of the firms. Toulouse appears as an attractor for future engineers,
technicians studying in its schools or universities, which search for local
employment whatever their geographical origin (Grossetti 1990). In 2000,
Toulouse gathered more than 107,800 students, 70 % in the four universities and
their engineer schools, 5 % in the "Grandes Ecoles"; complemented by vocational

1It must be stressed that Toulouse is also one of the main national pole in health,

pharmaceuticals, biotechnologies, regrouping an important university and public or private

institutes of research. This cluster will not be analysed in this paper, as it does not meet the
process of local development induced by strong public policies it focuses on.
 From Exogenous to Endogenous Local Development 219

trainings for technicians (DRIRE (Direction Regionales de l'Industrie, de la

Recherche et de l'Environnement) 2001).
This concentration of education and research is the key of the development of
Toulouse. The process of decentralization has often been in France the result of
the eviction in regions of standard activities to the benefit of the high value added
ones developed in Paris. In Toulouse, it appears that the French National System
of Innovation has in some sense been decentralized in some activities. The system
works indeed on the basis of the close association of the "Ecoles" of engineers and
related research institutes, the large public and private firms, the public agencies
or establishments like CNES or ONERA, the whole being articulated by national
programs (Grands Programmes). These elements, usually concentrated in Paris,
are characteristic of the contemporaneous aeronautics and space activities in
Toulouse.

2.1.3. Re-configuration of the Local System

Space Industry. The location of the CNES and the research institutes in Toulouse
have been followed by the decentralisation of Matra Espace in 1979 and Alcatel
Espace in 1982, which have been two main actors of the development of space
industries in France, the third one, Aerospatiale Espace, being located in Cannes
near Sophia Antipolis. This process complements the regional development
strategy of the government.
The European Space Agency (ESA) has presided over the development of the
space industry in Europe. Given the important share of the French government in
the ESA budget, the CNES has played a pivotal role in the definition and design of
the projects, national but also European, as well as in the choice of firms in charge
of the development and design. The technological standards required by the CNES
were very high and the satellite programs calibrated for short series, more oriented
towards R&D than competitive markets. The growth of the activity has followed
the evolutions of the public programs, with some downturns like the Hermes
project for example. Given the principles governing ESA, the space industry has
been from the beginning organized in cooperative networks of firms (alternatively
Aerospatiale, Alcatel, Alenia, DASA, Matra ... ) built to complete each program.
Toulouse has been a locus of the construction of the European space industry;
obviously, the state control on the space policy and thus the definition of the
programs is not questioned. But the knowledge base has grown from the local
linkages established between the different actors of the field. Alcatel and Matra
have both grown today up to 2,000 employees, drawing on the local labour market
to feed their growth, but also to develop new capabilities. The industrial strategy
of Matra Espace has been very influential on the constitution of the milieu, and
complemented the action the CNES (Dupuy and Gilly 1999). The firm was
organized without central laboratory of research and multiplied the local links
with the CNES, the institutes of research and the potential subcontractors to solve
the complex technological problem-solving activities raised by the space industry.
Matra acted as prime contractor of the projects and specialized in integration of
systems, implying the coherent merging of the different knowledge bases involved
220 C. Longhi

in the definition of highly complex products like satellites. As underlined by

Dupuis and Gilly (1999), these problem-solving activities call for multiple skills
and confrontations of non-formalized knowledge, which continuously reshape the
solution. Proximity appears pivotal: 80 % of the Matra subcontracting has been
allocated to local research institutes and firms, essentially in computing services or
electronics, but also in firms like Zodiac, for multi-layer insulation equipment for
satellites for instance. Matra is also involved in one of the different joint-institutes
between firms and CNRS created in Toulouse, which are devoted to fundamental
research problems. The relations of Alcatel Space with its local environment have
been less intense, its needs in R&D were mostly covered internally from the
national integrated group it was part of. The firm specialized on payloads and on
board equipments, often in joint programs leaded by Aerospatiale (Cannes), which
displayed also important internal potential of research. The situation evolved
recently when these entities have been merged, under the control of Alcatel Space
in Toulouse. The positive feedback induced by the interactions between research
and industry in problem solving activities have rapidly led to the emergence of an
innovative milieu in space activities. The coordination of the learning process
implemented by the CNES, as well as the pivotal role played by research institutes
like LAAS or IRlT involved in research programs with different industrial
partners, have been essential in the emergence of localized knowledge.

Aeronautic industry. In the aeronautic activities, the development of the Airbus

program, and the induced reorganization of Aerospatiale, had an important role in
the reinforcement of the industrial context in Toulouse, and its evolution towards
collective learning processes and conception oriented activities. Given the cost of
development of programs in aeronautic activities, the cooperation of different
companies was necessary in Europe; Concorde was already the result of such
cooperation, but, among many other reasons, the institutional arrangements that
prevailed in this process worked as constraints upon the commercial success of the
project. After many resistances from the defenders of the "arsenal logic" (Muller
1988), the GlE Airbus (Groupement d'lnteret Economique) was created to design
new arrangements, which decreased the prevalence of states in the programs and
the inefficiencies it entailed. For example, two assembly lines were working for
Concorde, one in France and one in UK, to allow strict equality in the cooperation
and equal control over the process. Clearly, aeronautics were considered as
strategic activities by the states, and closely controlled by the Defence authorities.
The institutional arrangements under the GlE Airbus have been promoted by two
key actors, R. Beteille, from Sud Aviation, and F. Kracht, a former engineer of
Nord Aviation, from DASA, who had experienced the discrepancies of the past
programs. They intended to promote the definition of programs directed by the
needs of the clients and the markets. The resulting arrangement has been a GlE
owned by the companies associated in the consortium, Aerospatiale, Daimler-
Chrysler Aerospace (DASA) and Construcciones Aeronauticas (CASA), but prime
contractor for these companies. In 1978, British Aerospace (BAe) joined the
consortium. The GlE has been entrusted of the control on design and production,
of the relations with the clients, and of test flights. The principle of equality
 From Exogenous to Endogenous Local Development 221

working under the "arsenal logic" is given up, the division of labour obeys the
specialisation of the companies for the strategic components of the planes
(Aerospatiale is endowed with the integration of systems, the cockpit and the
controls), the industrial parts being divided according the financial involvement of
the partners (Muller 1988). The same principles are applied in the other European
consortium, the GIE ATR (Avions de Transport Regional) linking Aerospatiale
and Alenia. The implementation of GIE has allowed to implement an efficient
cooperation without merging, any process of privatisation of the aeronautic
activities being impossible to consider in France at this time.
The head office, management and design activities of the GIE have been located
in Toulouse, giving the area the role of European centre regarding aeronautics.
The GIE was an entity without capital and without profit, but it gained a strategic
position. There is indeed an asymmetry between Airbus Industry and the partners
of the consortium as, through work sharing, the GIE increases the specialization of
the partner and holds the control on the whole process and the commercial
expertise (Talbot 2000). Airbus has continuously grown, to reach 2,400 persons in
2000.
The development of the Airbus family has induced the restructuring of the
activities of Aerospatiale, from an organisation based on largely autonomous sites
to a multidivisional organisation (Kechidi 1996). The restructuring has mainly
concerned the "Aircraft Division", where all the establishments have been
integrated through the creation of a "Direction of Production". The design and
conception department have also been merged, and located in Toulouse along with
the headquarters. The specialisation process induced by the GIE Airbus in the
consortium has been internalised by the tirm. It obeys the decomposition of the
aircraft on sub-systems, allocated to the partners depending of their specialization,
and the process of standardisation induced by the development of the Airbus
family, optimised through flexible factories; accordingly, the division of labour
between the site obeys their technological specialisation. This process is gone with
the complete redefinition of the subcontracting strategy, from its decentralization
by sites to its co-ordination from the Toulouse direction. The new internal
principles of specialisation, standardisation, automation, have been applied to the
subcontractors, and the internal tools of computer-aided technology extended to
the network. The share and nature of the subcontracting has also evolved.
Aerospatiale internalises only the strategic activities, design, control, cockpit,
controls, electronic systems, assembling, engine pylons, assembling, but 10 to 40
% of these activities are planned to be subcontracted. The share of the sub-systems
based on diffused technologies is planned from 80 to 90 %, and of 100 % for
standard technologies (Kechidi 1996). A further consequence of the specialisation
is that subcontractors are entrusted with the whole responsibility of sub-systems,
from design to production, meaning the development of internal capacities of
conception, involvement in the financing of programs, in maintenance. A multi-
layer organisation of sub-contracting is adopted, the first layer firms organising
the subsequent levels, with the agreement of the prime contractor; the direct links
of the Aerospatiale have fallen from 650 in 1987 to less than 100. The
externalisation of sub-systems from design to production has induced new
222 C. Longhi

coordination processes between Aerospatiale and the first layer of subcontractors.

For example, the Program Meeting Review, i.e. regular bimonthly meetings where
the evolution of the plans are discussed and the coherence of the whole process
ascertained (Talbot 2000).
The consequence of this reorganisation has been the concentration of the first-
layer subcontractors in Toulouse, because of the coordination process going with
design activities, and new arrivals of firms, to the detriment of the others regions
or Midi-Pyrenees, where volume subcontractors are dispersed. Giving the pressure
on costs implied by the market-oriented strategy of Airbus, and thus Aerospatiale,
these last are indeed more and more under competitive pressure from foreign
firms. For example, Lacetoere, first layer firm, subcontracts activities in Poland,
Korea, and has acquired firms in Czech republic. This process is reinforced by the
principle of compensation working with the aircraft market. As a consequence,
subcontracting covers today the whole countries clients from Airbus. On the
contrary, the competitive and technological level of firms located in Toulouse has
increased, and they have been able to diversify their markets. Latecoere,
Microturbo, Sextant Avionique, Liebherr Aerospace work also for Boeing or
Bombardier. Toulouse is thus the node of a global networks characterizing the
contemporaneous innovative systems, but with decreasing links with its close
periphery. An other consequence of the reorganisation of aeronautic activities is
the growth of independent firms in design, conception, consultancies, computing
activities in the area of Toulouse, as the market of conception is no more
monopolized by an integrated firm. The local aeronautic activities are thus shifting
from manufacturing to R&D, design, management, integration of systems, and
from internalisation to externalisation, inducing the multiplication of linkages
between firms and institutions, and thus the emergence of a technopolitan milieu

On board systems. The existing knowledge base in the university and institutes in
electronics, robotics, automation, computing, has induced others related industrial
developments, characterized also by the pivotal role initially given to large firms
and obeying also to the policy of regional development directed by the
government. Motorola for example, which employs 2,500 persons in Toulouse,
has specialized the site on on-board systems for cars. Or Siemens Automotive,
1,500 employees, specialized in car safety and comfort technologies, like
navigational aids. These locations have resulted in further arrivals, like Bosh,
Valeo, Actia on diagnostic apparatus for the car industry. On the whole, an on-
board systems cluster has emerged in the Toulouse area, which covers the whole
range of applications, from aerospace to car industries, involving design, software
development, testing. It results from close interactions between public and private
research ranging from electronics to materials, between small and large firms, and
it is integrated in international markets. The public or private institutions of
research (like LAAS, LEEI, IRIT) are involved in many networks related to
different technological domains, and help to the building of specific local
competences on the technopolitan area.
 From Exogenous to Endogenous Local Development 223

2.1.4. From French to European Technopo/e

The recent period has been highly challenging for Toulouse, as since 2000, the
landscape of the aerospace industry has tremendously changed. The first challenge
has been the A380, the new long-haul aircraft launched by Airbus Industry. An
aspect is the innovative content of the plane. Another aspect has been the location
of the assembly line of the A380, which was a strategic element regarding the
future centre of the European industry.
From a long time, Toulouse had planned to dedicate industrial wasteland from
Dassault-Breguet to the project. But the area resulted undersized for the definitive
program, and Hamburg proposed an alternative location open to the sea and
adapted to the logistic implied by the A380 assembling. The local public response
in the Toulouse area resulted in the creation of a new zone dedicated to aeronautic
activities, AeroConstellation, localized in Blagnac. The project has been build by
the major of the city, B. Keller, who associated different communes to gather 380
hectares dedicated to the zone. The investment amounts to 4 billions francs: 700
millions from the cities, 300 millions from the "Departement" for the development
of the zone, 2.5 billions from Airbus to build the assembly line, and 200 millions
from Air France, which will locate its maintenance facilities on site. The airport,
as well as the existing Airbus assembly lines, the head offices of Airbus, ATR,
British Aerospace were already localized in Blagnac. Nevertheless, the city could
not stand this new investment alone. The institutional arrangement has been the
creation of a grouping of communes around Toulouse (Communaute
d' Agglomeration), gathering the resources dedicated to local development.
AeroConsteliation illustrates to socio-economic context of Toulouse. Since the
works of Becattini (1990), the understanding of economic successful regions has
placed considerable emphasis on the key role of local institutions or on the
importance of regional institutional thickness, the interlocking web of institutions
including firms, chamber of commerce, local governments (Keeble et al. 1998).
Indeed, Keller was consultant in Airbus France, coming from Aerospatiale, 1.
Diebold, the responsible of the AeroConstellation project in the Toulouse
municipality, was engineer in Aerospatiale Airbus, the head of the Chamber of
Commerce is also the head of Latecoere. This strong socio-politico-economic
context helps to understand the development of the region, its capacity to face
changes and to adapt to new conditions; as we have seen, it takes roots in the
history of the region, as the development of the university obeyed the same
interlocking hundred years ago. But, contrary to the usual district socio-economic
context, the regional system of innovation in Toulouse is based on the importance
of large firms and the interdependence of public - private institutions.
The "Communaute d' Agglomeration" has also promoted the development of an
optical regional network to meet the development of the Information and
Communications Technologies. This investment complements the activity of
France Telecom, which employs 4,000 persons in Toulouse for the development
and operating of interurban and pan-European networks. These activities have
created more than 12,000 jobs in ten years, and stand as an engine of growth in the
region (INSEE 2000). They are closely linked with the space industry, and the
224 C Longhi

satellite telecommunications technologies, leaded by Alcatel. They find also an

important regional market, the large firms moving to internet technologies for
their control and management systems. The emergence of the internet
technologies has reinforced the movement of creation of specialized services
SMEs, which emerged with the externalisation of activities in large firms, and
sustains the deepening of the technopolitan feature of the local innovation system.
The contemporaneous development of the area, traditionally leaded by large firms
standing as the "natural" market for the engineers and PhD students, is
encouraging entrepreneurship; the recent creation of an incubator by the
universities, the institutes of research and the "Grandes Ecoles" to promote the
creation of new firms evidences the changes of the local socio-economic context.
The preparation and launch of the A380 program coincides with a major
restructuring of the European scene in aerospace industries, from European
cooperation networks to European firms. In France, the successive major
restructuring of the economy has traditionally gone with nationalisations and re-
allocations of assets between firms. This restructuring of the aerospace activities is
gone with privatisation, moreover of firms traditionally linked to defence ... This
process of emergence of European firms has first emerged in the space activities,
and covers today the whole range of activities. The France reluctance to abandon
the state ownerships in aerospace industry has in fact been overcome by the
German defence of pan-European mergers, the new institutional rules resulting
from the US-Europe agreements, and need to adapt firms to competitive markets.
In space industries, the process began with the merger of Matra Espace with
Marconi Space System, and the acquisition of BAe Space Systems by Matra
Marconi Space in 1994. But the whole process has been allowed by the surprise
Aerospatiale and Matra Hautes Technologies merger set to combine the state-
owned company and the Largardere group into a shareholder driven firm. The
pan-European strategy promoted by Germany was strongly shared by Lagardere,
and the merger has been rapidly followed by the creation of EADS (European
Aeronautic Defence and Space), merging Aerospatiale Matra, DASA and CASA.
In space activities, two major groups have been constituted. Astrium, from Matra
Marconi Space and DASA Space, and Alcatel Space, from Alcated and
Aerospatiale Espace. EADS owns Eurocopter, 80 % of Airbus, 75 % of Astrium,
45 % of Dassault Aviation, and a 29.5 % stake in Arianespace. This reorganisation
of the industry has been followed by the transformation of the GIE structures into
new organisations, for Airbus as well as for ATR. EADS and BAE Systems will
hold respectively 80 % and 20 % of the newly created corporate Airbus company,
and an equal partnership EADS - Finmeccanica Alenia Aerospazio has been built
for ATR Integrated. The GIE structures have allowed to pool the European
capabilities in order to build an important resource base and gain an increasing
share of the world market, under the constraint maintained national independence.
But the growth of the activity has resulted in the need of more management
control over the whole industrial process, from design to subcontracting strategy.
This process is in some sense the outcome of the growing asymmetry between the
GIE and the partners underlined in Talbot (2000).
 From Exogenous to Endogenous Local Development 225

This reshaping of the European landscape gives Toulouse the central role in the
sector. In addition to the assembling of the A380 and the new ATR, it will receive
the Airbus and ATR headquarters, activities of design, programme planning,
delivery and customer services, and training, confirming the embeddedness of the
firms in the local area. The creation of Airbus as an independent corporate
organisation will imply a restructuring and rationalisation of all related activities
from design, engineering to manufacturing, and in Toulouse, a deepening of the
process initiated with the former reorganisation of Aerospatiale. One can expect a
reinforcement of the context of first-layer subcontractors, and of design and
engineering activities, as well as an increased pressure over the volume
subcontractors. deepening the technopolitan nature of the area. Airbus has already
announced a planned decrease of 20 % on the cost to allow the commercial
viability of the A380 program. The innovative content of the new program has
induced, as soon as approved, a strong pressure on the local market on engineers
and high qualifications. Large as well as small-specialized firms are adapting their
capacities in R&D, conception, to fill the competitive requirement implied by the
new institutional arrangements governing Airbus. The trend is an increasing
concentration of the innovation process in Toulouse, at the expense of Paris.
The creation of Astrium and Alcatel Space, strongly territorialized in the local
milieu, and involved as prime contractors in many new programs, from earth
observation to telecommunication and internet, is to reinforce the position of
Toulouse in these activities. The restructuring of the firms has allowed their
adaptation, from commands of short series, high tech prototypes from the CNES
or ESA, to competition and mass-production going with for example the growing
market of telecommunication satellite. The role of the CNES will certainly have to
evolve in this new setting, and new institutional arrangements to emerge. Indeed,
as a consequence of the restructuring of the aerospace industry, Toulouse leaves in
some sense the French national system of innovation, in which it played a pivotal
role, to move more deeply towards the European one and to competitive global
markets.

2.2. The Sophia Antipolis Technopole

The Sophia Antipolis Park is located in the South of France, on the French Riviera
between Nice and Cannes. It is often presented as a project which initiated a new
form of local development, the "first technopolis in Europe", built on the idea of
cross-fertilization between research and industry championed by its creator, Pierre
Laffitte. Indeed, the experiment, built on greenfield, is today an important centre
of high technology activities in Europe; 1,193 firms or institutions were located in
the park in 2000, offering 21,536 jobs, and some 5,000 researchers and students
are present on the park. The experiment is only 30 years old; this short period
allows drawing the main characteristics of the history of the project, to trace and
understand the different phases of growth and crisis.
226 C. Longhi

2.2.1. From Vacant Space

The Sophia-Antipolis experiment of development based on high tech activities is
unusual in that the region was one with no industrial or university tradition.
Nevertheless, the region was endowed by significant assets, including an
international airport, a welcoming climate and a cosmopolitan tradition. The
infrastructures were linked to the tourist industry, the main activity of the area.
But nothing predisposed the region to economic development based on high
technology.
The experience of Sophia-Antipolis is the outcome of "small events" that
occurred during the nineteen sixties. It was believed in the first instance that "sun
and sea" could attract enterprise in research and leading edge technology, as had
occurred in the United States. This view was reinforced by the installation in 1960
of an IBM research centre near Nice, followed by a similar move by Texas
Instruments affording credibility to the concept of a new mode of industrialisation
based on the sun belt effect. These ideas might well have come to nothing without
the influence of one man, Pierre Laffitte, then Director of the renowned "Grande
Ecole", "1'Ecole Nationale des Mines de Paris". Laffitte had a utopian vision of a
rural "Quartier Latin", a "City of Science and Wisdom" and proceeded to embody
these ideas in a specific project. In addition to the initiative of Pierre Laffitte,
another "small event" proved to be a crucial impulse. This was the deep
involvement of France Te/(!com in the project and the creation in Sophia-Antipolis
of an infrastructure which allowed nodes of a national and international
communications networks, giving to the area an important comparative advantage
at this time.
The embodiment of Pierre Laffitte's ideas took place rapidly on a dedicatt:d
zone located in a forest of pine trees. But the viability of the project, run by a
private association, soon came up against financial constraints. From the
beginning of the 1970's the direction of the project was transferred to the public
sector, with a major role for local authorities, through the creation of a syndicate
gathering, the towns on which the project has been settled, the Departement des
Alpes Maritimes and the Chamber of Commerce. The syndicate is responsible for
the general management, the financial policy, the international relations, the
promotion and the services to corporations; the control of operations has been
delegated to a semi-public company, the SAEM Sophia Antipolis. The state was a
leading actor for a period when the project benefited from support from the
DAT AR, the French regional planning authority, whose involvement allowed
directing investments in the area.

2.2.2. Take-off Stage

The take-off stage for Sophia-Antipolis was based predominantly on the location
of external organisations on the site which were attracted by the quality of the
infrastructure made available to them. The implant companies had their own
internal resource base but this was but external to Sophia-Antipolis. This form of
development resulted in minimal interaction and interdependence between the
 From Exogenous to Endogenous Local Development 227

organisations on the site. Decision centres and industrial strategy making were
external to Sophia-Antipolis itself.
However, the participation of the public authorities in the initiative made
possible a shift in the scale of the project. After the first period where large public
sector companies and laboratories were introduced onto the site, linked to the
French decentralisation movement, new locations emerged as the outcome of a
commercialisation strategy for the science park based on a major international
marketing effort, mainly directed at the USA. Large multinational companies,
mostly American, adopted Sophia-Antipolis either as their European
administrative base or as an R&D centre charged with adapting their products to
European markets. The availability of high-quality telecommunication
infrastructures able to support hubs and Nice international airport explain most of
these location decisions. These international firms have made major investments,
establishing self-contained entities able to operate without local linkages.
In this take-off stage the "Ecole Nationale Superieure des Mines de Paris a
Sophia Antipolis", the "Institut National de Recherche en Informatique et
Automatique" (INRI A) , institutes from the "Centre National de la Recherche
Scientifique" (CNRS), were decentralized from Paris. After the initial impetus of
France Telecom, the computer centres of large firms like Air France or group
insurance like Organic located in the area, followed by Thomson, Digital
Equipment, Rohm and Haas, Dow Corning, Allergan etc.

2.2.3. "Satellite Platform"

The mass effect created by the process of accumulation of external activities has
obviously resulted in internal dynamics and positive feedbacks in Sophia-
Antipolis, which will be analysed later, but basically the project has grown as a
periphery, a "satellite platform" in Markusen (1996) taxonomy, that is, a system
directed from outside, rich in external connections but deprived of internal
linkages, according to corporate logic independent of any considerations of
economic exchange with others economic actors on the site (Longhi and Quere
1993). Nevertheless, a process reinforcing early expansion set in. As shown in
Figure 1, the take-off of the project is emerging in the seventies, and the eighties
are characterized by an important process of growth, feed by new locations; this
first phase will be followed by a severe slowdown to be discussed further.
228 C. Longhi

AnnuaJ Total
Growt

2000
2000
a

a a

Fig. 1. Sophia Antipolis :growth of employment.

Source: SAEM Sophia Antipolis.

The process of accumulation initially resulted in the emergence of three

dominant clusters of technological activity. The first set of activities is made up of
the computer science, telecommunications and electronics cluster. This provided
the impetus for the growth of the project and structured the development of its
industrial context. These activities now represent around 75% of technology jobs
in the technopole, with a balance of large French and international (centres of
R&D) companies on the one hand and units of large research institutions on the
other. These soon reached a critical mass with positive feedback effects.
The second cluster of activities is related to life sciences and health sciences,
which gradually built up a significant presence on the site, without actually
reaching the coherence of the previous cluster. Mostly large international
companies were located in Sophia Antipolis to coordinate their European activities
from there and to adapt their products to the "European" market, without no
significant local links. As a result the activities and specialities to be found are
highly diverse and the critical mass to set off an endogenous growth dynamic has
never been reached. The third cluster of activities is linked to natural sciences,
environment; historically, it was first to emerge in Sophia Antipolis, but without
further growth, and constitutes today a small share of the activity.
 From Exogenous to Endogenous Local Development 229

2.2.4. Toward a Technopolitan Area

Limits to growth in Sophia Antipolis were closely linked to the initial conditions
of the project, i.e. lack of a local labour market, lack of endogenous creation of
resources, of SMEs or specialized services. Nevertheless, the accumulation of
economic activities on site has triggered positive processes and catch lip effects
which smoothed progressively these limits until the beginning of the nineties,
when a crisis emerged in Sophia Antipolis
A first element concerns the site's potential for training and research. The sitting
of the first public institutes on Sophia Antipolis was the result of a decentralisation
policy of the major research institutes: the "Ecole des Mines", the CNRS, the
INRIA. Their role in the development of the project was very important; INRIA in
particular played a fundamental role in structuring local technological
development, the emergence of an endogenous innovative capacity and of start-
ups on the site. But this potential was not sufficient to train the graduates needed
by the developing labour market. A significant development was the sitting in
Sophia-Antipolis of research institutes and doctoral studies of the University of
Nice in 1986. The university is young and was not associated to the development
of the park. Its research institutes and doctoral training began to move from Nice
to Sophia from 1986, and progressively a critical mass was reached. The
development of university training for engineers has given impetus to the local
labour market. In this respect, the process of development of Sophia-Antipolis was
the reverse of the "classical" science park processes: the development of the
University and the nature of specialist education received their impetus from the
accumulation of industrial activity. The large companies arrived first, with
decentralised research institutes and only then the university and small and
medium high tech enterprises.
A second element concerns the high tech small firms which have grown up in
Sophia-Antipolis. The origin of the high tech small firms is very diverse. Many
were attracted to Sophia-Antipolis by a site suitable for growth or to build
linkages with research institutes (in which their technologies have often been
born) or with firms (as subcontractors providing specialized services). Most of the
success stories of the park were such implants. Others were spin-offs from the
research institutions (INRIA, Ecole des Mines, INSERM in health sciences). The
university has also had some successful spin-offs, leaders in their niche. They
retain close relationships with research departments, and often try to build local
networks around their activities. Pre-production and production of their range of
products are often subcontracted to different firms in the region; they are linked
with local service activities (such as publishing, translation). These spin-offs were
mainly concentrated in computer sciences and telecommunications.
There are also some high tech firms which have located in Sophia-Antipolis to
benefit from the prestige of the address without having any links with the local
industrial environment. Those that did have linkages were concentrated in the
telecommunications sector, in GIS and image processing, software, or biomedical
activities. Some acquisitions of high tech SMEs have been made by large firms to
renew their technological base, mainly to shift from centralized to decentralized
230 C. Longhi

systems of information. On the contrary, spin-offs of large firms have been very
scarce on this period. The strategy of internalisation of human resources within
large firms and the attractiveness of jobs in these firms have resulted in a "reverse
spin-off effect", i.e. their systematic recruitment of high qualified resources from
existing SMEs to feed their growth. This characteristic of the project has severely
constrained the development of a local labour market and of localized knowledge.
A final element is made up of service activities. This cluster is heterogeneous
and its growth has been considerable since 1982 in terms of jobs, so that it makes
up 28% of current employment and more than 50% of all firms. This growth is
based in part on a catch up effect, due to the absence of service provision on a
greenfield site like Sophia-Antipolis; they are basic services required by those
living and working on the site and specialized services required by the growing
accumulation of firms. The two former categories of firms have made a
considerable contribution to the viability of the development process, even if part
of service growth has also been make up by a third category of firms without any
links with the technological activities of the site. These are essentially the firms
linked to BTP or marketing and distribution which have located on in the
periphery of Sophia-Antipolis to take advantage of the prestige of the technopolis
or of favourable real estate conditions.
These different elements are very important as they contribute basically to the
reversal of previous established trends and to the building of an innovative milieu.
But the nineties have questioned these positive developments.
The shocks that characterized the early 1990s have indeed been of considerable
importance for Sophia-Antipolis. First, it has hit the engine of growth of the
project, the computer science activities, implying downsizing in the large firms.
Second, it has questioned the strategy of development of the project, based on a
marketing strategy aiming at attracting investments of large firms. Indeed, the
nineties brought on new forms of globalisation. The 1980s were characterized by
multinational developments driven by market processes. The major corporations
organized themselves into semi- autonomous regional groups. The attractiveness
of Sophia Antipolis for inward investment gave it a comparative advantage over
alternative locations. Corporations were not concerned with the specific
competencies or capabilities of regions in choosing their location, but were
influenced mainly by general conditions in terms of cost and facilities. As we have
seen, Sophia-Antipolis was well endowed in this respect.
The nature of this process of globalisation in high tech activities has however
dramatically changed in the 1990s (Gordon 1996; Veltz 1993). It is no longer
mainly oriented towards manufacturing products or supplying particular markets
through adaptation of existing home technologies but towards the process of
innovation. As innovation depends on diverse specific capabilities and knowledge,
locational differences in these specificities become important. Locations are not
easily substitutable, and choices of location are not driven by general advantages
but by specific advantages of regions in terms of core competencies. The regime
of growth of Sophia Antipolis, based on the attraction of new resources, was
clearly a strategy for coping with the 1980s but at odds with the 1990s. It resulted
in the location of more conventional activities without any technological content,
 From Exogenous to Endogenous Local Development 231

and in the depreciation of the park's image. A new strategy of development, based
on local resources and competencies, had to be generated.

2.2.5. Emergence of an Innovative Milieu

After a period of instability in the beginning nineties, different new trends have
emerged in the area. Mainly, the unexpected consequence of the crises has been
the change in the regime of growth, from exogenous to endogenous driven
processes, and the development of local relations between firms or firms and
research institutions involved in 10caIly implemented innovative processes. A
main actor of this change has been the professional associations existing on the
park, which stood for the absence of clear public strategy to face the crises and to
reconfigure the local system.
Various business associations and clubs exist in Sophia-Antipolis. Their role has
evolved considerably, from defence of specific interest to more coIlective
purposes. The Telecom VaIley Association was the first initiative in this direction
from the large firms; created by Texas Instruments and France Telecom, it now
includes all local actors in the field of telecommunications, both public and
private, large and small. In the face of the dimming of the high-technology image
of Sophia-Antipolis, and the risk of relocations away from the park, the
Association has attempted, on the basis of the important amount of resources
accumulated 10caIly, to label it as the strategic European location for
telecommunications. It has brought about a positive dynamic with the large
companies beginning to open themselves to the other members of the Association,
including smaIler firms, to share certain technological problems and resources, to
take advantage of complementary expertise and businesses. It can be argued that
the park is indeed today effectively a strategic centre for telecommunication
activities in Europe, with a specific advantage on wireless technologies. The main
European institutions defining the standards for telecommunications (ETSI) and
internet (WWW) are located in Sophia-Antipolis, which is thus at the heart of the
innovative process and a strategic area to gather information on the future
technologies.
Numerous associations and clubs exist in fact in Sophia Antipolis, linking SMEs
(Imet, MITSA), research institutes (Persan), specific technologies (Club High
Tech), with the explicit aim of generating networking and processes of local
coIlective learning. These local associations and clubs embody in fact the ideas of
cross-fertilization developed by P. Laffitte when creating Sophia Antipolis; they
are fundamental elements in the coordination of the innovative activities of firms,
providing information on technologies as weIl as on markets, in the building of
local ties.
Another important trend to be noticed is an important movement of creation of
new SMEs, which was lacking in Sophia-Antipolis. As stressed by Keeble et al.
(1998), the creation of new high-technology SMEs and spin-offs is a "major
process whereby research ideas, technological innovations and expertise are
diffused and shared within a region, rather than being confined and internalised
within a single, secretive firm". This process is indeed conducive to coIlective
232 C. Longhi

learning and local positive feedback. The history of Sophia-Antipolis has revealed
that in the past opposite processes were in fact at work, namely internaIisation of
resources within large firms.
Historically, the process of spin-off was restricted to research institutions. With
the crisis, it has been extended to large firms. Paradoxically, the most important
movement of creation of new SMEs correspond to the major slowdown of the
project in terms of employment growth at the beginning of the nineties, and to the
process of downsizing and outsourcing induced by large firms. The number of
firms increased significantly during the crisis. Downsizing resulted in the
diffusion of capabilities up to there internalised in the large firms through the local
labour market, and not in a loss of resources. Indeed despite the weakness of
linkages between firms or institutions a strong sophipolitan milieu effect has
always existed for the people involved in the project, which have tried to find
local solution to the crisis. A lot of "forced" spin-offs from large firms have
emerged, which have helped to diffuse technological capabilities and competences
locally. These firms, created by former engineers of large firms (Thomson, Texas
Instruments, DEC) reluctant to leave the project when confronted to
"downsizing", have often subcontracting relationships with their parent firms, and
important informal networking relations between them, to share problems and
solutions. The fact that most of these SMEs have been able to adapt and grow has
enhanced the movement.
Recently, numerous local start-ups and spin-offs have emerged in the software
industry, multimedia, telecommunication and internet technologies. An important
resource base exists locally in these activities. Indeed, many of these new
technologies are developed in INRIA or France Telecom. The consequence has
been the multiplication of small and micro firms created from local resources,
with strong links with research and close interrelationships. This cluster is
supported by local business associations, and is embedded within numerous
different local formal and informal networks.
This trend towards an endogenous regime of growth could not have existed
without the development of the training and research capacities of the area, and
their interdependence with the industrial context. The existence of a local labour
market, allowing the skilled labour mobility, the diffusion of embodied and tacit
expertise and technological know-how is recognized in the literature as the basic
element of local development (Camagni 1991; Keeble et al. 1998). As we have
stressed, Sophia Antipolis was initially deprived from such processes. The
changes initiated by the location of the University has never stopped, and the
growth and multiplication of higher education and research institutes is also
quantitatively one of the important trend of the recent years. Roughly 3000
students and as many researchers are today in Sophia-Antipolis. The advent of the
Institut National de la Recherche Agronomique (INRA) and of the "Institut de
Pharmacologie" (CNRS) significantly reinforced the pharmaceutical and health
science cluster of activities, even if industry-research linkages were in both cases
mainly turned towards outside players. The critical mass effect and the visibility
of these developments were insufficient to promote innovative local behaviour.
This critical mass has on the contrary been reached in the information
 From Exogenous to Endogenous Local Development 233

technologies with the multiplication of institutes developing specialized doctoral

studies, the creation of new schools of engineers from the University (ESSI,
ESINSA) or from professional consortia (THESEUS, EURECOM). This potential
of higher education and research is continuously growing; with the University, the
research institutions (CNRS, INRIA, INRA, INSERM ... ), the Ecole des Mines,
the research and training impulsed by France Telecom, the business schools
(CERAM), most of the major French education and research institutions are
currently represented in Sophia-Antipolis.
In addition to the qualified resources supplied by the higher education
institutions, the process of downsizing and externalisation induced by the large
firms has resulted in the reinforcement of this growing labour market, and made
qualified resources available to feed the growth of new firms on site, which has
been exponential in different cases. But clearly, this local labour market is
restricted to the sectors of information and telecommunication technologies. In
this sector, the turnover is also very important, showing that new behaviours are
nowadays set up in Sophia Antipolis. For example in 1997, the evolution of local
information technology employment revealed a net growth of + 1,512 (l,460 of
which was from existing local firms); this net total is the result of 2,596
recruitments and 1084 resignations, revealing a considerable turnover and
mobility of skilled labour. As stressed by Storper (1993) in an other context, these
processes of turnover as well as spin-offs are made possible because of the
existence of an already institutionalised industry, not only technologically, but
also in terms of a specific professional culture, of the existence of human capital
specific to an industry. They allow the existence of a local labour market which,
together with the local inter-enterprises relationships and spin-offs processes,
impulse the emergence of local specific competences belonging to the innovative
milieu itself, and not only of individual firms.
The last important change concerns the key actors of the development of
Sophia-Antipolis, i.e. the large firms. As we have emphasized, the growth of
Sophia Antipolis has resulted from the coincidence of a supply of general
advantages (physical infrastructures, climate ... ) and the needs induced by the
multinationalisation of large firms aimed at organizing their European operations.
The huge decrease in the arrival of new multinational firms and new investments
in the beginning of the nineties has also been the indication of the crisis. The
viability of the experience has been dependent on its capability to induce a shift in
the regime of growth, from accumulation of external resources to the development
of activities based on local specific competences. The attractiveness is no more
only dependent of physical infrastructures, but mainly from specific capabilities
and human infrastructures. The location of large firms is nevertheless still crucial
to the project, but it takes a different sense, it eventually validates its technological
and innovative potential and its position of "pole of excellence" in some fields,
and constitutes opportunities for the existing local labour market.
Indeed, the arrival of new firms does not now involve neither huge investments
nor simply adaptation of existing products or services. The new firms usually
settle in the park as tenants with small units and grow locally drawing on local
capabilities and on the existing local labour market to develop their innovative
234 C. Longhi

activities. For instance, Andersen Consulting located on the park with a small unit
of ten people and has subsequently grown locally to employ more than three
hundred staff on the base of new markets created from their local activities. The
same strategy has been followed by the new comers in Sophia Antipolis, whose
location have been clearly induced by the specific capabilities in wireless
communication, GSM technologies or internet technologies which have grown
locally, from the public research institutes as well as from the R&D developed in
the area by large (Texas Instruments, IBM) or medium (VLSI for instance) firms.
Most of the main actors in this innovative field have recently located in Sophia
Antipolis, with a precise strategy of mobilization of local capabilities and
knowledge. This is the case of Siemens, Lucent, Bay networks, Ericsson, whose
units have grown on the basis of local resources, or Philips, which acquired VLSI,
but also of many small firms which find in Sophia Antipolis an area suitable for
growth or for attracting easily the specialized human resources needed in this
innovative technological field. The process of outsourcing and the development of
information technologies have also considerably changed the local strategy of the
large firms; a lot of local relationships linked to the innovative process have
emerged locally, through subcontracting with high tech specialized SMEs.
Growth is now quasi-exclusively endogenous, i.e. results from the growth of
existing local firms, when before 1990, it resulted mainly from the balance
between entry and exit of firms in the project. Besides, the growth is concentrated
in the information technologies sector. An efficient local labour market thus exists
today for qualified information technology staff. More than half of sophipolitan
jobs are highly qualified jobs (executives and engineers); in the information
technologies sector, this proportion was 70.8% in 1997.
In the contemporaneous process of innovation, the challenge for Sophia
Antipolis was to enter the global networks set up by the high tech firms to sustain
their innovative activities, which links large and small and specialized firms as
well as public research institutions. The emergence of a new regime of growth and
of new institutional arrangements after a crisis of adaptation show that the
resources accumulated in the eighties were sufficient to face the shocks and
reconfigure the system. But the shocks of the nineties have resulted in an
unbalanced local system, in which information and telecom activities have been
able to adapt to the new environment and to regenerate growth while other
activities (health sciences) are still locked into the previous regime and failing to
fulfil their potential. Information technologies have recovered the role of engine of
growth they had in the first phase of the project. The engine is no more made of
self-contained establishments of large international firms; it results from the
positive feedbacks induced by the local interactions of firms involved in
innovative activities. This strong movement of specialization of Sophia-Antipolis
can raise problems for the balanced growth of the area, as it could increase its
sensitivity to asymmetric shocks, despite its specialization in R&D and design
activities.
 From Exogenous to Endogenous Local Development 235

3. Conclusion

Fundamental features of territories, and specifically high tech complexes like

Toulouse or Sophia Antipolis can be related to the processes through which
resources are created and mobilised. The two areas have many differences and
many similarities. They are different in terms of size and scope, of industrial
activities, or of socio-economic features. They have in common to be
technopolitan areas born from state-led processes, government and local policies
based on decentralisation of large public firms and institutions, exogenous
processes of growth whose unintended consequences have been the emergence of
endogenous processes of creation of resources, of innovative milieux in different
clusters of high tech activities. The contemporaneous arrangements of the
innovative milieux could not be planned, they are the consequence of adaptations
and changes of the local systems, of reconfiguration of their economic working
following evolving internal and external circumstances. Diversity, usually
considered across different areas, is also a characteristic of the dynamic evolution
of the local systems, where reconfigurations have given rise to new internal
economic and institutional arrangements, new coherences with evolving
environments. Key actors, collective agents play a fundamental role in the
bifurcation of the enacted systems that are territories; turning points of evolution
of Sophia Antipolis and Toulouse towards technopolitan milieux are examples of
these dynamics.
Toulouse had a long university tradition closely associated to economic
development and a constituted industrial base in aeronautics, a strong socio-
economic system, as well as an identity which explain the further regional and
industrial policies that have led to a technopole. Sophia Antipolis on the contrary
stood as a vacant space, and the regional policy intended to create a new socio-
economic context and change the nature of the development. The reactions to
external shocks and internal reorganisations of the local systems have been thus
very different, from coordinated initiatives linking the local governments and the
public and private institutions in Toulouse, to unexpected self-organized processes
in Sophia Antipolis.
The viability of local development based on high tech activities is gone with the
steady reinforcement and specialisation of the areas in activities of conception,
R&D, problem-solving processes, as well as the increasing necessity of different
technological expertises in the economic process. The continuous confrontation of
non-formalized knowledge and collective learning feeding the innovative process
of high technologies has strengthened the local systems and the decisive role of
proximity. The two areas have built in their technological fields specific
competences that have attracted new investments and inserted the local systems in
global networks of innovation. The structuring of the local and global dimensions
of the innovative process is indeed the key of the viability of contemporaneous
complex of high technology.
In Toulouse, institutional thickness - i.e. a dense and coherent set of territorial
relationships of all the economic actors, private and public, a specific culture, and
236 C. Longhi

a shared representation system and development strategy - has been one of the
main asset of development, but the role of the State has been permanent, because
of the strategic content of the local activities. In turn, the embeddedness of the
European aerospace industry in Toulouse, its pivotal role in the European scene,
induced by the endogenous process of creation of resources and local specific
knowledge, explains certainly the restructuring of the industry allowed by the
privatisation processes implemented by the government. This basic change in the
institutional arrangements governing the economic activities of the city will
certainly deepen its insertion in the global knowledge-based economy. In Sophia
Antipolis, the development is the result of the government regional development
policy, without the strategic content present in Toulouse. The emergence of an
innovative milieu from the accumulation of external resources has been the result
of an external shock, a change in the environment the project had to face. The
reconfiguration that resulted has allowed the project to go beyond the image effect
on which it has grown, towards the emergence of a socio-economic context and
the realization of the cross-fertilization concept which presided to its creation.

References

Allinne, 1. P. (1996), "Les technopoles soot-elles centenaires? Systcmes innovateurs et

creation d'entreprises en France", Revue d'Economie Regionale et Urbaine, 5, pp. 1870-
1960.
Andersen, E. S. and Teubal, M. (1999), High Tech Cluster Creation and Cluster Re-
configuration - A System Policy Perspective, Mimeo, DRUID, June.
Arthur, W. B. (1989), "Competing Technologies, Increasing Returns and Look-In by
Historical Events", The Economic Journal, Vol. 99.
Becattini, G. (1990), "The Marshallian Industrial District as a Socio-economic Notion", in:
Industrial districts and inter-firm cooperation in Italy, Pyke, F. and al. (eds), Geneve,
International Institute for Labour Studies.
Brusco. S. (1982), "The Emilian Model: Productive Decentralisation and Social
Integration", Cambridge Journal of Economics. vol. 6, pp. 167-184.
Camagni, R. (1991), Local "Milieu", Uncertainty and Innovation Networks: Towards a
New Dynamic Theory of Economic Space, in: Innovation networks: spatial perspective,
Camagni, R. (ed.), London, Belhaven Press.
DRIRE - INSEE (2001), L 'industrie en Midi-Pyrenees, Dossiers Etudes n° 104.
DRIRE (1998), Les industries de I 'aeronautique et du spatial en Midi-Pyrenees.
Dupuy, C. and Gilly, J. P. (1999), Industrial Groups and Territories: the Case of Matra-
Marconi-Space in Toulouse, Cambridge Journal of Economics, 23.
Garnsey, E. (1998). "The Genesis of the High Technology Milieu", International Journal
of Urban and Regional Research, vol. 22 n° 3.
Garnsey, E. and Longhi, C. (1997), High Technology Locations and Economic
Development, EUNIT (European Network on Industry, Innovation and Territory)
International Conference on Industry, Innovation and Territory, Lisbon.
Garnsey, E. and Longhi, C. (1999), "Auto-organisation et emergence des milieux
innovateurs", Revue d'Economie Regionale et Urbaine, I.
 From Exogenous to Endogenous Local Development 237

Gordon, R. (1996), "Industrial Districts and the Globalization of Innovation: Regions and
Networks in the new economic space", in: Wealth from Diversity, Vence-Deza, X. and
Metcalfe, J. S., Kluwer.
Gormand, C. (1993), L 'industrie aeronautique et spatiale, Paris, L' Harmattan.
Grossetti, M. (1990), "Enseignement superieur et technopole, l'exemple de l'informatique a
Toulouse", Revue Fram;aise de Sociologie, XXXI.3.
INSEE (1998), 15 ans d 'aeronautique et d 'espace en Midi-Pyrenees, Toulouse.
INSEE (2000a), "Aeronautique, espace et sous-traitance", Reperes, 100.
INSEE (2000b), Tableaux economiques de Midi-Pyrenees, Toulouse.
INSEE (2001), "Les NTIC en Midi-Pyrenees", 6 Pages, 46.
Kechidi, M. (1996), "Coordination inter-entreprises et relations de sous-traitance: Ie cas
d' Aerospatiale", Revue d'Economie Regionale et Urbaine, 1.
Keeble, D. et al. (1998), Collective Learning Processes and Inter-Firm Networking in
Innovative High·Technology Regions, ESRC (Economic & Social Research Council)
Centre for Business Research, Working Paper 86, March.
Longhi, C. (1999), "Networks, Collective Learning and Technology Development in
Innovative High Technology Regions: the Case of Sophia-Antipolis", Regional Studies.
Longhi, C. and Quere, M. (1993), "Innovative Networks and the Technopolis Phenomenon:
the Case of Sophia-Antipolis", Environment and Planning C: Government and Policy.
Luger, M. I. (2001), Science and Technology Parks at the Millennium: Concept, History,
and Metrics.
Markusen, A. (1996), "Sticky Places in Slippery Spaces: A Typology of Industrial
Districts", Economic Geography.
Metcalfe, J. S. (2000), "Propspects and Challenge for Research on Innovation", Co-
evolution o/Systems a/Innovation, Conference, Berlin, June 8th-9th.
Muller, P. (1988), Airbus, I 'ambition europeenne. Logique d'Etat, logique de marche.
Paris, L'Harmattan.
Porter, M. E. (1990), The Competitive Advantage o/Nations, New York, Free Press.
Saxenian. A. (1994), Regional Advantage, Cambridge, MA, Harvard University Press.
Storper, M. (1993), "Regional "Worlds" of Production: Learning and Innovation in the
Technology Districts of France, Italy and the USA", Regional Studies, vol. 27.5.
Talbot, D. (2000), Institutional Dynamics and Localized Inter-Firms Relations. The Case of
Aerospatiale and its Subcontractors in Toulouse, European Urban and Regional Studies,
7(3).
Veltz, P. (1993), D'une geographie des couts a une geographie de l'organisation. Quelques
theses sur l'evolution des rapports entreprises/territoires, Revue Economique, n04.
Biotechnology Development in Germany:
The Case of Nordrhein-Westfalen

Francesco Salamini, Anke Sohn and Hartmut Thomas

Abstract. Nordrhein-Westfalen is and has been a centre of excellence for

chemical and genetic research for a long time; advanced biology has been added
recently. The industrial development of the latter sector is logarithmic here. There
is no decline in the demand for research and development either for biomedical or
agricultural applications. The forecast of the future of the biotech industry in
NRW cannot be included in a pattern containing well-defined goals for the various
application fields. In this situation, it is not easy ei~her to estimate the capital
requirements in the long term or set priorities for the region. Structural conditions
to be kept in view are the enhancement of the "public-research-economy"
communication on subjects such as health care, food and agriculture, biotech
products and the community, environment and diagnostics. The improved regional
credit and functionality of the industrial incubators are also under study. On the
other hand, it is a good sign that the supply of venture capital for new start-ups is
rising in NRW; that the public opinion is more favourable than in the past; that the
federal and local governments are in favour of the biotech industry in general and
of transferring research know how to businesses. We can therefore be reasonably
certain that the region will be among the biotechnological hot spots that will
survive in the next fifteen years. The paper reviews the history and the players that
have been influencing the development of biotechnologies in NRW: the Bio-Regio
project; the role of the Bio-Gen-Tec-NRW agency; the first step of biotech
development in NRW; and a short reflection on the future.

1. The BioRegio Initiative

In the 1980s, Germany went through a period of complex - at times fractious -

debate on the future of technological societies. Complex because the discussion on
the function of science in society, traditionally central to German culture, became
more pointed on the heels of the new developments in biotechnology, thus
generating a clearer understanding of the meaning and relevance of bioethics
(Korff et al. 1998). Fractious because the debate was focused almost exclusively
on the risks inherent in genetic manipulations, with a clearly restrictive impact on
the laws that were being enacted to regulate this field. Faced with this situation,
the scientific and business communities reacted in a predictable manner: scientists
and biotechnology investments moved to the United States. However, starting in
1992, small groups of opinion makers, including politicians, businessmen, bankers
240 F. Salamini. A. Sohn and H. Thomas

and scientists, reopened the debate on the risks of biotechnology, broadening it

with discussions of the benefits that could be derived from this new wave of
innovation. Later on in this presentation, we shall discuss certain aspects of this
phase, viewed from the standpoint of the situation that was developing in NRW.
In 1994, the federal government decided to address these issues. The Ministry of
Education, Research and Technology (BMBF; Jiirgen Rlittgers Minister)
sponsored the BioRegio Initiative, which was based on a simple concept: how to
develop an integrated proposal for a regional system that would take advantage of
a wide range of locally available skills in such fields as advanced molecular
research, technology and innovation, and tap investment capital.
The federal government launched a competition, and 17 German states
participated. The proposed models were judged by an international panel. Each
state offered a different solution, availing itself of the support of the federal
government to create the requisite organizational structures (in NRW, the Bio-
Gen-Tec-NRW association). The three states selected as winners (NRW, Bayern
and Rheinland Pfalz) were each given 50 million German marks that they could
use to promote the development of biotechnology (Eichener et al. 2000),
encourage the flow of capital toward new fields and, in general, stimulate job
creation. The stated goal of Minister Rlittgers was to make Germany the leading
European country in the field of biotechnology research and development by the
year 2000. In February 2001, H. Thomas, Chief Executive Officer of Bio-G(!n-
Tec-NRW, could state publicly that Rlittgers's goal had been achieved (Bio-Gen-
Tec-NRW,2001a):
. "A few years ago, we set out to make Germany the European leader in
biotechnology. The objective was ambitious and many had their doubts at
that time, but today, it has become a reality. Based on the number of start-
ups of new biotech companies, Germany is considered worldwide as
occupying the number one slot in Europe: the greatly increased availability
of venture capital and the keen interest of investors in the shares of young
biotech companies, coupled with the set-up boom that is also in progress,
are clear signs of a consolidation and international positioning of Germany
as a biotechnology location".

2. Biotechnology in NRW: The Function of the Bio-Gen-

Tec-NRW Association

Germany got a late start in developing its biotechnology industry but has been
catching up. Today, there are 332 biotech companies, which together employ
11,000 people and generate annual gross revenues of US$670 million. In 2000,
they invested US$600 million (GFW-NRW 2001).
Many of these companies are located in Nordrhein-Westfalen, the most
populated German state (18 million inhabitants), which accounts for 22% of the
country's gross domestic product. It is host to a number of multinational
 Biotechnology Development in Germany: The Case of Nordrhein-Westfalen 241

companies (Bayer, Henkel, RWE, Krupp-Thyssen, Ford, and many others), and its
economy is strongly linked to the chemical industry. The industrial vocation of
NRW is supported by a high concentration of public research facilities: 53
universities (500,000 students); 21 scientific organizations, including Max Planck
and Fraunhofer Institutes and several National Research Centres; 32 other public
agencies financed with government funds; 58 scientific units participating in
special research projects (Sonderforschungsbereich, SFB, which are managed by
the Agency for University Research, Deutsche Forschung Gemeinschaft, DFG);
68 technology centres and industrial incubators; and 27 technology transfer offices
(Anonymous 2001, Bio-Gen-Tec-NRW 2000).
The engine that drives this public research network - the most extensive and
important in Europe - is not chemistry, but biology: eight graduate schools of
medicine, 12 SFB groups, eight interdisciplinary research centres (IZKF), one
Molecular Medicine Centre in Koln (ZMMK), one Biomaterials Centre in Aachen
(BIOMAT), eight graduate schools of biology, two biotechnology institutes in
Julich, two Max Planck Institutes in Koln (for plants and neurological research),
the Max Planck Molecular Physiology Institute in Dortmund, one Molecular
Biotechnology Institute in Aachen (Fraunhofer Foundation), and one Chemio and
Biosensor Institute in MUnster.
The BioRegio competition ended in 1996. Since then, the state of NRW has
been working to turn its biotechnology effort into economic expansion. Today,
the state is host to 130 biotech companies located within an area of 13,000 km 2 ;
220 organizations provide products and services to the biotech companies (GFW-
NRW 2001). The rise of this biotech industry has been followed by the Bio-Gen-
Tec-NRW association, which assisted the birth of 82 start-ups from 1997 to 2000,
requiring a total investment of 635 million German marks (50 from the federal
government, 195 from the local government and 390 from private investors). In
2000, the local government invested 100 million German marks (Bio-Gen-Tec-
NRW 2001a, Anonymous 2001). Fifty-two of the 80 start-ups created with the
support of Bio-Gen-Tec-NRW are located along the Rhine between Ktiln and
DUsseldorf. They engage in a wide variety of activities including purification of
nucleic acids, cell separation, molecular carriers, bioreactors, bioprocessing,
quality control, molecular chips, bioinformatics, protein design, diagnostics,
functional genomics and postgenomics, models of transgenic animals, and
molecules that prevent neurodegeneration and cardiovascular diseases.
The main cities in this area, named "Bioriver", are DUsseldorf, with 180 new
jobs and the Life Science Centre Incubator; K01n, which accounts for 300 new
jobs and is home to Bio-Gen-Tec-NRW, the Max Planck Institute for Plant
Improvement, the Institute of genetics of the university and the East Rhine
Technology (RTZ) Incubator; and Bonn, where four start-up projects were
approved in 2000. The Aachen-Julich region has been the site of interesting
interdisciplinary projects financed with private capital and was recently selected as
the home of the Fraunhofer Institute for Molecular Biotechnology. In the Ruhr
Basin, which specializes in bioprocessing and microstructure technologies, the
BioIndustry agency works to encourage the establishment of start-ups in
bioindustry and medicine. In addition, the city of Bielefeld has access to a
242 F. Salamini, A. Sohn and H. Thomas

Genomic Research Centre that has excellent potential for the establishment of
start-ups.
The Bio-Gen-Tec-NRW association is a private non-profit organization
financed with public funds, private capital and revenues generated by the projects
it manages directly. It also manages public funds earmarked for the development
of biotechnology in the state of NRW. This association, which has 172 members,
provides consulting and support to new start-ups during the establishment and
early development phases. New ventures receive financial support in the form of
federal and local government funds that are provided through Bio-Gen-Tec-NRW
and have access to a wide range of consulting resources. Bio-Gen-Tec-NRW sees
itself as a one-stop shop that can help clients define the scope of their endeavour
more clearly, prepare a business plan, obtain public financing, enter into contracts
with venture capitalists and present their projects at public events (Figure 1).

Management
Information and guidance (*) Patents

Introduction for
Technological Bio-Gen-Tec-NRW and securing
support (*) of private
financing (*)

Political Access to public

interface

Fig. 1. Services provided by Bio-Gen-Tec-NRW, a non-profit association that helps support

the development of biotechnology ventures in the state of Nordrhein-Westfalen.
* functions performed by special committees.

The Management, Consulting and Coaching Department makes use of financial,

legal and tax consultants (42 in 2000). Clients can also use a voucher program to
get help in preparing business plans and receive legal advice. Public funds pay for
75% of the cost of these services. Special attention is paid to securing patent
protection or acquiring patents. Bio-Gen-Tec-NRW organizes seminars devoted
specifically to patent issues and draws on a network of 26 lawyers who specialize
in biotechnology.
The financial network comprises more than 50 German, regional and
international investors, including venture capitalists and bank investment funds.
At the regional level, five companies with expertise in biotech formed a
consortium and invested about 30 million German marks between 1997 and 2000.
 Biotechnology Development in Germany: The Case of Nordrhein- Westfakn 243

Before a start-up project is accepted by Bio-Gen-Tec-NRW, its scientific and

business assumptions must be evaluated and approved by the Biotech Competence
Network. Special attention is paid to financial considerations, including project
risk assessment.
Bio-Gen-Tec-NRW introduces the companies that it sponsors at international
trade fairs. It also posts relevant data on Internet in a directory that contains 360
entries (Bio-Gen-Tec-NRW 2001b) on biotech companies, research institutions
and venture capital firms, along with information on regional research and
development areas. Interaction with the public at large occurs primarily at national
trade fairs, at Biomeetings and through press releases (120 in 1998) and press
conferences. The association also supports the "Genes We Use Every Day"
project, which uses a contest and prizes to disseminate information in the school
system.
The most important financial assistance programs used by Bio-Gen-Tec-NRW
to support the development of biotech companies are listed in Table 1, which
shows the eligible expenditures, the goals of the program and the level of
financing provided by financial institutions to support R&D projects or launch
start-ups. The most recent program is the Bio-Chip Initiative (16 million German
marks), which is funded by two ministries of the local government (Bio-Gen-Tec-
NRW 2000, 2001a).
 tv
t
:TJ
Table 1. Outline of the most important programs available for biotechnology research and startup projects in the state of Nordrhein-Westfalen. en
..,
Program Investments covered Focus % of funding from public capital ;:;
3
~:
~
FEDERAL GOVERNMENT en
o
:r

BIOTECHNOLOGY 2000
"..,
.R&D • Basic research • Up to 100% (institutions) 0-
'"
• Equipment not included • Up to 50% (startup) ;r:
;l
o
3
~
NRW REGIONAL GOVERNMENT

TIP. Support for small .R&D • New high-risk • Up to 35% (startup)

and medium-size businesses • Equipment included technologies • Up to 75% (institutions.
subcontr.)

INCUBATOR • Consulting services • Concept assessment • Up to 50% of consult. contracts

• Market research
• Patents
• Investments

VOUCHER • Consulting services • Pre seed and • Up to 75% of costs

• Market research seed stages
• Patents
Table 1. (continued)
0::1
BIO-CHIP. Support for small • Chips development in: • Preseed and o·
1>
and medium-size businesses - diagnostics seed stages • Up to 75% of costs ::r
"
- plant improvement o=
- neurochips ~
'<
- proteins )?
<
.g"
~
;:.
Ei"
?\l
3
~
':';

~
n
~
o
"....
Z
o
~
":;.
~
"~
OJ'
=
"
~
U\
246 F. Salamini, A. Sohn and H. Thomas

3. The Beginning of Biotechnology Development in NRW

Until the 1980s, biology research in Germany was funded by such public agencies
as the Max Planck Company and Deutsche Forschung Germishaft (DFG), and by
ministries of the federal and local government via research projects. Visible
changes in research policy came during the early 1980s, when the federal
government decided that it would encourage public agencies to manage the
development of cutting-edge technologies, provided their projects involved at least
one local industrial enterprise. In the wake of this decision, the Ministry of
Scientific and Technological Research (BMBF) created three public Gene Centres
(Heidelberg, Mlinchen and Koln). These Centres were to receive 100% financing
provided they met one condition: they would have to show that there was an
official relationship linking the institution responsible for setting up the Centre
with a privately-owned business. The K6ln Gene Centre (Zentrum fur Molekulare
Gen- und Zelltechologie in Koln, 1986, 1990, 1994) showed that one of its
cofounders, the Max Planck Institute for Plant Improvement, received an annual
grant from Bayer, Leverkusen, of one million German marks, which was used to
fund unspecified basic research programs. The three Gene Centres were financed
by the BMBF for three funding periods (1982-1994). The completion of the
program for the establishment of the three Centres was followed by a protracted
debate as to which projects should then be developed and supported. The result
was the decision, supported by Minister Rlittgers, to launch the BioRegio
competition among German states for three financing packages of 50 million
German marks each. A precondition for entering the competition was the
establishment of a regional network of public-sector managers, opinion makers,
scientists from universities and other institutions, private-sector managers, union
officials, and Chamber of Commerce and Industry representatives. An
organizational model was a key part of the competition proposal, which was also
required to include a description of the media that would be used to disseminate
information about new biotechnology projects, as well as a list of all of the entities
engaged in biotechnology work in the project area, showing the scientific and
practical significance of their work.
The definition and organization of the networks was not an easy task. It
involved balancing the flow of financial support from the federal and regional
governments to public - and private - sector institutions with the need to establish
quasi-public collaborative superstructures, while at the same time enlisting public
support for the selection of a given area as a preferred location for the
development of biotechnology businesses. The state of NRW - the city of K6ln in
particular - was in a privileged position. Between 1988 and 1990, it had been the
theatre of a public debate concerning the first release of a GMO in Germany (the
case of the transgenic petunia). In Koln, the media had gradually developed a
negative bias toward an experiment carried out by the Max Planck Institute for
Plant Improvement. In 1992, fearing that as a result of this developing situation
local politicians would choose not to pursue opportunities in biotechnology,
municipal and state officials, Chamber of Commerce representatives, journalists,
 Biotechnology Development in Germany: TIle Case of Nordrhein-Westfalen 247

scientists, union officials and private-sector managers established a joint

committee to coordinate, encourage and channel their efforts to support the
transfer of technology from public research to the private sector in a manner that
would foster the growth of biotechnology enterprises while addressing the role of
science in society. The knowledge gained through this process was instrumental in
creating the Bio-Gen-Tec-NRW network in 1994, which was then used in the
BioRegio competition as a conduit for the joint efforts of different institutions. It
is worth noting that, in this case, the experience gained over the 10 years during
which the three Gene Centres had been in operation and the existence of local
support, voiced (at least in Kbln) by a committee of people who shared the same
clearly-focused objectives, turned out to be determining factors of success: the
three states, where the Kbln, MUnchen and Heidelberg Gene Centres were located,
were chosen to receive the financing packages, in preference to 14 other states.
The less immediate reasons for the success achieved by the state of NRW -
even though initially the environment was particularly challenging and local
industrial development priorities were nebulous at best - are basically two: a high
and widespread level of scientific expertise and the organizational ability of a
group of people representing diverse professional interests who were sustained by
their belief that they were contributing to the transfer of scientific knowledge.

4. It Could Happen Tomorrow

The state of Nordrhein-Westfalen has been and remains a centre of excellence for
research in chemistry and genetics, to which it recently added advanced biology.
Industrial development in advanced biology is currently in a logarithmic phase.
There appears to be no slackening of demand for research and development for
biomedical and agricultural applications. It is clear, however, that current forecasts
of the future of the biotech industry cannot be pigeonholed within a model that
sets well-defined objectives for each separate field. As a result, it is difficult to
forecast long-term capital requirements and to define priorities at the state level.
However, close attention should be paid to key structural developments, including
an intensification of the debate involving the public, the research community and
businesses on such issues as medicine and health, food and farming, biotech
products and society at large, and the environment and diagnostics. The increased
availability of local credit and the success of industrial incubators should also be
taken into account. Positive developments in the state of NRW include an increase
in the availability of venture capital for start-ups, a more favourable public
opinion and a more constructive attitude by the federal and local government
toward the biotechnology industry in general and the transfer of research to
businesses in particular. Therefore, it is reasonable to expect that this German state
will become one of the biotechnology hot spots (among perhaps 20 or 30) that will
survive worldwide over the next 15 years.
248 F. Salamini, A. Sohn and H. Thomas

References

Anonymous (2001), "Spotlight on NRW", Nature. supplement March 22, Vol. 410.
Bio-Gen-Tec-NRW (2000). lllJo. no. 1+ 2, Bio-Gen-Tec-NRW.
Bio-Gen-Tec-NRW (200Ia), Allllual Report 2000, Bio-Gen-Tec-NRW.
Bio-Gen-Tec-NRW (200Ib), Bio-Cell-Tec Atlas 2001, Bio-Gen-Tec-NRW.
Eichcner, V., Schaaf, S., Schulte, F. and Weingarten, 1. (2000), ErJolgsJaktoren for die
Biotechnologie-Regionen. SozialwissenshaJtliche Begleitforschllng zu den BioRegios,
Hans Backler Stiftung.
GFW-NRW (2001), Welcome spezial. Biotech-NRW 2001, GFW-NRW.
Korff. W., Beck, L. and Mikat, P. (eds) (1998), Lexikon der Bioethik. 1m AIIJtrag der
Coerres-CesellschaJt, GUtersloher Verlagshaus.
Zentrum fUr Molekulare Gen- und Zelltechnologie in Kaln (1986), Abschlussbericht 1982-
1986. Forderllngsvorhaben BCT0365 2 (BMBF), Max-Planck-Institut fUr
ZUchtungsforschung, Institut fUr Gcnetik der Universitat zu Kaln.
Zcntrum fUr Molekulare Gcn- und Zelltechnologie in Kaln (1990), Abschlllssbericht 1987-
1990. Forderlllzgsvorhaben BCT0365 2 (BMBF), Max-Planck-Institut fUr
ZUchtungsforschung, Institut fUr Genetik der Universitat zu Kaln.
Zentrum fUr Molekulare Gen- und Zelltechnologie in Ki)ln (1994), Abschlussbericht 1991-
1994. Forderlllzgsvorhabell BCT0365 2 (BMBF), Max-Planck-Institut fUr
ZUchtungsforschung. Institut fUr Gcnctik dcr Universitat zu Kaln.
Support for Technology-Based Firms:
The Role Played by Property-Based
Science Parks

Don Siegel, Paul Westhead and Mike Wright

Abstract. To encourage the formation and development of independent

technology-based firms property-based Science Park initiatives in the United
Kingdom have been established. The performance of Science Park and off-Park
firms was compared with regard to several indicators. Also, the role played by the
Science Park manager/director in the development of firms located on "managed"
and "non-managed" Science Parks was explored.

1. Introduction

Policy-makers and practitioners are investing in people and are encouraging

locally based solutions to create "learning regions" (Keane and Allison 1999).
They are particularly interested in encouraging R&D and innovative intensive
areas. Policy-makers and practitioners believe new technology-based firms
(NTBFs) playa pivotal role in competitiveness, job generation, innovation, wealth
creation and regional development Organisation for Economic Co-Operation and
Development (OECD), 1993; Department for Trade and Industry (DTI), 1998;
(Tether and Storey 1998).
The phenomenal growth in the number of NTBFs (and resultant employment)
around, for example, university technology-based incubators such as Stanford
University in Palo Alto and Massachusetts Institute of Technology (MIT) in
Boston (Roberts 1991; Saxenian 1994) in the United States provided a role model
for the development of Science Parks in the United Kingdom. Adherents of the
"linear model" assume a chain of successive interrelated activities. These begin
with basic scientific research, pass through applied and more developmental
research activities, the development of new product and process ideas, the
evolution and testing of prototypes, to commercial production and finally to
diffusion. As a result, the local technological environment can attract and
encourage the formation and development of technology-based firms in localities.
This view has, however, been questioned. Massey et al. (1992) have asserted that
there is no simple linear relationship between science and production, and no
single model of the innovation process. Nevertheless, the linear model commands
widespread support and continues to guide the strategies selected by policy-makers
and practitioners.
250 D. Siegel, P. Wcsthead and M. Wright

Cooke and Morgan (1993) have asserted that "industrial districts" of firms
(Castells and Hall 1994) can be created by public policy. To stimulate NTBFs,
successive British Governments have encouraged (in conjunction with universities,
higher education institutes (HEIs), local authorities and various financial
institutions) the development of various property-based incubator initiatives.
Science Parks represent one form of property-based incubator initiative. After a
period of enthusiastic appreciation, scepticism is growing surrounding the value of
sponsored environments (Flynn 1993) and property-based initiatives (Massey et al.
1992; Felsenstein 1994; Jones 1996; Oakey and Mukhtar 1999). There is still
insufficient knowledge about the formation of "hub" (i.e., a Science Park acting as
the generator of externalities for tenant firms) and "spoke" (i.e., firms located on
Science Parks) districts. Despite considerable investment in the Science Park
movement, limited empirical research has focused upon the contextual and process
issues associated with the behaviour and attitudes reported by entrepreneurs who
own independent firms located on property-based Science Park initiatives.
This paper makes a contribution to this debate. To assess the benefits of
property-based Science Park initiatives in Great Britain, a demand-side perspective
was utilised. Information was gathered surrounding the perceived demand-side
advantages of a Science Park location. The performance of independent
technology-based firms located on Science Parks was compared with the
performance of a "control group" of similar firms located off-Park. Performance
contrasts between the two groups of firms were explored with regard to several
indicators reported by owners of independent Science Park and off-Park firms. The
role played by the Science Park manager/director in the development of firms
located on "managed" and "non-managed" Science Parks was explored.

2. Science Parks in the United Kingdom

Property-based initiatives that can be regarded as Science Parks have been

identified and defined by the United Kingdom Science Park Association
(UKSPA). According to UKSPA (1999), a Science Park as a property-based
initiative includes the following features: it encourages and supports the start-up
and incubation of innovation-led, high-growth, knowledge-based businesses; it
provides an environment where larger and international businesses can develop
specific and close interactions with a particular centre of knowledge creation for
their mutual benefit; and has formal and operational links with centres of
knowledge creation such as universities, higher education institutes and research
organisations. Within this context, Science Parks in the United Kingdom may
provide an intellectual and physical infrastructure as well as management support
that is actively engaged in the transfer of both technology and business skills
towards their clients.
Cambridge and Heriot-Watt Science Parks were established in the United
Kingdom in 1972. In 1983, seven Science Parks were fully operational in the
 Support for Technology-Based Firms: The Role Played by Property-Based Science Parks 251

United Kingdom. By 1989, there were 32 Science Parks in the United Kingdom
and this plateau was maintained until 1993 (UKSPA 1999). In 1999, the United
Kingdom had 46 fully operational Science Parks. This "new wave" growth in the
Science Park movement is mainly due to the fact that the "new universities",
previously the polytechnics, began to establish Science Parks.
Carter (1989) identified three types of management strategies leading to the
development of Science Parks in the United Kingdom. First, is a university led and
funded strategy (which is the least common model) in which a university or HEI
establishes a Science Park on its own and continues to be responsible for its
development and/or management. Second, is a joint venture company strategy,
with the Science Park being run by a separate legal entity. This mechanism ensures
a continuing positive role for the HEI and other investors in the joint venture
company. Third, and most common is a co-operative venture strategy where
partners work together within a flexible and informal framework. This is the
strategy with the least input from the HEI and the academic involvement tends to
be fairly limited, often via the HEls industrial liaison officer. Services available to
Science Park firms can be delivered through three main types of management
agreements (Carter 1989; Grayson 1993). The first is the informal team. This is the
cheapest and most flexible approach in which the Science Park partners divide the
various management tasks between themselves but there is no full-time, on-site
management presence. The second is the single on-site manager. This has the
advantage of allowing the build-up of expertise relating to the specific needs of
Science Park firms. Background and experience of the manager are important,
especially the balance of industrially related scientific skills, promotional skills
and skills in accessing finance. The third is the on-site management company.
Here, a formally integrated management structure, including representatives from
all investors in the Science Park, may provide a more secure basis for its long-term
development.
UKSP A has made a distinction between "managed" and "non-managed" Science
Parks in the United Kingdom. In 1992, UKSPA suggested that a "managed"
Science Park had a generally full time manager on site whose principal task was to
manage the Park (Westhead and Storey 1994). Science Park managers (i.e., gate-
keepers) can legitimise the activities of entrepreneurs and can increase and
diversify an entrepreneur's commercial, professional and social networks. In this
study, the role played by Science Park managers is explored. The following
Science Parks were regarded by UKSPA, in 1992, as being "non-managed":
Aberdeen, Aberystwyth, Cambridge, Highfields (Nottingham), Listerhills
(Bradford) and Loughborough (Westhead and Storey 1994, p. 170).
252 D. Siegel, P. Westhead and M. Wright

3. Research Questions

Science Parks have various objectives. These ohjectives include: the promotion of
HEI/industry linkages and the transfer of technology from HEIs to Science Park
firms; the promotion of the formation of new technology-based firms; the
encouragement of spin-off firms started by academics; the encouragement of the
growth of existing technology-based firms; attracting firms involved in leading-
edge technologies; creating synergy between firms; improving the performance of
the local economy; improving the image of the location, particularly for areas of
industrial decline; creating new jobs directly as well as indirectly; and enhancing
the competitiveness of new as well as existing firms in the region.
Developers of Science Parks assume that the owner of an independent
technology-based firm faces barriers to commercialising their knowledge into
profitable businesses due to their liabilities of small size and youthfulness. They
have assumed that these issues can be addressed by providing owners of NTBFs
and established firms with high specification and flexible premises located in
landscaped environments (although some tenants have to pay a rent premium);
shared resources and business services; links with local universities or HEIs; and
opportunities for co-operation with other tenants. Further, a Science Park manager
(i.e., a gate-keeper) can be a mechanism reducing uncertainty (and fixed costs) for
owner-managers of NTBFs. A gate-keeper can enhance the reputation of new and
small firms (i.e., those with limited social and business networks) and enable some
owner-managers of firms to gain access to an uninterrupted supply of crucial
resources from a variety of organisations including skills from HErs and finance
from banks, venture capitalists and development and enterprise agencies. Science
Park management can proactively stimulate transfer and promote links between
local HErs and tenant firms through informal contacts and visits from an HEI's
Industrial Liaison Officer, and thus enable firms to tap into important technical
information from HEIs (Westhead and Batstone 1998, 1999). Proximity ~o similar
firms and organisations with a technology focus (i.e., HEIs) can, in addition,
provide opportunities for business development through informal and formal
linkages. Network theory (Johannisson et al. 1994) suggests that entrepreneurs
(and firms) with dense and varied social and business networks of contacts can
gain information and resources to overcome business blind spots. Universities and
HEIs can act as a vehicle for the sponsorship of firms through the use of grants and
contracts and industry sponsored research. Moreover, firms utilising the resources
of HEIs (i.e., testing equipment, laboratories, the technical expertise of academics
and students, etc.) can control their own fixed costs.
The above arguments suggest that the performance of independent firms will be
enhanced if their owner-managers select a Science Park location. To guide policy,
the following research questions are explored in this review:
Rl. Were independent technology-based firms located on Science Parks in 1986
more likely to survive over the 1986 to 1992 period than off-Park firms?
 Support for Technology-Based Firms: The Role Played by Property-Based Science Parks 253

R2. Were independent technology-based firms more likely to survive than firms
engaged in more "conventional industries"?
R3. Were independent technology-based firms, reporting a link with a local HEI
in 1986 more likely to survive over the 1986 to 1992 period?
R4. Were independent technology-based firms located on "managed" Science
Parks in 1986 more likely to survive over the 1986 to 1992 period than firms
located on "non-managed" Science Parks?
R5. Did independent Science Park firms report higher levels of absolute
employment growth over the 1986 to 1992 period than off-Park firms?
R6. Was absolute employment growth over the 1986 to 1992 period
concentrated in a small number of independent technology-based firms?
R7. Did independent technology-based firms report higher levels of absolute
employment growth than firms engaged in "conventional" activities?

4. Research Methodology

Considerable debate remains over the appropriate research methodology to

measure the effectiveness of property-based initiatives. The above discussion has
highlighted the supply-side rationale leading to the development of Science Parks
in the United Kingdom. A demand-side perspective (i.e., a bottom up rather than
top down) was used in this study to assess the benefits of a Science Park location.
Researchers assessing the outcomes associated with programmes and initiatives to
assist NTBF formation and development rely on asking participants (e.g., owner-
managers of firms) their views. This procedure has several limitations:
• there is a clear risk that respondents may be unrepresentative of the
population of businesses located on Science Parks;
• respondents may be tempted to give the answers they think the questioner
expects to hear;
• the impact of a property-based Science Park initiative can only be judged by
comparing the performance of Science Park firms with comparable firms that
have not selected a Science Park environment;
• the effectiveness of property-based Science Park initiatives could be
exaggerated (or undervalued) by failing to take account of the demographic
characteristics of firms that have selected a Science Park environment; and
• whilst reported opinions of owner-managers of Science Park firms are of
interest, there is a need to collect "hard" indicators of actual subsequent firm
behaviour and performance.
To overcome the first limitation, the scale and nature of firms located on Science
Parks at specific points in time need to be ascertained. A longitudinal comparative
static methodology was used to assess the benefits associated with a Science Park
location. To gauge the benefits associated with a Science Park location in Great
Britain, the owner-managers of firms located on Parks were contacted. In 1986,
254 D. Siegel. P. Westhead and M. Wright

data was gathered from Science Park and off-Park firms (i.e., firms not located on
a Science Park) in Great Britain by Monck et al. (1988) I. A "follow-on"
questionnaire survey of Science Park and off-Park firms previously interviewed by
Monck et aI., was conducted in 1992/93 by Westhead and Storey (1994).
It is difficult to generalise results from a single Science Park environment. To
overcome this problem, interviews were conducted, in 1986, with a representative
sample of owner-managers of Science Park firms located in diverse range of
Science Park environments. Due to time and resource constraints, interviews could
not be conducted with the owner-managers of all firms located on Science Parks.
A random sample of Science Park firms was drawn, in 1986, by Monck et aI., from
the population of all firms located on Science Parks in Great Britain. Also,
structured questionnaires (i.e., developed with the advice from Steering
Committees) were used in 1986 and 1992 to gather information from owner-
managers of Science Park firms. Face-to-face interviews with the owner-managers
of Science Park firms were conducted (in 1986 and 1992) to ensure high response
rates to the surveys. Face-to-face interviews ensured that the owners of the
businesses responded and all appropriate information was collected.
To minimise the impact of the second limitation, a variety of objective as well as
more subjective outcomes associated with a Science Park environment were
monitored. The remaining limitations were overcome by collecting data (in 1986
and 1992) from a matched control group of owners-managers whose firms had
never been located on a property-based Science Park initiative. Using trade
directories and the Yellow Pages business directory, Monck et aI., identified a
random sample of technology-based firms not located on a Science Park in 1986.
Samples of Science Park and off-Park firms were matched, in 1986, with regard to
factors that influence business performance (i.e., business age, main industrial
activity, ownership status and standard region location; Storey 1994). To
overcome the final limitation, the Monck et aI., study conducted in 1986 and the
"follow-on" sample survey conducted in 1992/93 by Westhead and Storey (1994)
gathered employment data from Science Park and off-Park firms. The Monck et
aI., (1988) surveys as well as the Westhead and Storey (1994) "follow-on" surveys
are summarised below.
Monck et al. (1988) conducted a questionnaire survey of independent and
subsidiary organisations located on Science Parks in Great Britain. In total, 284
direct face-to-face interviews were conducted in 1986 by Monck et aI., with the
owner-managers of 183 Science Park firms (row 1 in Table 1).

IThe 1986 survey of Science Park firms conducted by Monck et al. (1988, p.64) and the
1992 survey conducted by Westhead and Storey (1994) adopted the following UKSPA
definition of a Science Park as a property-based initiative which included the following
features: has formal and operational links with a University. other Higher Education
Institution or Research Centre; is designed to encourage the formation and growth of
knowledge based businesses and other organisations normally resident on site; and has a
management function which is actively engaged in the transfer of technology and business
skills to the organisations on site.
Table 1. Survival and closure of surveyed science park organisations over the 1986 to 1992 Period (a).

Science Park sample (b) Off-Park sample C/)

r::
Total Ind Sub Total Ind Sub 't:l
't:l
g
~
r;l
l. Number of surveyed organisations in 1986 183 135 48 101 92 9 n
:r
::s
0
Sample of 35 organisations selected for interview 0-
~
in 1990 during the pilot follow-on survey but 6:l
excluded from the 1992/93 follow-on survey en
..,'"'
0-
::n
...
2. Survived - either original or new address 24 17 7 3
by the end of 1992 ~

3. Confirmed closure by the end of 1992 11 6 5 ;l

;:tl
'"
0

'"C
4. Survived - interview completed 59 49 10 50 44 6 0;-
"
'<
CT>
during 1992/93 0-
5. Survived - either original or new 31 26 5 19 18 0"
'<
address, 1992 :::\'
0
6. Confirmed closure by the end of 1992 46 25 21 21 19 2 't:l
CT>
7. Organization has no telephone listing / 12 12 0 11 11 0 ~
not recorded in telephone .,6:len
book or trade directory by the end of 1992 CT>
0-
C/)
n
Ci·
::s
n
8. Total sample closure rate (including 43 30 CT>

no telephone listings) over the 1986 (32%) (33%) '"C

to 1992 period ~
en

to.>
lJ\
I.Il
 ~
Table 1. (continued) Q\

9. Number of organisations in the valid 75 62

1992/93 follow on sample (excluding ~
en
those 23 independent Science Park iii·
organisations re-interviewed in 1990 ~
and the organisations that had closed :-c
over the 1986 to 1992 period) ::;:
!;l
10. Valid response rate to interview ;.
survey, 1992/93 65% 71% &
"
Q.
=
Notes: (a) The tracking of Science Park organisations was successfully achieved because information was collected
~
from Science Park managers surrounding organisation name changes and 1 or organisation relocations.
(b) Nine surveyed firms located on a Science Park in 1986 have subsequently survived and moved to an ::;:
:J.
off-Park location. og.
(In d) Independent organisation in 1986.
(Sub) Subsidiary organisation in 1986.
Source: Westhead, 1997, Table I, p.49.
 Support for Technology-Based Firms: The Role Played by Property-Based Science Parks 257

This sample constituted 53% of all tenants on Science Parks in Great Britain at
that time. The coverage, however, varied from one Science Park to another (for
example, only 35% of firms located on the Cambridge Science Park were
interviewed) and no firms were surveyed at Aberystwyth, Heriot-Watt, Kent and
St. Andrews. Monck et al. (1988) acknowledged that the 1986 survey contained an
over-representation of "new wave" Science Parks but argued "". that the firms in
this survey do provide an adequate sample of Britain's new high technology
industries, providing adequate geographical, technological, sectorial and
ownership coverage" (pp. 110-111).
In 1990, Storey and Strange (1992a) conducted a follow-on pilot study of 35
Science Park organisations (rows 2 and 3 in Table 1). During 1992/93, empirical
evidence from a second, and much more extensive, longitudinal "follow-on"
survey of independent and subsidiary organisations located on Science Parks in
1986 was collected (Westhead and Storey 1994). It was decided not to re-survey
those organisations contacted in 1990.
Table 1 shows that out of the 135 independent Science Park organisations
interviewed in 1986, 92 (68%) remained in business in 1992 (rows 2, 4 and sl
Rows 3 and 6 in Table 1 show 31 independent Science Park organisations had
closed. A further 12 independent Science Park organisations could not be traced
(row 7), and were regarded as closures. Hence, the total number of independent
Science Park organisation closures was 43 (32%; row 8).
Row 9 in Table 1 shows there were 75 independent surviving Science Park
organisations in the valid 1992/93 "follow-on" sample (excluding the 17 surviving
independent organisations re-interviewed in 1990 and 43 independent organisation
closures). Structured questionnaire interviews were collected from surviving
independent organisations (in 1986) during late 1992 and early 1993. Row 4 in
Table 1 indicates 49 "follow-on" interviews were conducted with independent
Science Park organisations (6S% valid response rate; row 10)3. These interviews
explored whether a Science Park environment provided technology-based firms
with a variety of benefits. The management function on site and the role-played by
the Science Park manager/director was also explored.

2The following widely used definition of organisation closure was used: "An independent
business is regarded as a closure if, in 1992, it is no longer identifiable as a trading
business. An independent business which moves locations but continues as a trading
business is not regarded as a closure. If the business is a subsidiary or a branch plant then it
is regarded as having ceased if it no longer trades at its previous location" (Westhead and
Storey 1994, p. 25).
3Responses to the "follow-on" survey were examined for non-response bias (Westhead and
Storey 1994, pp. 33-39). No marked differences between the 1986 and 1992 "follow-on"
survey returns were recorded with regard to ownership characteristics, sectorial or
geographical coverage or the age of the surveyed organisations.
258 D. Siegel, P. Westhead and M. Wright

5. Results from the Longitudinal Comparative Static

Study

Rl. Were independent technology-based firms located on Science Parks

in 1986 more likely to survive over the 1986 to 1992 period than off-Park
firms?
Westhead and Storey (1994) conducted a survival analysis of the independent and
subsidiary organisations located on and off Science Parks in Great Britain
interviewed by Monck et al. (1988). They monitored business survival over the
six-year period between the end of 1986 and 1992. In 1992/93, Westhead and
Storey attempted to contact all the 183 organisations interviewed in 1986 (Table
1). Using a widely used definition of business closure (Garnsey and Cannon-
Brookes 1993), they noted that 43 independent Science Park organisations had
closed. A further 30 independent off-Park organisation closures were identified.
No statistically significant difference in closure rates was detected between the
Science Park and off-Park samples (32% compared with 33%). A chi-square test
confirmed that this difference was not significant at the 0.05 level. Westhead et al.
(1995) concluded that sponsored Science Park environments had not significantly
improved the probability of business survival. It, however, must be appreciated
that several off-Park firms were located in sponsored Government designated
"assisted areas" for regional development assistance.
R2. Were independent technology-based firms more likeLy to survive than
firms engaged in more "conventional industries"?
It was difficult to compare the closure rates reported by independent Science
Park and off-Park firms over the 1986 to 1992 period with the closure rates
recorded by independent firms engaged in more "conventional industries". From
the outset, it needs to be appreciated that any comparison would be imperfect. It
was difficult to obtain any official statistics surrounding the survival of
independent businesses located in Great Britain. Data was, however, gathered from
a database that focused upon the number of businesses deregistered for Value-
Added-Tax (VAT) in the United Kingdom. Between 1980 and 1990, about 11 % of
the stock of firms de-registered each year (Daly 1991). Based on an annual 11%
deregistration rate over the six year period (i.e., 1986 to 1992), it was "expected"
that 115 total sample independent firms (i.e., the independent firms in the Science
Park and off-Park samples combined in 1986) would have closed. The "observed"
number of businesses closing in the total sample of independent firms was 42 firms
fewer than "expected" (73 "observed" closures compared with 115 "expected"
closures). This suggests that, subject to provisos, the closure rate of independent
essentially technology-based firm was lower that that recorded by United Kingdom
businesses in general.
R3. Were independent technology-based firms, reporting a link with a
local HEl in 1986 more likely to survive over the 1986 to 1992 period?
 Support for Technology-Based Firms: The Role Played by Property-Based Science Parks 259

In 1986, Monck et al. (1988) made a distinction between independent firms

reporting at least one link with a local HEI and those reporting no links. With
reference to the combined sample of independent Science Park and off-Park firms,
Westhead et al. (1995) charted the survival of firms with and without HEI links in
1986 to the end of 1992. The combined on and off-Park "follow-on sample",
however, cannot be regarded as a representative sample of all technology-based
firms in the United Kingdom in 1992/93. They found that a larger proportion of
firms (131 firms: 72%) that had reported a HEI link in 1986 were survivors in
1992 than firms that had not reported a link (19 firms: 53%). A chi-square test
confirmed that this difference was significant at the 0.05 level. Westhead et aI.,
inferred that links with a HEI had enabled some firms to overcome some of the
liabilities of "small size" and "newness,,4.
R4. Were independent technology-based firms located on "managed"
Science Parks in 1986 more likely to survive over the 1986 to 1992 period
than firms located on "non-managed" Science Parks?
Westhead and Batstone (1999) compared the closure rates between independent
firms (in 1986) located on "non-managed" (a stock of 41 firms in 1986) and
"managed" (a stock of 85 firms in 1986) Science Parks 5 . Over the 1986 to 1992
period, 10 firms (out of 41 firms) on "non-managed" Science Parks had closed
compared with 28 firms (out of 85 firms) located on "managed" Parks. It is
interesting to note that the closure rate was lower on the "non-managed" rather
than the "managed" Science Parks (24% compared with 33%). However, a chi-
square test confirmed that this difference was not significant at the 0.05 level.
Studies have found that older and larger firms are more likely to survive (Storey
1994). The lower closure rate on "non-managed" Science Parks may, therefore, be
simply due to the fact that more firms on "non-managed" rather than "managed"
Parks were older and larger in size. Further, the demographic differences between
firms on the two types of Parks may be due to the fact that many "non-managed"

4Westhead et al. (1995) used multivariate logistic regression analysis to detect the
combination of factors associated with the survival of technology-based firms in the Monck
et al. (1988) samples over the 1986 to 1992 period. A Science Park location, in 1986, was
not found to be a significant determinant of business survival. They found that younger
businesses and those competing against small firms were significantly less likely to survive.
However, "team start" businesses which were owned by more than one shareholder and
those that had a founder with a bachelor's degree or more were significantly more likely to
survive.
SIn 1986, 135 independent Science Park firms were interviewed on Science Parks defined
by UKSPA. The 9 independent firms surveyed, in 1986, located on the Bolton, East Anglia
and Springfield House (Leeds) Science Parks were regarded by UKSPA as Science Park
firms. However, in 1992, these locations did not conform to UKSP As Science Park
definition. Firms surveyed in Bolton, East Anglia and Leeds were no longer regarded as
Science Park firms. Westhead and Batstone (1999) excluded these firms from their study
that compared the closure rates between independent firms located on "managed" and
"non-managed" Science Parks.
260 D. Siegel. P. Wcsthead and M. Wright

Science Park managers have exhibited a reluctance to attract and incubate new and
small weak firms. The closure rate difference may, in part, be due to deticiencies
in the management of some "managed" Science Parks. Moreover, the difference
may be due to "managed" (i.e., those generally located in depressed and peripheral
regions) Park managers being less selective (and more prepared to take risks)
surrounding their choice of tenants. This, in itself, may stem from a desire to let
the highest appropriate proportion of property in order to maximise rental income
flow. Additional multivariate statistical analysis controlling for demographic
sample differences between firms in the two groups is, therefore, required to
explain this aggregate closure rate difference.
R5. Did independent Science Park firms report higher levels of absolute
employment growth over the 1986 to 1992 period than off-Park firms ?
Flynn (1993) warned that sponsored environments (i.e., Science Parks) may
encourage negative selection (i.e., non-survival attributes being transferred to the
next generation of firms) and the survival of competitively weak organisations.
The direct employment contributions reported by surviving independent Science
Park and off-Park firms in the "follow-on" samples were compared by Westhead
and Cowling (1995). Employment data for independent firms in 1986 was
available for two points in time for 46 surviving Science Park firms (out of 49
firms) and a further 31 surviving off-Park firms (out of 44 firms). The mean
employment size, in 1986, was 11.3 employees for Science Park firms compared
with 21.4 employees for off-Park firms. This difference between the two groups of
firms was not statistically significant at the 0.05 level. By 1992/93, the Science
Park firms had grown to employ on average 26.8 people whilst the mean
employment size for off-Park firms had grown to 37.8 employees. Again, this
difference was not statistically significant. The mean absolute employment growth
reported by Science Park and off-Park firms over the 1986 to 1992 period was
virtually identical (means of 15.5 employees and 16.4 employees). No statistically
significant difference was reported Science Park and off-Park firms. Westhead and
Cowling (1995) concluded that both types of firms had made contributions to job
generation. Moreover, there was no real evidence that the Science Park sample
included a large proportion of weak organisations that had been propped up by a
sponsored environment6 . Nevertheless, additional multivariate statistical analysis

6Westhead and Cowling (1995) used multiple correlation and regression analysis to detect
the combination of factors associated with log transformed employment change over the
1986 to 1992 period reported by surviving firms in the Monck et aI. (1988) Science Park
and off-Park samples. A Science Park location, in 1986, was not found to be a significant
factor explaining subsequent employment change. With respect to factors influencing high
levels of employment growth, a high firm size (in 1986) was found to act positively on
employment growth, as was a graduate level education for the key founder. On the finance
side, firms that had access to and used a multiplicity of sources of start-up finance tended to
report superior levels of employment growth.
 Support for Technology-Based Firms: The Role Played by Property-Based Science Parks 261

controlling for sample selection bias is still urgently required to confirm this
suggestion.
R6. Was absolute employment growth over the 1986 to 1992 period
concentrated in a small number of independent technology-based firms?
New and small firms are not a homogeneous group and only a small proportion
of NTBFs have the enthusiasm, ability, resources or inclination to grow (Oakey
1991). In aggregate, new and small firms make a significant contribution to the
total employment pool but this aggregate data masks a highly skewed distribution.
In short, whilst most firms employ a few people, only a few firms provide
significant employment. This assertion was conformed by evidence from the
"follow-on" samples of Science Park and off-Park firms.
Two firms in the Science Park sample reported no change in employment size
over the six-year period between 1986 and 1992. A further 9 Science Park firms
had reduced their employment size in total by 40 employees. The 35 employment
growing Science Park firms had generated an additional 753 jobs. Interestingly,
the 5 fastest employment growing Science Park firms (11 % of this sample)
accounted for 57% of these gross new jobs (i.e., 428 jobs). The skewed
distribution of employment was even more apparent in the off-Park sample. Four
off-Park firms reported no change in their employment size but a further eight
firms indicated that they had reduced their employment size in total by 115
employees. At the other extreme, the 19 employment-growing firms had generated
an additional 623 gross new jobs but the 5 fastest employment growers (i.e., 16%
of this sample) had generated 84% of these gross new jobs (i.e., 521 jobs).
Presented evidence suggests that few technology-based firms are significant
direct employment creators. Not surprisingly, a case for targeting support toward
firms (and entrepreneurs) with growth potential has been proposed (Westhead
1995). Considerable debate has, however, been generated surrounding the benefits
of targeting support and "picking winners" (Storey 1994).
R7. Did independent technology-based firms report higher levels of
absolute employment growth than firms engaged in "conventional"
activities?
Responses from the 46 Science Park firms and the 31 off-Park firms in the
"follow-on" sample were combined by Westhead and Cowling (1995) to generate
a "combined sample" containing 77 technology-based firms. They detected that
firms in the "combined sample" were larger and had recorded more absolute
employment growth than monitored firms engaged in technology-based (Oakey et
al. 1988) as well as "conventional" activities (Jones 1991; Storey and Strange
1992b).
262 D. Siegel. P. Wcsthead and M. Wright

6. Conclusions and Implications

At first sight, evidence presented in this review appears to be disappointing. This,

in part, is due to the unrealistic expectations surrounding the benefits associated
with property-based initiatives. Very few statistically significant performance
differences were detected between Science Park and off-Park firms with regard to
several indicators (i.e., business closure rates and absolute employment change)7.
It was, however, noted that Science Park and off-Park firms reporting at least one
link with a local HEI in 1986 were significantly more likely to survive over the
1986 to 1992 period than their counterparts that had failed to develop links with a
local HEI in 1986. Presented aggregate empirical evidence also confirmed that
technology-based firms were more likely to survive and report superior levels of
direct absolute employment growth than firms engaged in more "conventional"
industries. Policy-makers and practitioners concerned with encouraging wealth
creation and job generation need to appreciate that relatively few owners of
independent technology-based firms have the inclination or ability to grow the
employment size of their ventures. Whilst adding to the employment pool,
presented evidence suggests that relatively few technology-based firms are
significant direct employment generators in the short-term. Supporting the case for
the targeting of public policy support, a skewed pattern of absolute employment
growth, particularly by off-Park firms, was detected. As found elsewhere,
relatively few firms had generated the vast majority of absolute employment
growth. This evidence again warns that assessments of public policy interventions
should not solely focus upon business survival rates. Additional multivariate
statistical research is required that makes a distinction between low and high
performing surviving firms in terms of a broad range of "hard" (i.e., employment
growth, sales growth, profitability, ability to export, R&D inputs and outputs, etc.)
and "soft" (i.e., entrepreneur satisfaction, contribution to a local community,
number of links with a local HEI, etc.) performance indicators.
Several Science Parks are encouraging NTBF in depressed and turbulent
environments that do not have a culture of technology-based firm activity. Some
owner-managers located in these environments have selected "managed" Science
Parks to gain access to critical resources as well as to increase the legitimacy of
their ventures. This study has highlighted that different types of Science Parks
exist. Although the difference was not statistically significant, it was noted that
business closure rates on the "non-managed" Science Parks were lower than those
reported on "managed" Parks. It can be inferred here that Science Parks placing a
greater emphasis on increasing the supply of entrepreneurs who own technology-
based firms in depressed environments without a strong culture of technology-
based enterprise will be associated with high levels of business turbulence. In

7No statistically significant differences were detected between independent Science Park
and off-Park firms with regard to "inputs" into R&D and R&D "outputs" (Westhead 1997).
However, Science Park firms in the follow-on sample had reported in 1992/93 significantly
more links with local HErs than off-Park firms (Westhead and Storey 1995).
 Support for Technology-Based Firms: The Role Played by Property-Based Science Parks 263

<.:Ontrast, owners of "non-managed" Science Parks who place a greater emphasis on

developing rather than incubating new firms may be inevitably associated with
lower business closure rates. We believe it is unwise to assess the benefits
associated with "managed" Science Parks with regard to a single performance
indicator. A more balanced perspective of the benefits associated with the
heterogeneous Science Park movement needs to be appreciated.
One of the implications of these findings is that more effective ways of linking
tenant firms to the facilities and resources provided by a local HEI need to be
given greater attention by Science Park managers. For example, HEI industrial
liaison officers and Science Park managers need to become more proactive in
setting up systems which continually encourage the development of more formal
technical links between HEIs and industry located on and off Science Parks
(Westhead and Storey 1995). Areas that need to be considered by Park managers
include the need to enhance their technology and related market information
gathering as well as ensuring the dissemination of this information to their tenants.
Additional research evidence is needed to provide policy-makers (House of
Lords Select Committee on Science and Technology, 1997) with more accurate
assessments of the direct and indirect benefits as well as costs of future policy
decisions towards encouraging the supply of academic entrepreneurs and
technology-based firms. Large and representative surveys of entrepreneurs who
have selected a Science Park location for their businesses in the United Kingdum
(and other countries) need to be collected. Responses from the owners of NTBFs
should be selected to ascertain the benefits associated with different types of
Science Park initiatives. Several "hard" (i.e., business closure rates, employment
size, absolute employment change "inputs" into R&D and "outputs" from R&D,
etc.) and "soft" (i.e., social and business networks, trust, co-operation and
satisfaction, etc.) outcomes need to be monitored. Future studies need to carefully
monitor the deadweight and displacement outcomes associated with different types
of Science Parks. Most notably, how many owners of Science Park firms would
have gone into business anyway without the provision of supportive Science Park
environments.
The effectiveness of intermediaries (Perry 1996), such as Science Park
managers, in the commercialisation process and the linking of Science Park firms
with HEIs, other tenants on Parks a well as other firms located off-Park needs to
be carefully monitored. To promote mutual learning and the sharing of expertise
between technology-based firms, additional research is warranted surrounding the
role-played by gate-keepers in maximising goodwill trust between tenants located
on Science Parks and off-Park firms. Superior gate-keeper practice needs to be
highlighted and disseminated to other practitioners.
264 D. Siegel, P. Westhead and M. Wright

References

Carter, N. (1989), Science Parks Development and Management, London, The Estates
Gazette Limited.
Castells, M., and Hall, P. (1994), Technopoles of the World: The Making of 2Ft Century
Industrial Complexes, London, Routledge.
Cooke, P., and Morgan, K. (1993), "The Network Paradigm: New Departures in Corporate
and Regional Development", Environment and Planning C: Society and Space, 11, pp.
543-564.
Daly, M. (199]), "VAT Registrations and Deregistrations in 1990", Employment Gazette,
99, pp. 579-588.
Department for Trade and Industry (DTI) (1998), Our Competitive Future: Building the
Knowledge Driven Economy, Competitiveness White Paper, London, HMSO, Cm 4176.
Felsenstein, D. (1994), "University-Related Science Parks - 'Seedbeds' or 'Enclaves' of
Innovation?", Technovation. 14, pp. 93-110.
Flynn, D. M. (1993), "A Critical Exploration of Sponsorship, Infrastructure, and New
Organizations", Small Business Economics, 5, pp. 129-156.
Garnsey, E. W., and Cannon-Brookes, A. (1993), "The "Cambridge Phenomenon"
Revisited: Aggregate Change Among Cambridge High-Technology Companies Since
1985", Entrepreneurship and Regional Development, 5, pp. 179-207.
Grayson, L. (1993), Science Parks: An Experiment in High Technology Transfer, London,
The British Library.
House of Lords Select Committee on Science and Technology (1997), The Innovation
Exploitation Barrier, London, House of Lords, Paper 62.
lohannisson, B., Alexanderson, 0., Nowicki, K., and Senneseth, K. (1994), "Beyond
Anarchy and Organization: Entrepreneurs in Contextual Networks", Entrepreneurship
and Regional Developmellt, 6, pp. 329-356.
lones, C. (1996), "The Theory of Property-Lcd Local Economic Development Policies",
Regional Studies, 30, pp. 797-801.
lones, M. (1991), Small Firms Study 1985, 1988 and 1991, Coventry, University of
Warwick MBA, Unpublished Dissertation.
Keane, 1., and Allison, 1. (1999), "The Intersection of the Learning Region and Local and
Regional Economic Development: Analysing the Role of Higher Education", Regional
Studies, 33, pp. 896-901.
Massey, D., Quintas, P., and Wield, D. (1992), High Tech Fantasies: Science Parks in
Society, Science and Space, London, Routledge.
Monck, C. S. P., Porter, R. B., Quintas, P., Storey, D. 1., and Wynarczyk, P. (1988).
Science Parks and the Growth of High Technology Firms, London, Croom Helm.
Oakey, R. (1991), "High Technology Small Firms: Their Potential for RJ.pid Industrial
Growth", International Small Business Journal, 9, pp. 30-42.
Oakey, R. P., and Mukhtar, S-M. (1999), "United Kingdom High-Technology Small Firms
in Theory and Practice: A Review of Recent Trends", International Small Business
Journal, 17, pp. 48-64.
Oakey, R., Rothwell, R. and Cooper, S. Y. (1988), The Management of Innovation ill High
Technology Small Firms, London, Pinter Publishers.
 Support for Technology-Based Firms: The Role Played by Property-Based Science Parks 265

Organisation for Economic Co-Operation and Economic Development (OECD) (1993),

Small and Medium-Sized Enterprises: Technology and Competitiveness, Paris,
Organisation for Economic Co-Operation and Development.
Perry, M. (1996), "Network Intermediaries and Their Effectiveness", International Small
Business Journal, 14, pp. 72-80.
Roberts, E. B. (1991), Entrepreneurs in High Technology, New York, Oxford University
Press.
Saxenian, A. (1994), Regional Advantage: Culture and Competition in Silicon Valley and
Route 128, Cambridge, MA, Harvard University Press.
Storey, D. 1. (1994), Understanding the Small Business Sector, London, Routledge.
Storey, D. 1., and Strange, A. (l992a), "Where Are They Now? Some Changes in Firms
Located on UK Science Parks in 1986", New Technology, Work and Employment, 7, pp.
15-28.
Storey, D. 1., and Strange, A. (1992b), "New Players in the "Enterprise Culture"?", in:
Small Enterprise Development: Policy and Practice in Action, Caley, K., Chell, E.,
Chittenden, F. and Mason, C. (eds), London, Paul Chapman Publishing, pp. 85-95.
Tether, B., and Storey, D. 1. (1998), "Smaller Firms and Europe's High Technology
Sectors: A Framework for Analysis and Some Statistical Evidence", Research Policy, 26,
pp.947-971.
United Kingdom Science Park Association (UKSPA) (1999), 15th Anniversary 1984-1999,
Birmingham, The United Kingdom Science Park Association.
Westhead, P. (1995), "Survival and Employment Growth Contrasts Between Types of
Owner-Managed High-Technology Firms", Entrepreneurship Theory and Practice, 20,
pp. 5-'27.
Westhead, P. (1997), "R & 0 "Inputs" and "Outputs" of Technology-Based Firms Located
On and Off Science Parks", R&D Management Journal, 27, pp. 45-62.
Westhead, P. and Batstone, S. (1998), "Independent Technology-Based Firms: The
Perceived Added Value of a Science Park Location", Urban Studies, 35, pp. 2197-2219.
Westhead, P. and Batstone, S. (1999), "Perceived Benefits of a Managed Science Park
Location for Independent Technology-Based Firms", Entrepreneurship and Regional
Development, 11, pp. 129-154.
Westhead, P. and Cowling, M. (1995), "Employment Change in Independent Owner-
Managed High-Technology Firms in Great Britain", Small Business Economics, 7, pp.
111-140.
Westhead, P. and Storey, D. 1. (1994), An Assessment of Firms Located On and OjJScience
Parks in the United Kingdom, London, HMSO.
Westhead, P. and Storey, D. 1. (1995), "Links Between Higher Education Institutions and
High Technology Firms", International Journal of Management Science, Omega, 23, pp.
345-360.
Westhead, P., Storey, D. 1., and Cowling, M. (1995), "An Exploratory Analysis of the
Factors Associated with the Survival of Independent High-Technology Firms in Great
Britain", in: Small Firms: Partnerships for Growth, Chittenden, F., Robertson, M. and
Marshall, 1. (eds), London, Paul Chapman, pp. 63-99.
High-Tech Industrial Clusters
in the Asian Countries: The Case of the
Electronics Industry in the HSinchu-Taipei Region

Kung Wang

Abstract. This article mainly discusses Taiwan innovation system (TIS), that is,
the operation system by enterprises, government, universities and public research
laboratories of science and technology (S&T) industry development, and also the
most representative characteristics that can explain TIS, that is, 1) the technology
development program (TDP) of the Ministry of Economic Affairs (MOEA) for
instance, to illustrate the industry innovation policy by Taiwanese government; 2)
the Industry Technology Research Institute (ITRI) for instance, to explain R&D,
and technology diffusion strategy of research institutes; 3) the Hsinchu
Science-based Industrial Park (HSIP) for example, to show the way of enterprises
S&T commercialisation; 4) bringing back overseas scholars and also the talents
cultivation, to introduce supply of human resources in the academic circle. Finally,
it introduces the practical achievements of TIS by the cluster effect formed for the
development of Taiwan semiconductor industry.

1. Introduction

Taiwan is an island country, with limited natural resources and small market scale.
After the agricultural society in 1940s, there was the light industry of the people's
livelihood in 1950s, and then the petrifaction heavy industry in the 1960s. At the
beginning of 1970s, both the government and domestic and foreign scholars and
experts thought that, in order to maintain economic development, Taiwan should
develop towards hi-tech industry. Looking back at the industry situation of Taiwan
in the 70s, however, it was universally family business, and capital market has not
taken shape yet, with weak equipments and capacities in terms of basic researches
by universities. In order to enable domestic industry circle to obtain foundation for
hi-tech development, government, enterprises, and experts and scholars from
universities and public research laboratories all worked together, trying to build an
innovation system for the development of hi-tech industry. Important S&T
organizations, S&T policies and measures, and research institutes are as shown in
Figure 1.
 N
o.
00

Important S&T Organizations and Important S&T Orgallizatiolls alld Importallt Research Illstitutes ?'
Conferences Conferellces
~
Go
"
19591National Long-Tenn Development iI
I
I
Science Committee I
I
!Rewarding Investment Regulations (1960)
I Chung Shan Institute of S&T (1965)
I
I
I
I ITRI (1973)
I
19761Applied Technology R&D I
I
I
I

19791Advisors Group, Executive Yuan !S&T Development Program (1979) III (1979)
I
: Science Park (1980)
I
I
19821The Second National S&T Conference I
I
I
I
:The Program of Strengthening Cultivation
I
:and Recruitment of S&T Talents (1983) Biotechnology R&D Center (1984)
I
19861The Third National S&T Conference !National S&T Ten-Year Long-Tenn
!Program (1986)
1991 IThe Fourth National S&T Conference !Leading New Products (1991)
I Bicycle R&D Center (1993)
I

!Ten Top Emerging Industries (1994)

!Smart Industry Park (1995)
I
19961The Fifth National S&T Conference I
I
I
!National Promotion ProgramsTowards S&T
:(1 Q98)
I
I
2001 IThe Sixth National S&T Conference I
I
!

Fig. 1. Brief history of Taiwan S&T policy.

 High-Tech Industrial Clusters in the Asian Countries 269

Under its unique innovation system, after efforts for almosl 30 years, Taiwan
has outstanding performance concerning number of patents acquired in USA and
papers published on Science Citation Index (SCI) and (Engineering Index (EI),
comparing with other rising countries in Asia. Its overall development
achievements of S&T industry are quite good, as shown in Table l.
The main objective of this article is to introduce the innovation system with
Taiwan characteristics, with the main structure as follows: first, it discusses about
TIS, that is, the operation system by enterprises, government, universities and
public research laboratories in S&T industry development, and also the most
representative characteristics that can explain TIS, that is, 1) take TDP of the
MOEA for instance, to illustrates the industry innovation policy by Taiwanese
government; 2) Take ITRI for instance, to explain R&D, and technology diffusion
strategy of research institutes; 3) Take the HSIP for example, to show the way of
enterprises S&T commercialisation; 4) take the instance of bringing back overseas
scholars and also the talents cultivation, to introduce supply of human resources in
the academic circle. Finally, it introduces the practical achievements of TIS by the
cluster effect formed for the development of Taiwan semiconductor industry.

2. Documentation Discussions on National Innovation

System (NIS)

2.1. A Survey of Academic Thought: Definitions of, and Views on, a

National Innovation System

The concept of a national innovation system basically follows Schumpeter's (1939)

discussion on technology innovation. Among scholars who study this topic, some
of the more renowned academics are Freeman, Lundvall, and Nelson. They each
define a national innovation system from different research angles.
Freeman (1987) defines a national innovation system as " ... the network of
institutions in the public and private sectors whose activities and interactions
initiate, import, modify and diffuse new technologies". He discusses the interplay
between activities that lead to advances in technology and the organizations in the
public and private sector that comprise the national innovation system.
Lundvall (1922, p. 2) says of the national innovation system that " ... a system
of innovation is constituted by the elements and relationships which interact in the
production, diffusion and use of new, and economically useful, knowledge ... and
are either located within or rooted inside the borders of a nation state", and studies
the system from the perspective of the institutions that, taken together, make up
the national innovation system, as well as from the microcosmic point of view of
the individual.
Nelson (1993) calls the national innovation system a " ... a set of institutions
whose interactions determine the innovative performance ... of national firms".
 N
.....
Table 1. Comparison of R&D achievements b~ the four small dragons of Asia from 1993 to 1999. 0

Year 1993 1994 1995 1996 1997 1998 1999

R.O.C. Name of patents approved in USA 1,510 1,814 2,087 2,419
;-;
2,597 3,805 4,526
~
.,
Ranking 7 7 7 7 7 5 4 ::s
flC>
Number of reports published on SCI 4,793 5,852 6,682 7,500 7,758 8,601 8,931
Ranking 22 20 19 18 19 19 19
Number of reports published on EI 2,687 3,543 3,560 4,220 4,839 4,026 4,376
Ranking 11 12 11 11 11 9 11
South Name of patents approved in USA 830 1,008 1,240 1,567 1,965 3,537 3,679
Korea Ranking 11 10 9 8 8 8 7
Number of reports published on SCI 3,014 4,041 5,384 6,422 7,807 9,503 10,992
Ranking 29 24 23 21 18 16 16
Number of reports published on EI 1,342 2,135 2,528 3,479 4,911 3,645 4,777
Ranking 18 15 13 13 10 11 10
Singapore Name of patents approved in USA 44 59 61 97 100 136 152
Ranking 30 24 24 24 25 24 22
Number of reports published on SCI
Ranking
Number of reports published on EI 770 965 1,190 1,152 1,603 1,213 1,854
Ranking 22 23 21 25 24 23 19
H.K. Name of patents approved ill USA 182 220 248 247 261
Ranking 20 19 19 19 19
Number of reports published on SCI 1,539 1,779 2,915 3,090 3,703 --
Ranking 38 38 35 35 28
Number of reports published on EI 448 789 978 1,310 1,719 1,483 1,688
Ranking 31 28 24 23 21 20 21
Source: NSC, 2000b.
 High-Tech Industrial Clusters in the Asian Countries 271

Basically, the focus of his research is to compare the innovation systems

adopted by various countries.

2.2. The NIS and the Knowledge Economy

In today's era where business is moving towards the concept of a knowledge

economy, an understanding of how the national innovation system operates, and a
realization that the successful operation of the knowledge economy depends on
the smooth flow of knowledge among enterprises, universities and research
institutions, enables policy-makers to develop policies and measures that promote
successful innovation.
With this in mind, and following these academics' research, OEeD (1997)
attempted to apply to the national innovation system the indices used to evaluate
the development of the knowledge economy. The aim was to discuss and compare
interactions among the various players in the national innovation system, in order
to gauge how effectively those interactions were at facilitating the flow of
knowledge in the greater technology sector. Four types of interactions between
different constituents of the innovation system were compared:
1. Interactions among Enterprises.
2. Interactions among Enterprises, Universities and Public Research
Laboratories.
3. Diffusion of Knowledge and Technology to Firms.
4. Movement of Personnel.

2.3. Framework for a NIS Used in this Article

In this section, all the components of Taiwan's policy towards the NIS will first be
examined in broad outline. Then the focus shifts to a microcosmic analysis of the
way in which the individual institutions making up the innovation system interact
with one another. Interactions between manufacturers in the industry, co-operation
among enterprises, universities and public research laboratories, interactions that
lead to technology diffusion, movements in the labour force, international
knowledge exchanges, and other kinds of interactions will be examined to
facilitate discussion on these five aspects of communication within the NIS
(Figure 2).
272 K. Wang

Policy recomrrerrJa~fn
Ovcrse!lS Government
technology :md Professiornl ..., ...i
human resource; pvrh"nup i i

Policy suggestion
J i
~olicy suggestio

;' :r: ,! Study coorse plan

, .."
5·
5 ""co
n

""
i j
" i "
,,~

R&D plan R&D plan n

~
"8
0 ~o.
0-
~

Dl 1 i5-~
! ACldffi1ic and :=:q.
()Q
'< 0
"ri 8-« "i':l
5·
q Dl
i ""eareb institut, ~ -. j (JQ

"",05 .I {Jniversitics "8

J '""
0
f:;>operation
~---. Research
!':'.
o· imtitutcs Professiomls o·
? ?

~O~Mplru D["
~

;J Training Progr.

"n·
O,t n
00
IndU'Stry and '??..t-Q, IndU'Stry and
"""0;.
"
;:;. Research Institut • "'<>"b~ %~... a Universities
g O)operation ~<:r • q0~ ~ Cooperation
" ""
J Enterprises .. I
1 innovatifNl
Inlustry WVlSlOO lU"ll"K, HUllS I)' aggncgation
effect

Fig. 2. Taiwan innovation system illustration.

Source: This article.

3. The Framework of Taiwan S& T Innovation

Taiwan technology development adopts the principle of overall planning and

execution based on different labour division. For the industry technology
development, the MOEA takes charge of the technology development work of
applied researches, In addition to research units directly under its jurisdiction, the
R&D departments in state-run enterprises, and research institutions appointed by
Technology Developed Program (TDP), mainly engage in application and
development of industry technology, transferring then to enterprises circle to carry
out products technology development and commercialisation activities,
Technology of the enterprises circle can be originated from technology transfer or
research participation of research institutions and academic circle etc., as well as
 High-Tech Industrial Clusters in the Asian Countries 273

from technology introduction and strategic alliance with overseas companies or

research institutions. It is as shown in Figure 2.
Research institutions interact with the enterprises circle by the method of
technology transfer and cooperation with enterprises, and also assist in technology
development of enterprises by technology guidance, information supply, and
talents training etc. The special-topic research plans by NSC, promotes the
academic circle to assist enterprises to develop technology-by-technology
guidance and talents training etc. In the respect of development measures for
technology of applied research by the government, besides NSC's cooperation
program between enterprises and universities, research program of innovation
technology, and target-oriented research program, there are also the program of
introducing overseas technology, S&T development program and TDP by
enterprises mainly led by the MOEA. In terms of commercialisation and
application measures, there are the key components and parts by NSC and the
MOEA's leading new product program, enterprises cooperated research program,
and the R&D program of state-run enterprises assisting Small and Medium-Sized
Enterprises (SMEs) etc.
For the interaction among enterprises, government, universities and public
research laboratories, research institutions and academic circle transfer their R&D
results to enterprises by all means, and then relying on each research institution,
have interaction with overseas companies of enterprises or overseas research
institutions through technology introduction, cooperated research, overseas
investment and strategic alliance. The MOEA also implements subsidy funds,
cooperation funds, and investment offset etc. measures, to encourage enterprises
to engage in R&D. In conclusion, Taiwan industry technology development at
present mainly focuses on independent R&D, assisted by technology introduction,
and also stresses malignant interaction among enterprises, government,
universities and public research laboratories, with the final goal of promoting the
technology upgrade of the industrial circles.

4. TOP of the MOEA

In accordance with the "S&T Development Program" issued by the Executive

Yuan in 1979, TDP becomes an important policy tool helping the MOEA to drive
development of enterprises technology. TDP is used to research and develop key
components and parts or common key technology, and then transfers to industrial
circles for commercialisation and even quantity production investment, to reach
the goal of using enterprises technology to raise enterprises competitiveness. Thus,
the idea of TDP is to promote development of enterprises technology and stable
economic growth, and strengthen competitiveness of enterprises and national life
quality, through the R&D work of TDP.
The Technology Section of the MOE A is responsible for the promotion of TDP.
In order to help the upgrade of enterprises technology, technology that has the
common features of advantages in market, technology intensive, enterprises
274 K. Wang

connection and verifiability etc., and also high-benefit technology with the
capability to increase public welfare and public investment etc. are chosen as the
main R&D items, therefore, the promotion of TDP plays a decisive role in the
development of overall industry and economy.
The positioning of TDP is mainly to carry out the development of planned
technology based on requirement by the development of enterprises. Enterprises
development factors include capital, human resources, technology, market,
resources, environmental protection, infrastructure and related management
system etc., therefore, TDP selects technology needed for promoting enterprises
development as the prerequisite, adopts the development strategy of independent
technology development and bringing in overseas hi-tech industry, and engage in
product development of important components and parts, important examination
technology, decide technology standards and establish management system of
intellectual property, to further raise technology of enterprises.
In order to help forward industry upgrade, strengthen effects to industrial circles
by TDP, various forms and channels are used to implement the R&D outcomes of
TDP on enterprises. During the program execution period, enterprises can make
use of R&D outputs of research institutions by participation in advance,
cooperated research and authorization transfer etc. After the program is finished,
research institutions put into effect the R&D outcomes in the industrial circles by
technology transfer and patent authorization etc. Information exchange, seminar,
professionals diffusion and all kinds of industrial services can all help to establish
the exchange channels between research institutions and enterprises, and this kind
of channel also helps to diffuse R&D outcomes of TDP to enterprises without
taking a form, professionals diffusion is especially a method that applies
technology directly and practically in enterprises.
In order to fulfil the important mission of upgrading industry technology, the
Technology Section selects and adopts measures based on industry change
appropriately, furthermore, it has to have an insight into present and future
industry requirement from a foresighted and macroscopic point of vie"" to raise
proper measures during different stages (MOEA 2001). Looking back, in addition
to constantly appointing mainly ITRI, Development Centre for Biotechnology
(DCB), institute for Information Industry (III) etc. non-profit research institutions
to carry out technology development starting from 1979, and adopting all kinds of
channels like technology transfer, technology seminar, industrial services, and
derivative companies establishment etc. to implement R&D outcomes extensively
in the industrial circles, to really help S&T industries to become main force of
Taiwan industries today, it also uses proprietors as the direct object of measures to
promote enterprises to join in R&D in an all-round manner, to fulfil the measures
of upgrading enterprises technology. This can be traced back to 1994 when
promoting enterprises cooperation, proprietors participated in since the planning
stage of TDP, so that the R&D program by research institutions can be more
suitable to requirement by enterprises, and the participation by proprietors during
the research procedures can help to accelerate the commercialisation procedures of
technology as an answer to the shortening of technology life cycle. In order to
accelerate in helping the industrial circles to set up S&T type enterprises, using
 High-Tech Industrial Clusters in the Asian Countries 275

research institution capabilities, facilities, talents and knowledge etc. resources

accumulated by TDP, it provides "public laboratories" as the place for the
industrial circles to engage in R&D and business establishment. In order to
correspond to the gradual upgrade of industry technology and extensive
requirement, under the situation of appointment to juridical person research
institutions being insufficient for all these demands, the Technology Section tried
in 1997 to appoint the industrial circle to carry out TDP, namely the "enterprises
TDP". Following the development trends in the knowledge economy era,
strengthening the creation of knowledge is its important factor, so is the
development of human resources. Consequently, the first consideration is to
introduce abundant R&D and human resources in universities systematically into
the innovation of enterprises, the Technology Section tried the "Universities TDP"
of appointing universities to carry out TDP in 2000.

5. R&D and Technology Diffusion Strategy of Research

Institutions

In order to develop towards high added value and technology-intensive industry,

promote the establishment of emerging hi-tech industry, and improve the present
industry structure to reach sustainable development, government helps the
establishment of research institutions in phases and also takes the responsibility to
improve industry technology based on requirement by industry development. For
instance, ITRI was established in 1973, with a predominant role in innovation
technology and as the pilot who started up domestic industry technology; III was
established in 1979, to drive development of information software/hardware
industry in Taiwan. In addition, for the significance of biotechnology, government
also sets up DCB, with the expectation of improving our biotechnology. Since
industry structure of Taiwan is still with SMEs at the most, in order to assist
technology development of SMEs, seven non-profit research institutions were
helped to be set up in 1992 and 1993 successively, including printing, plastics,
shoemaking, bicycle, precision machinery, stone materials processing, and
pharmaceutical technology research centre.
Among them, ITRI is the largest non-profit industry technology application
research institute in Taiwan, and has engaged in R&D until now for over 20 years.
Starting from TDP execution appointed by the MOEA, it builds up its own R&D
ability, and then pushes enterprises to engage in technology development, further
drives our industry structure to advance towards the technology-intensive
development direction. There are around 6,000 employees in ITRI during these
recent years, with more than half owning master or Ph.D. degree. They cover
extremely extensive fields of technology specialty, and are distributed in seven
research laboratories and four research centres. Including Electronic Research &
Service Organization (ERSO), Computer & Communication Research
Laboratories (CCL), Opto-Electronic & System Laboratories (OES), Mechanical
Industry & Research Laboratories (MIRL), Union Chemical Laboratories (UeL),
276 K. Wang

Material Research Laboratories (MRL). Energy & Resources Laboratories (ERL),

and Centre for Measurement Standards (CMS), Centre for Aviation & Space
Technology (CAST), Biomedical Engineering Centre (BMEC). Centre for
Industrial Safety and Health Technology (CISH).
ITRI involves actively in the development of industry technology, to assist the
industrial circle to improve technology, further to establish oncoming S&T
industry and upgrade conventional industries. In order to reach the goal of
improving industry technology, the research and development management of
ITRI adjusts duly with the requirement by industries, for instance, it emphasized
R&D planning since early stages, selected technology items suitable to be
developed in Taiwan, and focused on bringing in technology. Following the ardent
demand by the industrial circles, it made great efforts in diffusing R&D
achievements, and also transferred experts appropriately to equally emphasize
R&D of emerging technology industry and putting it into effect. At present, it
drives the industrial circle to engage in R&D activities to actively create economic
benefits, and also cooperates with balanced industry development of government
to participate in newly risen fields. For the task of improving industry technology,
ITRI will still actively implement R&D outcomes on enterprises in the future,
besides bringing economic benefits, but also will open laboratories to accelerate
establishment of S&T innovation enterprises. On the other hand, it will also
engage in R&D of foresighted technology, to open up new technology and
foresighted fields for enterprises, and also keep an unfailing supply of motive
power of technology to create technology required by enterprises during the
opportune time. At the same time, it will accelerate internationalisation of R&D
and combine with the development of science-based industrial park (SIP).
The technology development strategy of ITRI is the R&D strategy oriented by
the target of establishing core technology, environment construction, R&D
innovation, and also improving R&D energy of enterprises. In the respect of
establishing core technology, it is to integrate each technological field and create
the maximum benefits; in the respect of environment construction, it is to build up
information network environment and promote exchange of specialists from
different fields; in the respect of R&D innovation, it is to strengthen foresighted
technology R&D, promote internationalisation and also recreation of innovation
culture, to enable ITRI to own an unfailing supply of motive power of technology;
in the respect of improving R&D energy of enterprises, it is to use active methods
of putting R&D outcomes into use by enterprises such as cooperation among
enterprises, technology transfer, R&D strategic alliance, and public laboratories
etc., to accelerate the improvement of R&D energy by enterprises.

6. The Technology Commercialisation Method in HSIP

Taiwan government adopts the policy of establishing Science Park to develop

hi-tech industry of Taiwan. Considering that the main conditions for developing
hi-tech industry are S&T professionals, the most advanced technology and fast
 High-Tech Industrial Clusters in the Asian Countries 277

acquirement and digestion of information, Hsinchu was chosen as the place to

establish the first SIP of our country, the reasons are that, there are Tsing Hua
University, Chiao Tung University and ITRI etc. academic and research
institutions in Hsinchu, they provide excellent human resources abundantly, who
can assist manufacturers to break through' technology bottleneck. There is also
convenient traffic condition in Hsinchu. The establishment of HSIP is an
important measure taken by the Taiwanese government for promoting
development of hi-tech industry. After development for 20 years, it is indeed the
representative for Taiwan hi-tech industry acknowledged universally by home and
abroad. There are six main industries formed already in HSIP, namely
(Science-based Industrial Park Administration 1998):
1. integrated circuits: this industry is the one with the fastest growth in HSIP
during the recent ten years, including integrated circuits design, manufacture,
materials, assembly and testing, manufacturing process equipments etc.
2. Computer and peripherals: computer and peripherals industry is the main
mdustry of HSIP, including computer system, storage equipments,
input/output equipments, network equipments, special software and related
mechanical and electrical components and parts etc.
3. Communication: communication industry has introduced the following four
categories of products in HSIP, electronics communication system
(including telephone set, modem, and exchange board), microwave system
and components, optical fiber system and components, and satellite
communication system.
4. Opto-electronics: this industry has brought in opto-electronics system
components (such as thin-film transistor flat display, color display tube,
CD-ROM drive, digital static picture camera), touch image sensor,
opto-electronics semiconductor, and photodiode etc.) and optical system
components (such as instruments and lens etc.).
5. Precision machinery: precision machinery industry has brought in
automation system in HSIP (e.g. Computer Numerical Control (CNC) tool
machinery, industry used controller, robot, water knife, vacuum machine,
factory information automation etc.) and automatic components (such as
casting, surface processing etc.).
6. Bioengineering: bioengineering technology has already brought in vaccine
reagent, medical appliances and seed vaccine etc. in HSIP.
Ever since it was established in 1980, HSIP has introduced 292 hi-tech
manufacturers until the end of 1999, altogether 82,778 employees and investment
at NT$6,183. After development for 20 years, HSIP has grown up to become a
place with significance for Taiwan hi-tech industry development.
Since the main objectives for establishing SIP in Taiwan is to promote
development of high technology and industry, and the government uses the
essential factors of hi-tech industry development as the object for carrying out its
policies, this section explains the ways of innovation by HSIP based on different
measures by the government to promote each essential factor to benefit
manufacturers or entrepreneurs who engage in development of hi-tech industry.
There are mainly technology sources, human resources, funds requirement,
278 K. Wang

preferential land tax and other levies, and management environment five items of
important factors, to be used to illustrate the way of bringing technology into
products by hi-tech manufacturers in HSIP.

6.1 Technological Resources

HSIP had a large external demand in early days, ITRI could be regarded as
important technological sources for HSIP, and manufacturers could also appoint
ITRI to develop product technology or solve technological problems. For
technological demand, government appointing applied research institutions such
as ITRI near HSIP to carry out R&D and also put R&D outcomes into use by the
industrial circles, has brought obvious results. With the growth of HSIP, the
technological sources have already included overseas or technology transfer and
cooperation, or from nearby Tsing Hua and Chiao Tung University. However,
manufacturers in HSIP presently are still searching for technology development
related information from ITRI, and also looking for new field, so its role has
changed from the predominant one to supporting functions like encouraging to
innovate, information, and assisting settlement of technological problems, only in
some innovation fields, it is still an important technological source for HSIP.
In order to encourage manufacturers to engage in R&D of innovation
technology, government provides grants for R&D of innovation technology to
manufacturers in HSIP. It has been 14 years since it was implemented in 1986,
and has become an important resource for helping HSIP to increase technology
competitiveness. The government has provided grants to 24 research programs in
1999, totalling NT$67 million, at 28.1 % of the total invested amount of NT$241
million. Up to 1999, altogether 478 programs received grants, at the amount of
NT$896 million. Furthermore, in order to encourage manufacturers to research
and develop key components & parts and products, the government has provided
key components & parts and products subsidy for R&D by manufacturer~ in HSIP.
From July 1992 to June 1999, there are 88 that have passed examination, and for
the total R&D funds of NT$6.33 billion, NT$2.472 billion was approved to
recei ve subsidy (Science-based Industrial Park Administration 1999).

6.2. Human Resources

For the development of high technology, talents are an important factor, and the
aggregation of high-level human resources is the most significant characteristics
of HSIP, and also an important resource promoting the gestation of hi-tech
industry. Concerning the supply of human resources, manufacturers can easily
recruit human resources needed, and can provide training for on-the-job personnel
and raise personnel quality, at the same time, it attracts talents, including bringing
in hi-tech entrepreneurs, that is why government chooses the place near
universities and research institutions to set up HSIP.
 High-Tech Industrial Clusters in the Asian Countries 279

6.2.1. Bring in High-Level Technology Entrepreneurs

Taiwan has started to engage in the development of hi-tech industry since 1980s.
Since local people were lack of R&D ability, overseas scholars were recruited,
especially those who have worked in Silicon Valley. They brought technology
ability, marketing information, equipments purchase information etc. to
Taiwanese, and based on their experience overseas, technology transfer and
cooperation were expanded and even branch companies were set up, and
Taiwanese also went to join in overseas technology seminar to master the
information about advanced technology. In order to create newly risen enterprises,
overseas scholars were recruited to bring in technology. The companies
established by Taiwanese scholars who went to study overseas, from its start-up to
healthy growth, all made significant contribution. There are 109 these kind of
companies in HSIP by the end of 1998, with nearly 3,056 people. The technology
and ideology brought back by these scholars have deeply rooted in the SIP and
developed, which shows the performance of this measure obviously.
In addition, many high-level technology talents from research institutions were
attracted to start enterprises in SIP, for example, the first non-governmental
integrated design company in Taiwan - Syntek Design Technology Co. Ltd. was
established in HSIP in 1981. The General Manager of the Company was formerly
the Manager of Circuit Design Department of Electronics Bureau, ITRI, and the
R&D Manager and Layout Engineer were also originally from ITRI. In 1983, it
established Greatek Electronics, and among its main shareholders, seven are from
ITRI, each takes on its main technology post.

6.2.2. Nearby Universities and Research Institutions

During the early days, HSIP manufacturers were in great demand of technological
professionals for starting new business. In addition to recruiting overseas scholars,
there are derived companies from ITRI, such as United Microelectronics
Corporation (UMC), Taiwan Semiconductor Manufacturing Corporation (TSMC)
etc., all bring human resources to HSIP. Moreover, there are also quite a lot who
resigned and started business in HSIP or joined in manufacturers, and technology
diffusion is also one of R&D results that government uses to evaluate ITRI. In
S&T circles, all consider the transfer of technology people is the best way for
technology transfer, thus, for electronics and information talents of HSIP, ITRI is
an important source.
The demand on human resources by HSIP has expanded gradually, including
those located at basic level and middle level. ITRI is still an important source for
middle-level manpower of manufacturers in HSIP, and for the gradual expansion
of basic-level manpower required, graduates from Tsing Hua and Chiao Tung
University provide the supply. In addition, the manpower demand by HSIP during
early phases required experience, therefore, students who just graduate regard
ITRI as the "Shaolin Temple", a place for studying R&D ability, and advance to
HSIP after three to five years, even manufacturers in HSIP regarded this as a
reference criterion when evaluating employees' value. After the demand increases
280 K. Wang

gradually and companies reach certain growth stage, by the method of hens
leading chickens, graduating students become the very important new force. In
particular, for the characteristics, prospect, payment and welfare of hi-tech
industries in SIP all have very strong attraction, there is no special difficulty in the
respect of recruiting good staff. Only for the change on industry technology, the
semiconductor industry in HSIP has greatly attracted many talents, thus brought
some squeezing phenomenon, consequently, what manufacturers recruit may not
necessarily be from the right department and can master the work within short
time, which increases the cost for on-the-job training and short-term training by
enterprises in HSIP, and also is a waste of education resources.
In order to cooperate with technology development and manpower demand by
enterprises, improve manpower quality, and shorten the time for new entrants'
before-the-job training, the Administration Bureau of HSIP has cooperated with
research institutions like Chiao Tung and Tsing Hua University, ITRI etc. to help
all kinds of training for professional technology human resources, such as
professionals training for deep sub-micrometer and liquid crystal display,
integration between communication and computer, computer and automation etc.
eight items of specialized technology, and also management, a second skill, and
evening on-the-job training etc. From 1993 to 1998, there are altogether 24,390
people undergoing training, and 18,271 participating in management course and
special topic discussion (MOEA 1999), which indicates that people who work in
HSIP are taking an active attitude towards participation in training courses when
facing fast development of technology, to keep their competitiveness. Recently,
the Ministry of Education has opened the on-the-job class for master degree,
which will be a catalyst to further promote improvement on manpower quality.
This is the convenience on manpower recruitment and training of human resources
brought by the government choosing places close to academic and research
institutions to set up science park.

6.2.3. Employees' Profit Sharing System

In order to attract participation by more professionals, manufacturers in HSIP
adopt the system of employees drawing dividend and becoming shareholders,
which indeed bring significant influence on HSIP. Basically, this system can
enable both the labour and capital to become rich, simplify the relationship
between the labour and capital, and reduce significantly the inner transaction cost
of enterprises. Its positive effect lies in keeping employees, so it can attract talents
for individual company, from the overall point of view, however, professionals
flow in this market, difficult to be deeply rooted (pulling force from the market).
On the other hand, since talents are easily attracted, the speed of enterprises
creation and experience exchange is even faster.
 High-Tech Industrial Clusters in the Asian Countries 281

6.3. Source for Capital

During the early stages, there was SIP providing land needed for setting up plants,
but difficulty in acquiring capital, especially for technology industry with high
risks. For the gradual establishment of capital market in Taiwan, capital changes
to mainly come from capital market, and the following gives an analysis on
influence of HSIP hi-tech manufacturers' capital demand concerning policies like
government providing funds for starting business and working capiLal etc.

6.3.1. Government Provides Funds for Starting Enterprises

During the early stages of development, the environment for HSIP to acquire
funds was not abundant. The government coordinates Bank of Communications to
provide financing for S&T enterprises or carry out investment on S&T enterprises,
and the Development Funds of Executive Yuan was established for investing in
S&T enterprises. In order to put its technology into practice by enterprises, ITRI
also establishes innovation investment company, besides assisting ITRI to transfer
technology to enterprise, it also assists manufacturers to acquire funds or get
investment from the company to accelerate the commercialisation of high-risk and
high-market potential technology of manufacturers. Furthermore, it rewards
establishment of entrepreneurial and investment enterprises, and related hi-tech
companies invested, which all have significant influence on the start-up funds for
manufacturers in HSIP. For instance, among the 32 manufacturers brought in
Taiwan in 1996, there were 9 Taiwan entrepreneurial investment companies,
having carried out investment to ten newly entered manufacturers among them
(IEK 2001).

6.3.2. Assist Manufacturers to Raise Working Capital

With the gradual establishment of Taiwan capital market, the stocks of
manufacturers in HSIP were listed, and for their proper management, professional
managers run the companies, and staff become shareholders, so that the
acquirement of capital for company operation or expansion also comes from
employees' investment, as well as capital from many entrepreneurs, venture
capital, bank investment, and investment transferred from related enterprises.
Since manufacturers in HSIP have close relationship in terms of personnel
exchange, information exchange, and technology transfer or cooperation, for
enterprises with great potential, when expanding factories and increasing
investment, all are very easy to obtain funds from non-official relationship.
Transferred investment is also one of the main methods to acquire funds in SIP,
especially the electronics industry. When manufacturers have very good operation,
for the upper, middle and lower-reach key components and parts and products,
they will transfer investment to set up new companies, thus to take shape within
short time in terms of funds, market, product, important high-level managers,
management system etc., shorten the preparation period for starting new business
and reduce cost. Therefore, very closely connected network is formed, as long as
282 K. Wang

the economy is in good cycle, these factors helping to start business can almost
create another manufacturer at any time.
In order to assist manufacturers to acquire working capital, the government has
adopted many measures, such as the measure of accelerated depreciation of
equipments, so that those with very large investment on equipments can delay in
paying taxes for accelerated depreciation, and also get working capital without
interest, which have quite significant influence on semiconductor and
optoelectronics mainly for manufacture. The government also looses its regulation
on retaining surplus and compulsory distribution. Following the trends of getting
funds easily by HSIP manufacturers, the influence on them is also relatively
reduced. Furthermore, the government provides middle and long-term low-interest
bank loan, and invests in manufacture of key products for important S&T
enterprises, this policy has the function of leading manufacturers to invest in
products with foresight and also risks. So it suits relatively more to the demand of
expanding funds by manufacturers of HSIP.

6.4. Preferential Land Tax and Other Levies

There was insufficient investment on high technology in Taiwan during the 1980s.
In order to encourage investment, the government has adopted many measures, for
instance, the income tax for profit business can enjoy 5 (4) years of exemption
from taxation, accelerated depreciation, paying taxes within limited amount,
investment offset, loosened expenses etc. Preferential taxes, which can really be
considered as great achievements in reducing and exempting from taxes.
Among all measures of reducing and exempting from taxes in HSIP, that the
income tax for profit business can enjoy 5 (4) years of exemption from taxation
brings the most favourable treatment for manufacturers, since it has significant
influence on manufacturers' profit, reduce payment of taxes. For the business
income tax of Taiwan enterprises, it is averagely 25%, in order to encourage
hi-tech manufacturers' investment in HSIP, the government has adopted
preferential measures on their business income tax. The first several years for
hi-tech new businesses are normally in deficit, however, it still encourages
investment, and its achievements can be detected from the fast accumulation of
capitals in HSIP.
The government gives five-year free business income tax (exemption five) for
businesses belonging to emerging and important S&T enterprises, and four-year
free business income tax for locating in HSIP (exemption four). Since almost all
enterprises in HSIP can enjoy five-year exemption and also four-year preference,
they enjoy more favourable treatment comparing with manufacturers outside HSIP.
The main goal of exemption five is to provide free tax during the initial stage of
establishment to S&T type enterprises with relatively great risks, to encourage
entrepreneurial spirit and start emerging S&T industries.
In order to reduce the cost for instruments and equipments of hi-tech enterprises
and raw material, and further to increase business profit, the government has
adopted import duties and goods tax reduction and exemption measures, to
 High-Tech Industrial Clusters in the Asian Countries 283

promote manufacturers in HSIP to mainly purchase machines, raw materials,

components and parts from overseas, which are more advanced than local
technology and therefore can raise domestic industry.

6.5. Operation Environment

In order to provide manufacturers with excellent hi-tech enterprises operation

environment, the government has adopted many measures such as single service
window, simplified administration procedures, and keeping improving
infrastructure etc. to build an excellent development environment for
manufacturers. The government also adopts cargo automation system, to create
competitive advantages for HSIP manufacturers and also grab the business
opportunities. HSIP has taken the lead to adopt cargo automation system since
Nov. 1992. This cargo automation system connects through SIP information
network (SPNET), to carry out goods import/export examination operations (visa,
hi-tech products control, software export examination), warehouse operation
(enter/exit storehouse) for related units (Administration Bureau and warehouse
and tr:msportation centre, manufacturers, customs declaring office, post office, III
software examination station inside HSIP), and also transform applications after
signature and take storehouse into electronics data interchange (ED I) message, to
transmit specially via trade-van, which not only provides fast and convenient
services to SIP enterprises, increases products competitiveness, but also improves
the administrational efficiency and image of government organs.
Since the Administration Bureau of HSIP provides single window
administrational service, which saves much time and documents travelling, so that
manufacturers can devote themselves to business, expanding product market, and
overseas and Taiwan hi-tech manufacturers all regard permission into SIP as a
great honour and pride. All operational activities needed after manufacturers enter
the HSIP can be obtained inside HSIP, so that enterprises operation can reduce
operational cost, and also prepare excellent living environment for employees,
reduce external cost, and also increase exchange of people from different
companies.

7. Supply of Human Resources from Universities

In order to cooperate with the demand on high-level manpower by the industrial

circles and development of Asia-Pacific Regional Operation Centre, the Executive
Yuan has specially combined enterprises, government, universities and public
research laboratories to organize employment-oriented short-term professional
training for high-level manpower, after the training, they can be directly employed
by enterprises, so that people with above master degree and related skill but
insufficient experience in practice, can serve enterprises after the short-term
training (MOEA 1999). For example, (1) assist talents from home and abroad
284 K. Wang

above master degree to engage in short-term research work: in order to assist those
from home and abroad at above master degree to get a job, before they decide
whom to work for, arrange them to engage in short-term research work to store up
talents. There are funds for subsidizing around 100 short-term researchers, to
employ by agreement, with the agreement term at most one year; (2) assist talents
at Ph.D. to engage in short-term research work: assist talents with Ph.D. degree to
work in non-governmental organizations, encourage non-governmental
organizations to use Ph.D. level talents, to cultivate talents needed for important
national construction plan, promote implementation of high-level human resources,
and also raise quality of non-governmental talents. The National Youth
Commission subsidizes research remuneration at 50 researchers every year, with
the longest subsidization term at most one year.
In the respect of bringing in R&D talents, to accelerate the upgrade of domestic
industries, the government brings in specialists lacked in Taiwan as well as trains
up actively high-level R&D talents, to improve S&T R&D ability, accelerate the
development of supporting industry to the most advanced S&T, and expand
recruitment of overseas high-level S&T talents, for example, NSC provides
subsidy to S&T figures from China or from China to the third region to come to
Taiwan to participate in S&T research program for public and private universities,
public research institutions and national laboratories. For the MOEA, in order to
assist domestic non-governmental enterprises to expand recruiting overseas
technology talents to return home to serve, any specialist beneficial to raise
technology and added value of enterprises can raise work program together with
their employer, after examination and approval by "Examination Committee for
Recruiting Overseas Industry Specialists", they can receive subsidy for the
specialists' salary part, at the limit of three years.
In order to strengthen the implementation of overseas high-level technology
talents and domestic talents above master degree, the government has adopted
many recruitment measures (NSC 2000a), such as: (1) set up service window of
"strengthen implementing high-level S&T talents program", to provide
introduction of this program and also information needed by each domestic
institution, and the execution unit includes each unit under the Executive Yuan,
state-run enterprises and ITRI etc.; (2) hold high-level human resources seminar
and exhibition, to assist talents above master degree to understand working
environment of Taiwan, and encourage enterprises to recruit high-level talents; (3)
starts to hold seminars for introducing overseas new S&T since May 1994, to
discuss the ten top emerging industries, and also assist scholars travelling overseas
to come home to give lecture and pay visits.
In the respect of training industry technology talents, in order to cooperate with
industry upgrade and train talents required by industry development, the MOEA
has advanced comprehensively the "Phase I five-year plan for training of industry
technology talents" from July 1990 to June 1995, including raising product quality,
industry design, industry automation, industry manufacture process technology etc.
talents training, and started carrying out phase II five-year plan since 1996. It
carried on advancing "Five-year plan for training of industry technology talents"
in 2000, to strengthen combining academic, research institutions and
 High-Tech Industrial Clusters in the Asian Countries 285

non-governmental enterprises teachers, facilities and teaching materials to raise

substantially the percentage for graduates from the science and engineering
department of junior college to engage in manufacturer through on-the-job
training and professional switching training, to strengthen the technological level
of workers presently in conventional industries, and also solve the problems of
generally insufficient manpower in conventional industries (MOEA 1999).
Corresponding to the serious insufficiency in hi-tech engineer-level talents of
emerging S&T key industries, the MOEA and Employment and Vocational
Training Administration Council of Labour Affairs have advanced the "Hi-tech
key industries talents training program" in July 1998, lasting three years. The
MOEA also cooperates with the "S&T talents training and implementation
program" by the Executive Yuan to establish the "Cultivating high-level
cross-industry talents with S&T background program" in 2000. It is estimated that
around 60 R&D or management talents with S&T background will be selected
during three years from Taiwan emerging hi-tech fields related enterprises and
research institutions, recommended by the institution they1\Old a post, subsidized
by the government, mainly to participate in overseas practice and assisted by
short-term training, to have six to nine-month investment evaluation, technology
transfer, and intellectual property etc. professional training and practice. In 2000,
among these 220 people recommended, 50 have passed examination, and after
around two-month training in Taiwan, it is planned that 20 will be chosen to have
further practice overseas.
In addition, in order to cooperate with the development rends of 3C integrated
S&T, the government subsidizes each circle to engage in R&D of 3C integrated
S&T and products, and the Ministry of Education will also started to encourage all
universities to have courses that teach 3C integrated S&T since 2000. It is
estimated that 750 3C S&T integrated talents can be training each year, and also
150 high-level talents with master and Ph.D. degrees. Corresponding to the highly
development of electronics, information and electric machinery in Taiwan, there is
serious insufficiency in related talents, especially wireless communication,
Internet, and software talents, and also for the demand on Business to Business (B
to B). E-commerce talents, electronics industry technology talents cultivation
program, industry computer talents cultivation program, and strengthening
information electronics hi-tech talents cultivation programs etc. were implemented,
by on-the-job class and cultivation class, having trained up more than 13 thousand
people from 2000 to 2002, which solves the lack of talents in domestic S&T
industries. Among them, the demand for communication and network talents is
even more pressing, so the cultivation for information electronics hi-tech talents
will be expanded.
286 K. Wang

8. Innovation System Forms Cluster Effect

An important case of Taiwan hi-tech industry development - the innovation course

of the semiconductor industry, mainly includes the function played by the
government, research institutions, universities and enterprises. During the
development courses of Taiwan semiconductor industry, the government
sponsored R&D funds of ITRI, and ITRI plays the sower of planning industry
development and also the cultivator who engages in R&D, to assist the
establishment of emerging industries, and create the prosperity for Taiwan
semiconductor industry, so that it becomes the representative of domestic hi-tech
industries.
The formation of overall integrated circuits industry can be described as that, the
government chooses integrated circuits technology, ITRI brings in 7-micrometer
integrated circuits technology from USA, after digestion and R&D, transfers to
industrial circles to carry out commercialisation. Further combines with
enterprises to develop sub-micron technology, and develop towards independent
memory products. During the industry development course for as long as 30 years,
the government subsidizes funds for R&D of ITRI, and ITRI plays the sower of
planning industry development and also the cultivator who engages in R&D, to
assist the establishment of emerging industries, and create the prosperity for
Taiwan semiconductor industry.
Our semiconductor industry started to establish factory by Kaohsiung
Electronics Company in 1966 to assemble transistor. From 1969 to 1973, there
was Philips, TI, RCA etc. foreign companies establishing assemble factories of
integrated circuits in Taiwan. Among Taiwan capital factories, there are Lingsen,
Orient, Siliconix etc. During this period, our semiconductor industry has been in
the assemble stage of integrated circuits. In the respect of academic field, only
Chiao Tung University has semiconductor lab for cultivating semiconductor
talents. After introduction by ITRI and establishment of HSIP, the clus~er effect
was incubated. For the efforts made for more than ten years, it finally helps
Taiwan semiconductor industry to take a place on global stage.
From the cluster effect of Taiwan information electronics industry, from Taipei,
Taoyuan County to Hsinchu County, in such a SO-Kilometer long and narrow
geographic area, it is almost the gathering place for main manufacturers. If just
checking integrated circuits industry, the obvious cluster situation can be found.
Except for Nanya Technology Corp., the left 15 integrated circuits wafer factories
are all situated in HSIP, and integrated circuits design industry is distributed in
Hsinchu and Taipei, package manufacturers base in Hsinchu, Taichung, and
Kaohsiung. Under the situation of insufficient land in HSIP, integrated circuits
manufacture develops towards Tainan Science Park, and package manufacturers
change to Hukou. Integrated circuits design industry still stays in Hsinchu and
Taipei for its small land demand and high requirement on professionals and
information.
For peripheral supporting industries, such as mask, conducting wire frame,
chemicals, equipment agents etc. related proprietors, under the consideration of
 High-Tech Industrial Clusters in the Asian Countries 287

serving customers without having to go far, most establish their base inside HSIP
or in its vicinity, so that the upper, middle, and lower-reach system of integrated
circuits industry are almost all concentrated in a closed geographic area. We can
regard this kind of industries cluster situation as a gigantic production satellite
system or imagine it as a "virtual" large company with vertical integration of
upper and lower reach. Almost without geographic and transportation limitation,
more economical ways can be used to carry out mutual communication or
products transportation. Comparing with large enterprises system of USA, Japan
and Korea, for this large "visual" company of this mode can have the effect of
each "enterprise system" concentrating on their own specialized fields, to develop
the flexibility and fast emergency reaction ability of small enterprises into full,
thus their competitiveness is greatly raised.
The unique innovation system of Taiwan causes the cluster effect of HSIP
without the intention. For semiconductor industry, more than 90% of its upper and
lower-reach enterprises are concentrated in Taipei, Taoyuan, and Hsinchu etc.
County and City, so it is the most obvious one, as shown in Figure 3. This kind of
cluster effect helps Taiwan semiconductor industry to take a place globally. In
2000, Taiwan has a market share as high as 70% of global Original Equipment
Manufactor (OEM) market, and around 20% share of Dynamic Random Access
Memory (DRAM) production.
The cluster effect of semiconductor industry in HSIP has quite evident effect on
related businesses establishment. Take integrated circuits design for instance,
there were 3 integrated circuits design companies in Taiwan during the first half of
the 1980s, only one was in HSIP. For the 84 companies established after 1995,40
were established in HSIP, and as high as 62 if calculating together with Hsinchu
County and City. Please refer to Table 2.

Table 2. Relation table of Taiwan integrated circuits (IC) design industry establishment
time and geographic location.
Number 1980- 1985- 1990- 1995-
Taipei City 7 9 8
Taipei County 6 4 3
HSIP 4 12 40
Hsinchu City 3 12
Hsinchu County 10
Tainan County 1
Source: Semiconductor Industry Yearbook, June, 2001.
288 K. Wang

Ie Design 28
Wafer
Design Tools 5

Ie Design
Ie Manufacture
Ie Package
Ie Testing
Wafer
Mask
Tools
Optoelectronics

o•
... ..
Ie Design

o
I
40KM

Fig. 3. Taiwan semiconductor industry manufactures cluster.

Source: Industrial Economics & Knowledge Centre (IEK). June 200 I.
 High-Tech Industrial Clusters in the Asian Countries 289

9. Conclusions

In order to promote the development of Taiwan hi-tech industry, based on the

development environment in 1980s, aiming at the factors needed for developing
hi-tech industries, all kinds of measures as many as seeds have been implemented,
to advance the germination and development of high technology. After
development for 20 years, these seeds have grown up to become large trees, and
today's hi-tech forest. For this forest, these seeds can bring new vitality but their
function has gradually withdrawn.
Taiwan's policy didn't pre-select electronics information industry, but all
factors pushed the growth of electronics information industry in Taiwan at the
fastest speed. The global development environment and electronics information
industry, especially semiconductor industry, have driven the development of
overall high technology. It can be further expected from the phenomenon of
communication and Internet deepening into every level of mankind, that they will
bring limitless opportunities for Taiwan to develop high technology.
For the innovation system that is built in Taiwan after 20 years, semiconductor
industry is a successful case from null to prosperous development. In the future,
the innovation system will put forth efforts on established system. Facing the
knowledge economy era and fast development of S&T, knowledge has become the
driving force for economic development, and gradually replacing the leading
positions of traditional manufacturer factors of land, labour, and resources etc.
Industry based on S&T knowledge, therefore, has become the main power for
economic growth of advanced countries in technology. For example, the
knowledge industry output value of OEeD countries, more than half are related
with creation and application of knowledge, especially electronics information,
communication, and biotechnology industry with gene as the principal part, will
become important S&T industries for the development ot knowledge economy.

References

Freeman, C. (1987), Technology and Economic Preference: Lesson from Japan, London,
Printer.
Industrial Economics & Knowledge Centre (June 2001), 2001 Semiconductor Industry
Yearbook, Taiwan, IEK.
Lundvall, B-A. (ed.) (1922), National Innovation Systems: Towards a Theory of Innovation
and Interactive Learning, London, Printer.
Ministry of Economic Affairs (1999), White Book of Industrial Technology 1998 Taiwan,
R.O.C., MOEA.
Ministry of Economic Affairs (2001), MDEA Technology Development Program 2000,
Taiwan, MOEA.
290 K. Wang

National Science Council (2000a), The Annual Report of Science and Technology in R. 0. C,
Executive Yuan, Taiwan, NSC.
National Science Council (2000b), Indicators of Science and Technology R.o.C 2000,
Executive Yuan, Taiwan, NSC.
Nelson, R. (ed.) (l993), National Innovation Systems: a Comparative Analysis, New
York/Oxford, Oxford University Press.
OECD (1997), National Innovation Systems, OECD.
Science-based Industrial Park Administration (l998), The Annual Report of Hsinchu
Science-based Industrial Park, Taiwan, NSC.
Science-based Industrial Park Administration (1999), The Statistics of Quarterly Report of
Science-based Industrial Park, Taiwan, NSC.
Financial Markets, Industrial Clusters
and Small and Medium-Size Enterprises

Angelo Tantazzi I

Abstract. While the quantitative importance of Italian industrial clusters is

indisputable, the structural weaknesses that are inherent in the cluster system have
been pointed out repeatedly in recent years. The study identifies the economic
conditions where the Stock Market can help strengthen and energize Italy's
industrial clusters. The cluster companies already listed on the Stock Market are
identified, and their importance for capitalisation and turnover of the market are
calculated. The contribution of cluster companies seems to be particularly relevant
in the group of industrial small and medium-size listed companies.
The new segmentation of the Stock Exchange is finally described. The recent
introduction of the STAR Segment (Segmento Titoli con Alti Requisiti) in the
stock market is designed to create a new regulatory, organizational and
microstructural framework that is better suited to enhance the value of small and
medium-size industrial enterprises. This solution
provides a source of diversification and offers retail and institutional investors
attractive investment opportunities in areas that previously had been neglected due
to problems of visibility, transparency and liquidity.

1. Italian Industrial Clusters: Definitions

A study of Italian industrial clusters entails an analysis of a segment of the

economy the importance of which is comparable to the contribution that large
multinational groups or capital-intensive industries provide to the national
economy in Italy and in other countries. The fact that industrial clusters have also
been called "spontaneous multinationals" (Fortis 1998) is indicati ve of their
significance. It has been estimated that, in absolute terms, the value added
generated in the early 1990s by the products that are emblematic of Italian
industry, a large portion of which is manufactured in industrial clusters, was
greater than the value added created in the United States by the pharmaceutical, oil
refining, office machines and beverage industries combined. Italian textile,
garment and footwear manufacturers generate more value added than the German
automobile industry. Industrial clusters and specialized manufacturing systems

IMy thanks to Alessandra Franzosi and Enrico Pellizzoni (Market Research and Analysis,
Borsa Italiana) for their invaluable contribution to the analytical portion of this
presentation.
292 A. Tantazzi

account for an estimated two million jobs, and industrial clusters contribute one-
third of Italy's exports.
While the quantitative importance of industrial clusters is indisputable, the
structural weaknesses that are inherent in the cluster system have been pointed out
repeatedly in recent years. The main weaknesses include the vulnerability of
clusters to innovations that originate outside the cluster, the lack of economies of
scale at the individual company level (offset only in part by system-wide
economies of scale) and the inability of industrial clusters to formulate a well-
defined strategy in the absence of a system leader. It has been suggested often that
these problems could be overcome if the individual manufacturers can be made to
grow to a large enough size while, at the same time, retaining their entrepreneurial
spirit, flexibility and ties to local traditions. One of the obstacles to growth for
these enterprises (as well as for the Italian economy as a whole) is the absence of a
financial system that can provide adequate support to companies in their efforts to
expand.
In order to understand how the stock exchange and the financial markets in
general can help strengthen and energize Italy's industrial clusters, one must first
clearly define the concept of industrial cluster, wlth the goal of identifying the
different configurations that exist within clusters and of understanding, given
these different internal organizations, the various problems and specific challenges
they face. Studies of industrial economics usually make reference to "small
business systems" the main structural characteristics of which are summarized
below:
• There is a strong manufacturing specialization at the local level, which is
identifiable not just by the preponderance of one industry or area of
manufacturing, but also by a production system that links different
subsystems. These systems are all tied to a single product that is
characteristic of the local economy.
• The production of the cluster is sufficiently large to account for a sizable
percentage of the national and/or international output for the industry or
specific product that the cluster specializes in.
• There is a very clear division of labour among the companies in the local
system, which produces a high level of horizontal integration with low
transaction costs.
• There is a plurality of small and medium-size enterprises instead of one
leading company that dominates the local system.
• An information system has evolved spontaneously within the cluster,
transforming the knowledge of the individual entrepreneurs into a common
asset.
• The employees are highly skilled.
• There is a high rate of new business creation.
• There is an abundance of entrepreneurial spirit that fuels the growth of the
manufacturing system.
This definition is broad and applies to a number of possible system
configurations. In order to measure the importance of cluster companies
 Financial Markets, Industrial Clusters and Small and Medium-Size Enterprises 293

quantitatively at the stock market level and determine whether they have a
different relevance within the context of the overall financial system, according to
the characteristics that identify each area and its different growth potential, we
must first analyse these complex systems systematically, classifying them into
three types of small and medium-size business systems (Garofoli 1983):
• In the so-called specialized manufacturing areas, companies compete in the
same markets, since they produce the same goods or do the same processing
work. The system's structure is horizontal.
• In the case of "local manufacturing systems", small enterprises operate
within the same industry, as is the case for "specialized manufacturing
areas", but intra-industry transactions between companies are quite extensive
(quality subcontracting). There is no production integration among the
different areas of manufacturing, the structure of the industrial system is
adequate and technology is generally imported from outside the system.
• In the so-called system districts (for example, Biella, Vigevano, Prato, Carpi,
Sassuolo), the division of labour among the different enterprises is quite
extensive, resulting in the gradual development of a local manufacturing
system. There are intra - and inter - industry transactions that lead to the
creation of a system that produces capital goods used to manufacture finished
products. The development of the district is driven by the utilization of local
resources (entrepreneurs, capital, skilled workforce, local technology).
The different situations described above, while characterizing different
production systems, also provide an outline that could explain the evolution of
industrial clusters. The classic Italian industrial cluster, in its original incarnation,
meets the criteria of the third definition provided above - i.e., the system district.
The distinctive features of industrial clusters are small individual companies, a
fragmented manufacturing system and the existence of a deeply rooted individual
and/or family-based entrepreneurial culture. Under these circumstances, the
growth of cluster companies to a size large enough to permit access to financing as
a means of further expansion - including, ultimately, access to the capital markets
- generally occurs over the long term.
However, the world of traditional industrial clusters is particularly vulnerable to
structural change caused by the occurrence of extraordinary conditions resulting
from the shock of innovations and/or external competition.
When this occurs, industrial clusters generally are forced to alter their "natural"
growth path, changing their traditional configuration and morphing into "local
manufacturing systems" or "specialized manufacturing areas". This process of
transformation is driven by the development of new, dynamic organizations that
are better suited to grow rapidly. Obviously, the emergence of companies of
unequal size can undermine the specificity of a traditional industrial cluster,
because large enterprises lack the operating flexibility and adaptability that make a
cluster organization inherently efficient. However, in the presence of an external
shock, substantial changes, including the adoption of a new organizational
structure, are essential to the very survival of a manufacturing system.
The availability of adequate financial resources is crucial, particularly during
these transitional phases, and is of vital importance in allowing districts to
294 A. Tantazzi

reorganize their manufacturing systems and remain competitive. Let us now

review a few examples:
• The creation of large companies such as Benetton, Diadora, Dolomite and
Tecnica in the sports apparel cluster located in the province of BeHuno
followed the arrival of synthetic materials, which required the attainment of
economies of scale, and, during the last 20 years, the conversion from
leather-working to plastics technologies, as well as the movement of
production facilities to countries with lower labour costs and the ability to
control multiple brands in order to compete successfully in international
markets. As these changes occurred, the cluster responded to external
technological and competitive pressures by shedding its artisan structure to
become an organization in which small, highly specialized subcontractors
handle the simpler aspects of the manufacturing process and support the
leading companies in the cluster.
• In the case of the industrial cluster that produces packaging equipment in the
province of Bologna, the turning point in its evolution into a system that
gravitates around a few companies that are world leaders in their respective
markets (Sasib, Ima, Nuova Fima Imballaggi, Mg2 and Officine Meccaniche
Corazza) occurred with the integration of mechanical and data processing
technologies in the manufacturing process and the need to develop
international sales and marketing strategies to counter increasing pressure
from German and Japanese competitors.
When the definition of industrial cluster, according to which the common
denominator of all cluster companies is a strong local character and membership
in a manufacturing system with well defined territorial boundaries and a high
degree of similarity (in terms of number of employees, valued added produced and
propensity to export), is broadened to include a local economic system, it is
conceivable that competition in the same product segment could trigger a natural
process whereby the most nimble and dynamic companies grow the fastest. The
development of a sizable number of medium-size enterprises occurs more easily in
the absence of the strong system-wide inter- and intra-industry relationships that
normally exist in industrial clusters characterized by transactions "between
equals".
The financial markets can playa different function in each of the business
systems described above, which can also be viewed as different phases of the
same phenomenon as it evolves over time. Traditional industrial clusters obtain
new resources by forging close links with the banking system. Other types of
financing become necessary during the phase when companies grow from local
enterprises to cluster leaders with a strong international bias. Financing for such
companies first takes the form of private equity financing and can evolve into
direct access to the capital markets, provided the cultural environment is
favourable and the companies have developed sufficiently structured management
organizations.
 Financial Markets, Industrial Clusters and Small and Medium-Size Enterprises 295

2. Industrial Clusters and the Stock Market

Owing in part to the differences in the way industrial clusters evolve (as outlined
above), the number and size of cluster companies that are publicly traded in Italy
can vary widely, depending on which definition of "industrial cluster" is chosen.
If the most comprehensive (but also the most restrictive) definition of "system
district" is used - i.e., the definition that specifies a high degree of intra - and inter-
industry integration achieved as a result of local growth that is made possible by
the optimum utilization of local resources only, the percentage of publicly traded
companies is relatively small. Based on the classification of industrial clusters
developed by Ceris (Istituto di ricerca sull'impresa e 10 sviluppo) - CNR
(Consiglio Nazionale delle Ricerche) (1997), only four cluster leader companies
are listed on Borsa Italiana's capital markets. They are: Luxottica and Safilo,
which are the leaders of the Cadore (Veneto region) eyeglass cluster; Ratti for the
Como silk cluster; and Olivetti for the Canavese (Piemonte region) information
technology cluster.
Table 1 lists publicly traded companies that operate in Italy within districts
characterized by a high degree of manufacturing specialization, identified on the
basis of the percentage of the total national exports of a specific product provided
by the respective provinces. The definition used in this case is that of "specialized
manufacturing area".
Using this definition, the number of publicly traded companies increases to 25.
The most important companies include Benetton and Stefanel (province of
Treviso) and Marzotto (province of Vicenza) for the garment industry; Tod's
(province of Ascoli Piceno) for the footwear industry; and Merloni (province of
Ancona) for the household appliance industry. These large companies are the
product of manufacturing systems - the Northeast cluster and the Adriatic
backbone cluster - that started at the local level and, during the last ten years,
became important players in the national economy. Other examples include
smaller but very dynamic companies such as Gewiss, which produces electrical
equipment in the province of Bergamo; I.M.A. Industrie Meccaniche, which
manufactures packaging equipment in Bologna; and Emak, a producer of
agricultural equipment in the province of Reggio Emilia.
The relative importance of this group of "publicly traded cluster companies" is
significant both economically and in terms of its impact on stock market
indicators. Based on aggregate data taken from corporate annual reports for 1999
(source: BUREAU van Dijk 2001), this group of companies employed more than
48,000 people and had revenues of 15,941 billion lire and invested capital of
22,554 billion lire, including 6,705 billion lire in stockholders' equity.
With respect to stock market statistics, at the end of May 2001, this group of
companies accounted for 2.7% of the total capitalization of the Italian stock
market. If the focus is narrowed to the macro group of industrial companies (to
which all publicly traded cluster companies belong), the relative weight of these
companies is greater, since they account for 21.7% of the group's total
capitalization.
 N
Table 1. Quantitative imQact of cluster comQanies on the Italian Stock Market*. -.0
a-
COMPANY AREA CAPIT AUZA TION % TOT. JAN-MAY % TOT.
(millions of euros) CAP. TRADING TRADING ;>
...,
..,
VOLUME VOLUME
(millions of euros)
.,.'"
N
t:L

STOCK MARKET - BLUE CHIP SEGMENT

1 BENETTON GROUP Treviso - clothing 3,241.0 0.47% 561.3 0.195%

2 LUXOTTICA Belluno - eyeglasses 8,025.8 1.16% 144.8 0.050%
3 MARZOTTO Vicenza - clothing 1,050.3 0.15% 210.0 0.073%
4 SAFILO Padua - clothing 1,268.2 0.18% 216.9 0.075%
5 TOO'S Ascoli P. - footwear 1,452.6 0.21% 92.6 0.032%
TOTAL CLUSTER COMPANIES 15,037.9 2.17% 1,225.6 0.426%
TOTAL BLUE CHIP SEGMENT 660,876,2 95.47% 283,654.3 98.486%

STOCK MARKET - STAR SEGMENT

6 CSP INTERNATIONAL Mantua - hosiery 83.8 0.01% 11.1 0.004%

7 INTERPUMP GROUP R. Emilia - pumps 325.5 0.05% 59.6 0.021%
8 MARIELLA BURANI R. Emilia - clothing 220.0 0.03% 62.5 0.022%
9 SABAF Bn:scia - plumbing fixtures 147.0 0.02% 30.6 0.011%
TOTAL CLUSTER COMPANIES 776.3 0.11% 163.8 0.057%
TOTAL STAR SEGMENT 5,288,9 0.76% 911. 7** 0.317%

STOCK MARKET - REGULAR SEGMENT

10 DANIEll & C Udine - wood products 270.4 0.04% 14.5 0.005%

11 FILA TURA POLLONE Biella - wool 15.7 0.00% 2.3 0.001%
12 GEFRAN Brescia - electr. equipment 84.5 0.01% 29.9 0.010%
Table 1. (continued) ---------_._- ---

13 GEWISS Bergamo - electr. equipment 592.1 0.09% 51.8 0.018%

14 GILDEMEISTER Bergamo - metalwork. equip. 119.5 0.02% 1.7 0.001%
15 CERAMICHE RICCHETTI Modena - floor tiles 203.1 0.03% 12.7 0.004%
16 IDRA PRESSE Brescia - metalwork. equip. 303 0.00% 1.3 0.000%
17 IMA Bologna - packaging equip. 316.3 0.05% 37.1 0.013% ""T1
18 MARCOLIN Belluno - eyeglasses 74.9 0.01% 4.6 0.002% .,5·
:::
(")
19 MERLONI Ancona - appliances 471.7 0.07% 56.5 0.020%
20 SIMINT Modena - clothing 293.2 0.04% 109.4 0.038% E
21 STEFANEL Treviso - clothing 141.3 0.02% 6.6 0.002% .,3:::
22 CEMBRE Brescia - metalwork. equip. 46.2 0.01% 1.6 0.001% *
0.002%
F"
23 EMAK R. Emilia - farm equipment 60.6 0.01% 4.8 s
~
24 LINIFICIO Bergamo - cotton textiles 28.7 0.00% 2.7 0.001% :;;
25 RATTI Como - silk 40.4 0.01% 1.1 0.000% 5'
TOTAL CLUSTER COMPANIES 2,788.8 0.41% 338.6 0.118% ::.
(J
TOTAL REGULAR SEGMENT 26,043.0 3.76% 3,448.9 1.198% [
Vi
TOTAL CLUSTER COMPANIES 18,603.1 2.69% 1,728.0 0.600% ".,
:::
Q.
TOTAL STOCK MARKET 692.208.1 100.00% 288,014.8 100.00% (10

* This list includes only companies that operate in the "specialized manufacturing areas" as defined by Fortis in II Made in Italy, Editrice II
Mulino. .,~
:::
** Trading volume is also for the January-March period, during which the STAR Segment was not yet operational. Q.

Source: Figures based on data provided by Borsa Italiana Spa. 3:::

(1)
Q.
C;.
:3
N·
'"
"
~
.a"
[;1-
(1)

'"
N
~
-.J
298 A. Tantazzi

When stated as a percentage of the 2001 trading volume (the Blue Chip
Segment accounts for 95% of the total), the contribution of publicly traded cluster
companies is relatively small (0.195% of the total). However, when the blue chip
shares are excluded, the relative weight of the cluster companies rises
significantly, reaching 11.4% of capitalization and 11.5% of the trading volume.
It is important to remember that cluster companies represent a considerable
portion of all publicly traded small and medium-size companies and that the
potential impact of unlisted cluster companies is highly significant when viewed
against the need to expand, both quantitatively and qualitatively, the roster of
Italian listed companies and bring the Italian stock market on a par with those of
other major European countries. At this point, about 2,000 small and medium-size
companies operating in traditional industries (internal estimate of Borsa Italiana)
are large enough to be considered for listing. More than one-third of these
companies belong to manufacturing sectors that are concentrated in industrial
clusters operating in the following fields: mechanical engineering (200
companies), textiles/garments (191 companies), food (134 companies), electrical
equipment (65 companies), furniture and furnishings (57 companies), ceramics
(31 companies) and jewellery (9 companies).

3. New Borsa Italiana Initiatives for Small and Medium-

Size Italian Enterprises

The recent introduction of the STAR Segment in the stock market is designed to
create a new regulatory, organizational and microstructural framework that is
better suited to enhancing the value of small and medium-size industrial
enterprises while providing a source of diversification and offering retail and
institutional investors attractive investment opportunities in areas that previously
had been neglected due to problems of visibility, transparency and liquidity. In
keeping with this approach, the stock market listing of cluster companies should
meet with a more favourable reception than in the past.
Borsa Italiana has designated liquidity and the transparency of information as
the two key areas in which STAR Segment companies will be required to make a
major effort.
The presence of a sponsor in the primary market, with responsibility for
performing the requisite due diligence work and ensuring periodic follow-up
research after the initial listing, and of a specialist in the secondary market (who
can be the same as the sponsor), coupled with regulatory requirements that ensure
a steady flow of information (prompt release of quarterly financial data in Italian
and English for STAR companies), helped solve the problems of limited direct
and indirect information, which is typical of small companies, and lack of
liquidity. This approach removes the danger of an inefficient pricing mechanism
and eliminates cost of capital distortions that occur when the market demands a
liquidity premium at the time of offering. For the same reason, the specialist is
required to ensure an orderly market, setting bid and asked quotes within a limited
 Financial Marl,.ets. Industrial Clusters and Small and Medium-Size Enterprises 299

spread while at the same time doing research that is designed to support the issuer
company through the publication of information on its performance and the
performance of the industry within which it operates.
ST AR companies must pay special attention to establishing effective corporate
governance systems so as to be able to provide the market with adequate disclosure
on their organization and on their general and operating performance. These
companies must comply with the Code of Conduct for Listed Companies.
Accordingly, they must appoint a sufficient number of independent directors to their
boards, establish an audit committee to oversee internal control systems and appoint
an "investor relator" to ensure continuity in their relations with investors and
intermediaries. The online services that Borsa Italiana has developed to meet the
special needs of institutional investors complete the framework of programs
designed with the specific purpose of increasing the visibility of new issuers and
creating the conditions necessary to trigger a virtuous circle that will be beneficial
for:
• Issuers, who will be able to successfully float secondary offerings with less
burdensome costs.
• Institutional investors, who will operate in a market that will be more
supportive of their interest in small and medium-size enterprises.
• Intermediaries, who will have an opportunity to specialize in origination and
research in the area of small and medium-size companies and in the
establishment of small-cap funds.
• The quantitative and qualitative expansion of the market, since a listing
system that ensures maximum transparency and orderly trading conditions
will help foster interest among companies that are potential candidates for
listing.
Even though the STAR Segment was created only recently, making the analysis
period relatively short, early evidence on trading in STAR securities has already
shown positive results:
• During the April-May period, the average number of contracts for STAR
securities executed daily was 14.6% higher than in the first three months of
the year, while the overall market declined by 2.3%.
• A similar trend was observed in the average daily value of the transactions
(+9.3% for STAR, -2.0% for the overall market).
• Volatility decreased by 17% for STAR securities, as against a drop of just
10.9% for the overall market.
At present, four of the 25 listed companies from the clusters shown in the
preceding table have agreed to be included in the STAR Segment. They are: CSP
International (a company based in the Castel Goffredo cluster, in the province of
Mantova, that controls some of the top women's hosiery brands), Interpump
Group (hydraulic pumps in the province of Reggio Emilia), Mariella Burani
(garments and fashion, also in the province of Reggio Emilia) and Sabaf (kitchen
furniture in the province of Brescia). At present, they represent 20% of the listed
companies included in the STAR Segment, 14.7% of the Segment capitalization
and 35.9% of the value of all trades executed in 2001.
300 A. Tantazzi

The fact that these companies chose to join STAR during the start-up phase
(trading in the Segment's securities began on April 2, 2001) shows that the
Segment's organizational and structural characteristics are being well received by
enterprises. The STAR Segment was launched to fill the relationship gaps that
existed until now in certain areas between the financial markets and medium-size
companies, despite these companies' excellent fundamentals and strong
competitive positions.
Economic literature and the case histories of individual companies concur in
showing that the financial markets playa decisive role in the development of
economic systems. The entities that oversee the operations of these markets are
responsible for stimulating the growth of their national economies by providing
enterprises with the tools necessary to seize the opportunities created by changing
competitive conditions. In pursuit of this goal, Borsa Italiana has focused its
efforts on restructuring the Italian financial markets. Easier access to the financial
markets and, consequently, to new sources of capital for small and medium-size
companies means making available to industrial clusters - the very heart of Italian
industry - a vital tool in their effort to manage the competitive repositioning
required by an accelerating process of globalisation.

References

BUREAU van Oijk (2001), CORom AIDA, march.

Ccris-CNR (1997), Innova=ione, piccole imprese e distretti industriali, Roma, Oocumcnti
CNEL III, Quadcmo 7.
Fortis, M. (1998), II "made in Italy" - Quando stile e creativitcl non sono .1'010 moda.
Bologna, II Mulino.
Garofoli, G. (1983), "Le arcc-sistcma in Italia", Politica ed economia, No.1!.
Garofoli, G. (1984), Modelli locali di sviluppo, Milano, Franco Angeli.
 List of Authors

Giacomo Becattini. Professor of Political Economics at Firenze University.

Member of the Accademia Nazionale dei Lincei (Roma), member of Accademia
Toscana di Scienze e Lettere "La Colombaria" (Firenze) and of the Accademia dei
Georgofili (Firenze). Chairman (1992-1995) of the Societa Italiana degli
Economisti. Received a number of awards, both scientific (Scan no Award for
economics) and literary (Prato and Firenze Awards). Since 1985 honorary member
of Trinity Hall college, Cambridge, England, and honorary citizen of Prato since
2001. Chief editor of Sviluppo Locale (Torino, Rosenberg & Sellier) and co-editor
of Il Ponte (Firenze) since 1957; member of the Boards and Scientific Advisory
Committee of the following journals: Economia e politica industriale (Milano,
IEFE), Economia Marche (Fabriano, Fondazione Merloni), Economia e societa
(Lecce University), Marshall Studies Bulletin (Firenze University), Storia del
pensiero economico (Firenze University), History of Economic Ideas (Pisa
University), Ricerche di storia sociale e religiosa, (Vicenza, Istituto di Storia
Sociale e Religiosa). He is also a member of the Scientific Committee of a number
of research institutions such as IRPET, Firenze, and IRIS, Prato. He has been
coordinating the "Prato weeks on local development", known as "Artimino's
Weeks", since the foundation in 1991. Member of the Firenze's Municipality
Government from 1980 to 1985.

Giovanni Dosi. He attended a post-graduate course at ISTAO, Ancona, 1977-78,

and holds a Ph.D. in Economics, 1983, Sussex University, GB. From 1984 to 1987
he was researcher at lUAU, Venezia. From 1987 to 1998 he was Professor of
Applied Economy at the Faculty of Statistics of Roma's Universita "La Sapienza".
Since 1998 he is Professor of Economic politics at the Scuola Superiore
Sant'Anna, Pisa, Italy. From 1992 to 1997 he was head of the international
research project "Technological and Economic Dynamics" at the International
Institute of Applied System Analysis (IIASA), Laxenburg, Austria.
Editor for Europe for the journal "Industrial and Corporate Change ". In the last
two decades he was a Visiting Scholar/visiting Professor at a number of
Universities, including UC Berkeley, Rio de Janeiro, Strasbourg, Stanford,
Sussex, Columbia University.

Maryann P. Feldman. Johns Hopkins University, Baltimore, Maryland Research

Professor, Department of Mathematical Sciences Director, Centre for Policy
Research, Whiting School of Engineering Adjunct Associate Professor,
Economics. Dr. Feldman holds a Ph.D. in economics and management and an
M.S. in policy analysis and management from Carnegie Mellon University. She
received a Bachelors of Arts in economics and geography from Ohio State
University.
302 List of Authors

Her book, The Geography of Innovation (Kluwer 1994), is in its second printing.
She is also editor of The Oxford University Handbook of Economic Geography
(with Gordon Clark and Meric Gertler, Oxford University Press, 2000),
Innovation Policy in the Knowledge-Based Economy (with AI Link, Kluwer,
2000) and the forthcoming Institutions and Systems in the Geography of
Innovation (with Nadine Massard, Kluwer, 2001).

Marco Fortis. Director of the Department of Economic Studies of Edison,

contract Professor of Industrial Economics at the Faculty of Political Sciences of
Universita Cattolica (Milano) since 1989. Member of the Scientific Committee of
the Research Centre in Economic Analysis, International Economics and
Economic Development (CRANEC) of Universita Cattolica (Milano), member of
the Scientific Committee and Vice-President of the Fondazione Comunita e
Innovazione, Vice-President of the Fondazione Guido Donegani, member of the
Board of Directors of the Fondazione Carlo Erba and of the Universita Carlo
Cattaneo, Castellanza.
Publications: numerous books and articles in books, journals and newspapers on
the subjects of Italian economy, industry and local production systems,
technology, development and international trade. Among his main books:
"Prodotti di base e cieli economici" (II Mulino, 1988, Iglesias Prize winner) and
"I! made in Italy" (II Mulino 1998). With Alberto Quadrio Curzio he edited "I!
made in Italy oltre il 2000" (II Mulino, 2000), belonging to the Fondazione
Comunita e Innovazione series of books.

Murray Gell-Mann. He received a Bachelor of Science from Yale, 1948 and a

PhD from MIT, 1951. Married poet Marcia Southwick in 1992. Distinguished
Fellow, founding member, and former Chairman of the Board, Santa Fe Institute,
Santa Fe, New Mexico, USA. His research interests are on topics ro;!lated to
simplicity and complexity. He is author of The Quark and the Jaguar (translation
I! quark e if Giaguaro) and co-editor of The Evolution of Human Languages and
Understanding Complexity in the Prehistoric Southwest. R. A. Millikan Professor
Emeritus of Theoretical Physics, California Institute of Technology, where he
taught from 1955 to 1993. He worked on weak interaction theory, "Eightfold
Way" scheme for particle classification, quarks, quantum chromodynamics,
renormalization group. He received the Nobel Prize in physics, 1969. Member of
Board, J. D. and C. T. MacArthur Foundation, helped start World Resources
Institute and MacArthur grants for nature conservation, was co-inventor of Rapid
Assessment Program for timely evaluation of natural areas considered for
protection. On UN Environment Program's Global Roll of Honour for
Environmental Achievement. He shared Ettore Majorana Science for Peace Prize.
He was Member of the US President's Science Advisory Committee, from 1969 to
1972 and of the Council of Scientific Advisors from 1994 to 2001.
 List of Authors 303

John H. Holland. Professor of Computer Science and Electrical Engineering and

Professor of Psychology at the University of Michigan; member of the External
Faculty of the Santa Fe Institute and member of the Executive Committee of its
Board of Trustees.
Holland has received numerous awards and honours, including the Levy Medal of
the Franklin Institute, the Henry Russel Lectureship of the University of Michigan
(the highest honour the University can confer upon a member of the faculty), the
prestigious MacArthur Fellowship, and the Pender Award of the Moore School of
Engineering at the University of Pennsylvania. Holland is known internationally
as "the father of genetic algorithms". His current interests include the theory of
complex adaptive systems and cognitive processes. His most recent book is
Emergence, Perseus Books (1998).

David A. Lane. He is Professor of Statistics in the Faculty of Economics at the

University of Modena and Reggio Emilia. In 2002 he will join the new Faculty of
Communication, Economics and Information in that University's Reggio Emilia
campus. Lane is also a member of the External Faculty and the Science Board of
the Santa Fe Institute. Among other academic honours, Lane has been elected a
fellow of the Guggenheim Foundation, the Institute of Mathematical Statistics, the
American Statistical Association and the International Statistical Association. He
has published articles and books in a variety of fields, including probability
theory, statistical inference, epidemiology, medical decision-making, economics
and business. He was Chairman of one of the sessions of the conference whose
proceedings are collected in this book.

Christian Longhi. Professor at L.A.T.A.P.S.E.S. - Laboratoire 'Trasformations

de l' Appareil Productif et Strategies Economiques Sectorielles" (Centre National
de la Recherche Scientifique), Val bonne (France) and IDEFI - Institut de droit et
d'Economie de la Firme et de l'Industrie (CNRS).
His publications include: "Technopolises and Technological Development, in: Y.
M. Rabkin (ed.), Diffusion of New Technologies in the Post-Communist World,
Dordrecht, Kluwer Academic Publishers, 1998; "Networks, Collective Learning
and Technology Development in Innovative High-technology Areas: The case of
Sophia-Antipolis", Regional Studies, 1999; Regional Evolutionary Trends in the
1990s, in: D. Keeble and F. Wilkinson (eds), High-Technology Clusters,
Networking and Collective Learning in Europe, 2000 (with D. Keeble);
Integration europeenne et dynamiques regionales, in: D. Torre, E. Tosi (eds),
Integration europeenne et institutions economiques, Paris, De Boeck, 2001.

Mario A. Maggioni. He holds a Master of Science and a Ph.D. from the Warwick
University, Coventry (UK). Now Associate Professor of Economics at the Faculty
of Political Sciences at Universita Cattolica (Milano). Member of the Steering
304 List of Authors

Committee of AISRe (Associazione Italiana di Scienze Regionali) and Associate

Editor of "Network and Spatial Economics".
He is the author of a number of articles and books on the subjects of economic
development, innovation and technological change, interactions between
educational processes and labour market and economic structure and local
production specialization at the local, national and European levels, and
interactions between "old" and "new" economy. Member of a number of Italian
and foreign inter-University research groups. His methodological interests span a
broad range from reticular analysis to ecologic-population patterns and
multi sectorial anal ysis.

Franco Malerba. He received a Ph.D. in Economics at Yale University (USA),

1983. He is now full Professor of Industrial Economics, Universita Bocconi and
Director of CESPRI (Centre on Innovation and Internationalisation Processes);
Eeditor of Journal of Industrial and Corporate Change; Associate Editor of
Journal of Evolutionary Economics, Advisory Editor of Research Policy. Director
of the European Summer School of Industrial Dynamics. President-elect of the
International Joseph Schumpeter Society.
His main research interests are: Industrial Economics, Economics of Innovation,
Industrial dynamics. Recent publications: S. Breschi, F. Malerba, L. Orsenigo,
'Technological Regimes and Schumpeterian Patterns of Innovation", Economic
Journal, vol. 110 (463), April 2000, pp. 388-410; F. Malerba (ed.), Ecol1omia
dell'Innovazione, Roma, Carrocci, 2000; A. Gambardella, F. Malerba (ed.), The
Organization of Econolllic Innovation in Europe, Cambridge University Press,
1999; F. Malerba, L. Orsenigo, Technological Entry, Exit and Survival: An
Empirical Analysis of Patent Data Research Policy, vol. 28 (6), August 1999, pp.
643-60. He was chairman of one of the sessions of the conference whose
proceedings are collected in this book.

Alberto Quadrio Curzio. Professor of Political Economy, Dean of the Faculty of

Political Sciences and Director of the Research Centre in Economic Analysis,
International Economics and Economic Development (CRANEC), Universita
Cattolica, Milano. He is member of Istituto Lombardo and Socio Corrispondente
of the Accademia Nazionale dei Lincei. Chief editor of the journal "Economia
Politica", II Mulino and member of the Scientific Committee of other journals
among which "Structural Change and Economic Dynamics"; "Journal of Policy
Modelling".
He has been Chairman of Societa Italiana degli Economisti (1995-98) and elected
member of CNR for ten years. He received the St. Vincent Award for economics
and the international Cortina Ulisse A ward. He received from the President of the
Italian Republic the "gold medal Award". He wrote approximately 300 scientific
publications, about 30 of which in English; one has been translated into Chinese.
His main subjects of study are: economic theory of scarce resources, income
distribution, education, economic development and economic-institutional forms
 List of Authors 305

thereof. He is Chairman of the Scientific Committee and Vice-President of the

Fondazione Comunita e Innovazione.

Enzo Rullani. He is Professor of Business Strategies at the Faculty of Economics,

Venezia U ni versity, after a period spent at the Udine and Verona U ni versities.
Professor of Economies of Industrial Enterprises and Technological Processes at
the Ph.D. in Business Economics, Bocconi University, Milano, Italy. He was a
Visiting Scholar at the MIT, Boston, in 2000. Chairman of the Technical-
Scientific Committee of the CUOA Foundation, Vicenza Business School, set up
by the North-East University. He is Chairman of COSES, the Venezia consortium
specialized in territorial analyses, and of TEDIS, a research centre on new
communications technologies at the Venezia International University.
He is author of numerous works on management and economics of industrial
businesses, among which Sistemi ed evoluzione nel management (edited with S.
Vicari), Milano, Etas Libri, 1999; "Complessita e informazione nella scienza
economica", Pluriverso 2, 1996.

Charles F. Sabel. He received an A.B. from Harvard in 1969 and a Ph.D. in 1978.
Currently he is Professor of Law and Social Science, Columbia Law School.
Publications include: "Quale volano per gli standard internazionali di protezione
sociale? Proposta per un progressivo miglioramento delle condizioni di lavoro
nell'economia globalizzata" with Archon Fung and Dara O'Rourke, in: II diriUo
del Mercato del Lavoro, n.1I200 1; "Treatment Courts and Emergent
Experimentalist Government" (with Michael Dort), Vanderbilt Law Review (April
2000); "An Unlikely Democracy: The US at the Millennium", lnternationale
Politik und Gesellschaft 112000 and presented at the OECD Forum for the Future
Conference, Berlin, December 1999; Design. Deliberation and Democracy,
(1998); The Second Industrial Divide: Possibilities for Prosperity (with Michael
Priore, 1984).

Francesco Salamini. In 1971 he received the degree of "Professor of Plant

Genetics", Roma University (libera docenza). He now is Director of the Max-
Planck-Institut fUr Zilchtungsforschung in Koln, Department of Plant Breeding
and Yield Physiology, Honorary Professor at he University of Koln, Professor (on
leave) in the University of Milano. From 1971 to 1974 he was Professor of Botany
and Plant Physiology, Universita Cattolica, Piacenza, Italy. In 1969 he was
Research Associate, Dept. of Botany and Plant Pathology, Purdue University,
West Lafayette, Indiana, USA. In 1988 he received the Medaglia dei XL,
Accademia delle Scienze, Roma. Since 1990 he is member of the Accademia
Nazionale dei Lincei, Roma. In 1992 he was member of the Accademia Europea,
London. In 1995 he received the Fondazione Invernizzi Price for the career. From
1985 to 1999 he was National Coordinator Program "Tecnologie Avanzate
Applicate aile Piante", Ministry of Agriculture, Roma, Italy. Since 1998 he is
306 List of Authors

Member of the Accademia dei Georgofili, Firenze. Since 1998 he is member, of

the Accademia Nazionale di Agricoltura, Bologna. Since 2000 he became
Chairman of Scientific Committee, Plant Genome Program, Genoplante, France.
He published more than 480 articles in specialized journals and is editor of 4
books. Member of the editorial board of Theoretical and Applied Genetics, J Gen.
& Breed., J Plant Breeding, Maydica, Plant Molecular Breeding, Genetical
Research, Atti Lincei, Plant Biology.

Angelo Tantazzi. He is Chairman and Economist at Prometeia Calcolo, Bologna.

He is also Chairman of Borsa Italiana SpA and Vice-president of the publishing
company II Mulino. He is member of several government's study commissions
and Associate Professor of Economic Politics at the Faculty of Political Sciences
of the Bologna University.

Kung Wang. He is Professor at the Graduate Institute of Industrial Economics,

National Central University, and Adviser at the Industrial Technology Research
Institute.
Among his past experience, he has been Adviser of the National Science Council;
Director General of the Science Based Industrial Park Administration;
Commissioner of the Fair Trade Commission The Executi ve Yuan. He has also
been visiting Assistant Professor at the City University of New York; Researcher
at the M.LT. in the International Motor Vehicle Program; National Expert of the
Republic of China in the Asian Productivity Organization; National Expert in the
Pacific Economic Cooperation Committee; Director of the Board of Directors,
Asia Pacific Science and Technology Association; Adviser of the Ministry of
Economic Affairs of the Council for Economic Planning and Development.
Ministry of Transportation and Communication.

Paul Westhead. He is Professor of Entrepreneurship in the Institute for Enterprise

Development at the University of Nottingham. He was formerly Professor of
Entrepreneurship at the University of Stirling after being a research fellow at
Cranfield University, Imperial College and Warwick University. He has
conducted numerous research projects sponsored by agencies such as the
Department for Trade and Industry, the Department for Education and
Employment, the Leverhulme Trust, the Confederation of British Industry, L10yds
Bank, NatWest Bank, KPMG Peat Marwick, PricewaterhouseCoopers, Shell UK
Limited and the United Kingdom Science Park Association. His research interests
include habitual entrepreneurs, technology-based firms, Science Parks and the
internationalisation of small firms. He has published over 60 refereed journal
articles. In collaboration with Professor Mike Wright, he recently edited
"Advances in Entrepreneurship: 3 Volumes" (2000) published by Edward Elgar.
He is on the editorial boards of Entrepreneurship Theory and Practice,
 List of Authors 307

International Small Business Journal and the International Journal of

Entrepreneurial Behaviour and Research.
Contributions to Economics
Michele Ba"ellal Johannes Briicker/Hayo Herrmann
Leonardo B~cchetti (Eds.) (Eds.)
The Competitive Advantage Spatial Change and Interregional
of Industrial Districts Flows in the Integrating Europe
:WOO. ISBN 3-7908-1254-4 2001. ISBN 3-7908-13.+4-3

Frank Bohn Kirstin Hubrich

Monetary Union and Cointegration Analysis
Fiscal Stability in a German Monetary System
2000. ISBN 3-7908-12fio-S 200 I. ISBN 3-7908-1352-4

Jaime Behar Nico Heerink et al. (Eds.)

Cooperation and Competition in a Economic Policy
Common Market and Sustainahle Land Usc
2000. ISBN 3-7908-12S0-3 2001. ISBN 3-7908-1351-6
Friedel Bolle/Michael Carlherl': (Eds.)
Michael Malakellis
Advances in Bebavioral Eco~]()mics
Integrated Macro-Micro-iVIodelling 200 I. ISBN 3-7908-1358-3
Under Rational Expectations
2000. ISBN 3-7908-127'+-9 Volker Grossmann
Inequality, Economic Growth, and
Slefan Baumgiirlner Technological Change
Ambivalent .loint Production and 200 I. ISBN 3-7908-1364-8
the Natural Environment
2000. ISBN 3-7908-1290-0 Thomas Riechmann
Learning in Economics
Henri Capron, Wim Mccusen (Eds.) 200 I. ISBN 3-7908-1J8'+-2
The National Innovation System
of Belgium Miriam Behlo
2000. ISBN 3-790S-130S-7 Bargaining over Time Allocation
2001. ISBN 3-7908-1391-5
Tohias Miarka
Financial Intermediation and Peter Mcushurl':cr/Hcikc Jiins (Eds.)
Deregulation Transtilrmatiiins in Hungary
2()OO. ISBN 3-7908-1307-9 2001. ISBN 3-79()8-1'+12-1

Chisalo Yoshida Claus Brand

Illegal Immigration and Economic Money Stock Control and Inflation
Welfare Targeting in Germany
2000. ISBN 3-7908-1315-X 200 I. ISBN 3-7908-1393-1
Erik LOth
Nikolaus Thumm
Private Intergenerational Transfers
Intellectual Property Rights and Population Aging
2000. ISBN 3-7908-1329-X
2001. ISBN 3-790S-1402-4
Max Keilhach Nicole Pohl
Spatial Knowledge Spillovers and Mobility in Space and Time
the Dynamics of Agglomeration and 2001. ISBN 3-7908-1380-X
Regional Growth
2000. ISBN 3-7908-1321-4 Pablo Coto-Millan (Ed.)
Essays on Microeconomics
Alexander Karmann (Ed.) and Industrial Organisation
Financial Structure and Stability 2002. ISBN 3-7908-1390-7
2000. ISBN 3-7908-1332-X
Ludwig Schatzl/Javier Revilla Diez
Joos P. A. van VugtlJan M. Peet (Eds.) (Eds.)
Social Security and Solidarity in tbe Technological Change and Regional
European Union Development in Europe
2000. ISBN 3-7908-1334-0 2002. ISBN 3-790S-14fiO-1