Industrial Agglomeration And New Technologies: A Global Perspective (New Horizons in Regional Science)

Industrial Agglomeration and New Technologies NEW HORIZONS IN REGIONAL SCIENCE Series Editor: Philip McCann, Professo...

Author: Masatsugu Tsuji | Emanuele Giovannetti | Mitsuhiro Kagami

135 downloads 2195 Views 4MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Industrial Agglomeration and New Technologies

NEW HORIZONS IN REGIONAL SCIENCE Series Editor: Philip McCann, Professor of Economics, University of Waikato, New Zealand and Professor of Urban and Regional Economics, University of Reading, UK Regional science analyses important issues surrounding the growth and development of urban and regional systems and is emerging as a major social science discipline. This series provides an invaluable forum for the publication of high quality scholarly work on urban and regional studies, industrial location economics, transport systems, economic geography and networks. New Horizons in Regional Science aims to publish the best work by economists, geographers, urban and regional planners and other researchers from throughout the world. It is intended to serve a wide readership including academics, students and policymakers. Titles in the series include: Industrial Clusters and Inter-Firm Networks Edited by Charlie Karlsson, Börje Johansson and Roger R. Stough Regions, Land Consumption and Sustainable Growth Assessing the Impact of the Public and Private Sectors Edited by Oedzge Atzema, Piet Rietveld and Daniel Shefer Spatial Dynamics, Networks and Modelling Edited by Aura Reggiani and Peter Nijkamp Entrepreneurship, Investment and Spatial Dynamics Lessons and Implications for an Enlarged EU Edited by Peter Nijkamp, Ronald L. Moomaw and Iulia Traistaru-Siedschlag Regional Climate Change and Variability Impacts and Responses Edited by Matthias Ruth, Kieran Donaghy and Paul Kirshen Industrial Agglomeration and New Technologies A Global Perspective Edited by Masatsugu Tsuji, Emanuele Giovannetti and Mitsuhiro Kagami

Industrial Agglomeration and New Technologies A Global Perspective Edited by

Masatsugu Tsuji Professor of Economics, Graduate School of Applied Informatics, University of Hyogo and Professor Emeritus, Osaka Unversity, Japan

Emanuele Giovannetti Associate Professor of Economics, University of Verona, Italy and University of Cape Town, South Africa

Mitsuhiro Kagami Ambassador for Nicaragua, Embassy of Japan, Nicaragua

NEW HORIZONS IN REGIONAL SCIENCE

Edward Elgar Cheltenham, UK • Northampton, MA, USA

© Masatsugu Tsuji, Emanuele Giovannetti and Mitsuhiro Kagami, 2007 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical or photocopying, recording, or otherwise without the prior permission of the publisher. Published by Edward Elgar Publishing Limited Glensanda House Montpellier Parade Cheltenham Glos GL50 1UA UK Edward Elgar Publishing, Inc. William Pratt House 9 Dewey Court Northampton Massachusetts 01060 USA

A catalogue record for this book is available from the British Library Library of Congress Cataloguing in Publication Data Industrial agglomeration and new technologies: a global perspective/edited by Masatsugu Tsuji, Emanuele Giovannetti, Mitsuhiro Kagami. p. cm. — (New horizons in regional science) Includes bibliographical references and index. 1. Economic geography—Research—Cases studies. 2. Commercial geography—Research—Case studies. I. Tsuji, Masatsugu. II. Giovannetti, Emanuele. III. Kagami, Mitsuhiro. IV. Series. HF1025.I54 2006 338.8’7—dc22 2006011735

ISBN 978 1 84542 396 4 Printed and bound in Great Britain by MPG Books Ltd, Bodmin, Cornwall

Contents vii ix

List of contributors Preface 1

Introduction Masatsugu Tsuji, Emanuele Giovannetti and Mitsuhiro Kagami

PART I 2

1

AGGLOMERATION IN ASIA

The relationship between Toyota and its parts suppliers in the age of information and globalization: concentration versus dispersion Masatsugu Tsuji

9

3

Iron town cluster: Yawata, its glory, decline and rebirth Mitsuhiro Kagami

4

Information technology and economic growth: discovering the informational role of density Takuo Imagawa

63

Agglomeration of exporting ﬁrms in industrial zones in northern Vietnam: players and institutions Akifumi Kuchiki

97

5

6

7

Industrial agglomeration and regional growth in Korea: focusing on the software and IT service sector Yasushi Ueki

139

China’s regional industrial disparity from the viewpoint of industrial agglomeration Koichiro Kimura

173

PART II 8

34

AGGLOMERATION IN ITALY

Italian comparative advantages, persistence and change in overall specialization Luca De Benedictis v

205

vi

Contents

9

Globalization, industrial districts and value chains Roberta Rabellotti

10

The competitive advantage of a region: industrial districts in Emilia-Romagna Enrico Santarelli

11

Where is the Internet? Agglomeration in space and cyberspace Emanuele Giovannetti, Karsten Neuhoﬀ and Giancarlo Spagnolo

PART III 12

13

225

247

268

AGGLOMERATION IN THE AMERICAS

The software industry in North America: human capital, international migration and foreign trade Andrew Schrank

289

Mexico: the management revolution and the emergence of the software industry Clemente Ruiz Durán

312

PART IV CONCLUSION 14

Conclusions Masatsugu Tsuji, Mitsuhiro Kagami and Emanuele Giovannetti

Index

353

359

Contributors Luca De Benedictis Macerata, Italy.

Professor of International Economics, University of

Emanuele Giovannetti Associate Professor of Economics, University of Verona, Italy and University of Cape Town, South Africa. Takuo Imagawa Director of Fair Competition Promotion Oﬃce, Ministry of Internal Aﬀairs and Communications, Japan. Mitsuhiro Kagami Nicaragua.

Ambassador for Nicaragua, Embassy of Japan,

Koichiro Kimura Research Fellow, Institute of Developing Economies, Japan External Trade Organization (IDE-JETRO), Japan and Visiting Research Fellow, Institute of Industrial Economics, Chinese Academy of Social Sciences, China. Akifumi Kuchiki Executive Vice President, Japan External Trade Organization (JETRO), Japan. Karsten Neuhoﬀ Department of Economics, University of Cambridge, UK and German Institute for Economic Research, Germany. Roberta Rabellotti Orientale, Italy.

Professor of Economics, University of Piemonte

Clemente Ruiz Durán Professor of Economics, Graduate School of Economics, National Autonomous University of Mexico, Mexico. Enrico Santarelli Professor, Department of Economics, University of Bologna, Italy. Andrew Schrank Assistant Professor, Department of University of New Mexico, USA.

Sociology,

Giancarlo Spagnolo Professor of Economics, University of Rome ‘Tor Vergata’, Italy, Stockholm School of Economics, Sweden and Centre for Economic Policy Research (CEPR), USA.

vii

viii

Contributors

Masatsugu Tsuji Professor of Economics, Graduate School of Applied Informatics, University of Hyogo and Professor Emeritus, Osaka University, Japan. Yasushi Ueki Research Fellow, Institute of Developing Economies, Japan External Trade Organization (IDE-JETRO), Japan.

Preface The question of where economic activity takes place is fundamentally related to the basic structure of society. Whether a region becomes an agricultural backwater, a bustling manufacturing hub, or an innovative hightech center will play a large role in determining the economic choices available to its inhabitants and the inﬂuence wielded by its government. As Asia and Latin America develop and developed countries ﬁght to maintain their edge in a globalized, informatized world, the notion of industrial agglomeration or clustering has increasingly been regarded as a source of long-term economic growth for regions and for entire economies. The question of why ﬁrms choose to locate in a particular region, given the kaleidoscope of location choices available to them, has thus become somewhat of a holy grail. Fortunately, ﬁne theoretical work continues to be done in this area. But theory must be complemented by evidence; no policymaker in his/her right mind would pursue an agglomeration strategy without being able to examine detailed evidence of how clusters have succeeded (and failed) in the past. Thus we, the authors of the chapters of this book, have directed our eﬀorts in that direction for several years now. This volume represents a compilation of the results of those case studies. Though these speciﬁc studies do not provide a ﬁnal answer to the microeconomic question of why ﬁrms gather in a region, they do provide key insights into the cluster formation process that will be essential in the future crafting of such a theory. The book’s chapters are naturally grouped by the region in which each study was performed: East Asia, Europe, or the Americas. The studies examine a broad cross-section of clusters: developed and developing countries, manufacturing and service industries. We tilt the balance of our attention slightly toward IT-related industries, both because these industries are relatively new (and thus present the greatest opportunity for new cluster formation) and because IT itself is a crucial element in the rapidly evolving global economy. Although no uniﬁed theory of agglomeration emerges from the chapters, several important themes do arise which we feel are inseparably tied to the clustering phenomenon. These include the importance of multinational corporations (MNCs) and their supply chains; the role of governments in creating infrastructure; regional specialization as a response to global ix

x

Preface

competition; and the essential role of R&D and innovative activities. These factors are all important elements of the increasing returns to scale that any cluster must generate in order to grow and thrive. We intend this book to be a road map for policymakers seeking to craft cluster-related policies, for business leaders choosing where to locate their operations, and for economic theorists attempting to explain the clustering process. Though we might not yet have found the holy grail, we have certainly narrowed down the space in which it may be found. This book received crucial support from a number of people and institutions. The book owes most of its results to a research project entitled ‘Supply Chain, Industrial Location, and Agglomeration in Knowledgebased Society’, which was organized and conducted by the Institute of Developing Economies (IDE)/Japan External Trade Organization (JETRO) in 2002–2003. We would like to express our deepest gratitude to Professor Masahisa Fujita, President of IDE, and Mr Osamu Watanabe, Chairman of JETRO, for their support and for their permission for publication. We are also deeply indebted to Professor Phillip McCann, University of Waikato and University of Reading, the editor of the series ‘New Horizons in Regional Science’. Professor McCann read our entire manuscript, provided many useful comments and suggestions, and approved the addition of this volume to his excellent series. Thanks are also due to the following people who contributed greatly to the improvement of this volume and to whom we thus owe a great deal: Dr Tetsushi Sonobe (Foundation for Advanced Studies on International Development), Dr Koji Nishikimi (IDE-JETRO), Michael Piore (Massachusetts Institute of Technology), Carlos Maroto (Director of the Asociación Mexicana de la Industria de las Tecnologías de Información), Eugenio Godard and Braulio Laveaga (President and Director of the Camara Nacional de la Industria Electrónica y de las Tecnologías de la Información), Luis Fernando Flores (Innovatia Cluster in Aguascalientes), Kie Ono, Shigeru Togashi, Giancarlo Spagnolo, Massimo Tamberi, Danny Breznitz, Seán Ó Riain and Ricardo Zermeño, as well as all the commentators and participants of seminars and conferences related to this research. This book is truly a group eﬀort and each of these people deserves a piece of the credit. We also express our thanks to Mr Noah Smith and Mr John Gallagher for their excellent editing work. We also thank Ms Nep Elverd and Ms Caroline Cornish of Edward Elgar for their patience and encouragement throughout the long publication process – thanks for putting up with us. Masatsugu Tsuji Graduate School of Applied Informatics, University of Hyogo and Professor Emeritus, Osaka University, Japan

1.

Introduction Masatsugu Tsuji, Emanuele Giovannetti and Mitsuhiro Kagami

For several years, the editors of this book have been studying technological transfer, multinational corporations (MNCs), deregulation and globalization processes (see, for example, Kagami and Tsuji 2000) and information technology (IT) (Giovannetti et al. 2003). In our studies, questions naturally arose as to why MNCs choose some locations for their plants and oﬃces as opposed to others. Also, why do IT-related ﬁrms agglomerate in certain places even though the Internet supersedes distance? Examples of the former phenomenon are industrial agglomerations in certain Eastern European countries, coastal areas of China and some border cities of Mexico, while examples of the latter are Silicon Valley in California, Bit Valley in Tokyo and Bangalore in India. Increasingly, we came to the realization that the answers to these questions are absolutely essential to understanding, predicting and explaining economic growth in developing countries. Industrial ‘clusters’, as they are called by leading theorists, are often the drivers of regional and even national economic growth. As perhaps the most remarkable example, consider the IT cluster in Bangalore. Without the IT industry explosion concentrated mainly in this city, India’s economy would be growing at a slower pace, and would almost certainly not be commanding the international attention that it now does. Smaller nations are even more dependent on the economic output of one or two key regions. Thus, we decided to undertake a project of considerable magnitude – the observation and analysis of industrial agglomeration, taking into account recent changes in the global economy. Our eﬀort, if successful, will provide invaluable information to key individuals in developing countries. With a comprehensive understanding of the clustering process in a contemporary context, business leaders and entrepreneurs will be able to take maximum advantage of existing or emerging clusters; policy makers and business leaders will be able to deliberately promote the development of clusters in their nations and regions; and researchers will have a head start toward a deeper understanding of the various processes at work. 1

2

Industrial agglomeration and new technologies

This book represents the necessary ﬁrst step in that process – namely, gathering and describing relevant examples of industrial agglomeration. Although the development of a general theory is our ultimate aim, we believe that this goal is beyond the scope of a single book, especially since our understanding of agglomeration is as yet incomplete. What we seek to do here is present and explore in-depth anecdotal evidence. In future works, we shall attempt to generalize; however, this book should not be viewed simply as an installment in a series. By examining the well-researched examples provided in this book, readers should be able to learn by example, and make their own generalizations. In this volume, comparison is key. Several aspects of this book’s scope and methodology set it apart from previous studies of industrial clusters. To accomplish our goals, we decided to draw examples from all over the world – speciﬁcally, from the key regions of Asia, Europe and North America – and to cover a wide array of industries, from traditional heavy industry to more modern software-based industries. Although studies of agglomeration and clustering are numerous, this book is one of the ﬁrst large-scale surveys to cover many examples and many regions. It is also one of the ﬁrst surveys to focus on developing nations, including studies from China, Mexico and Vietnam. We also decided to direct much of our attention toward IT, and its eﬀect on old industries as well as its creation of new ones. Again, this book’s IT focus separates it from most earlier clustering studies. The basic question is: why do ﬁrms gather, or agglomerate, in a certain region? Intuitively, MNCs will tend to construct their factories near large markets such as Europe, East Asia (including Japan), and the USA in order to save transportation costs. In addition, chosen locations can usually provide skilled and quality labor (information spillovers) at relatively low cost as well as copious funds supplied through various ﬁnancial institutions and capital markets. In the case of software clusters, the existence of a pool of computer-literate laborers is essential. However, this reasoning does not fully answer the question of why a special location was chosen among many possible candidates. These questions are fundamentally related to main themes of economic geography, or spatial economics, such as why certain cities grow as opposed to others, or why certain cities decline. There are some factors which ignite regional growth, and the process seems to be self-reinforcing or self-organizing once it starts. However, when other factors begin to take eﬀect, a cluster will shrink or decay. This cycle of concentration and dissolution (or dispersion) is thus the product of ‘centripetal’ forces working against ‘centrifugal’ ones. A standard textbook expresses this as follows: ‘The spatial structure of an economy is the result of a tug-of-war between external economies and diseconomies, between the linkages and

Introduction

3

information spillovers that foster concentration, and between congestion and other diseconomies that discourage it’ (Fujita et al. 1999, p. 349). The same authors also summarized that the centripetal forces are the Marshallian trinity of external economies: linkages, thick markets and knowledge spillovers (research and development: R&D), as well as other pure external economies, while the centrifugal forces opposing agglomeration include immobile factors, land rent/commuting, and congestion and other pure diseconomies. One key aspect of our survey was to be its longitudinal nature. By mixing in studies of old, established clusters with examples of newer, up-andcoming ones, we hoped to paint a more complete picture of a cluster’s life cycle. Old clusters can often experience a surprising rebirth after a long period of decline, sometimes spurred by new technology such as IT. This, we hope, will provide readers with a glimpse into the future, as well as allowing researchers to single out the centripetal and centrifugal forces involved in clustering. Another important concept regarding clustering is increasing returns, which are diﬀerent from the classical models of perfect competition and constant returns. Krugman (1995) wrote: ‘Increasing returns in production activities are needed if we want to explain economic agglomerations without appealing to the attributes of physical geography. In particular, the trade-oﬀ between increasing returns in production and transportation costs is central to the understanding of the geography of economic activities’ (cited by Fujita and Thisse 2002, p. 7). If scale merits work, we must use diﬀerent sets of analytical instruments, such as imperfect competition and monopolistic competition frameworks, combined with dynamic aspects. In this volume, the terms ‘agglomeration’ and ‘cluster’ are used interchangeably. The latter term, however, additionally implies the presence in a particular area of knowledge spillovers among ﬁrms or persons in the area. ‘Clustering’, in the formal sense, is therefore a new concept that focuses on the external eﬀects of information ﬂows. Recently, these information ﬂows have become increasingly important to industrial agglomerations. For example, countries such as China, India and Thailand were once widely considered to be relatively simple production bases for MNCs attempting to exploit cheap natural and human resources. In some agglomerations in those countries, however, R&D facilities or institutions have been established by MNCs or R&D, and innovation processes have developed endogenously (Kuchiki and Tsuji 2005). These agglomerations have thus been elevated to a ‘higher’ stage, at which time they may be properly referred to as ‘clusters’. However, because this book’s primary goal is to answer the fundamental question ‘Why do ﬁrms agglomerate?’, the word ‘agglomerate’ is placed in the title in lieu of ‘cluster’.

4

Industrial agglomeration and new technologies

Industrial clusters include small-, medium- and large-scale ﬁrms. Interactions among these, in terms of economic activities, are quite frequent and close. Famous examples of industrial cluster, in Porter’s (1990) terminology, are printing equipment in Germany and robotics in Japan. The term ‘industrial districts’, in contrast, is used in the original Marshallian sense, particularly for Italian cases where small-scale ﬁrms mainly gather to form an industrial town with a particular socio-cultural ﬂavor. More precisely, industrial districts are characterized as (a) local systems of active integration between a community of people and a community of industrial ﬁrms, and (b) a ﬂexible specialization characterized by the widespread presence of small-sized ﬁrms. Until recently, industrial clusters could be explained successfully by the various existing theories mentioned above. The contemporary world economy, however, has undergone a new evolution, with transformations such as globalization, emerging of developing countries, demand shifts and rapid technological changes such as the Internet revolution. Industrial clusters in advanced countries have therefore had to adjust to these new shocks and survive under new conditions (see for example, Small- and MediumScale Enterprise Agency 1994, 1996 and 1997). Industrial clusters in developing countries, as we mentioned, may serve as an engine of growth for the nation as a whole. The role of recent software industry clusters in certain developing countries, for example, has become important to these nations’ national economies, in terms of employment, income and exports. These developments require new insights in the analysis of industrial clusters, since existing theories cannot explain all the forces at work. This volume thus intends to examine examples that will eventually allow us to identify, understand and model these new forces. To accomplish our research goals, we formed three research teams, in Japan, Italy and the USA. These teams comprised respected researchers who had done outstanding work studying industrial clusters and the IT industry in their respective regions. After signing on to our project, the teams were charged with the task of examining recent IT clusters including software industries, as well as traditional industries such as iron/steel and automotive industries. The Japanese team studied the Toyota Motor Corporation; an iron and steel city; urban–rural relationships in the Internet era; industrial estates in Vietnam; Korean regional growth with IT-related industries; bridging of activities between industrial clusters (the Region-to-Region Initiatives Program by Japan External Trade Organization); and China’s regional industrial disparity. Italy was chosen due to the active role of its small and medium-sized enterprises and its unique but vivid industrial districts. The Italian team analysed

Introduction

5

peculiar patterns of Italian specialization and comparative advantage; involvement of some traditional firms in industrial districts in globalization (subcontracting production processes in the global value chain); the recent evolution of the famous ‘Third Italy’, i.e., the Emilia-Romagna region; and clustering forces at work in the Internet industry in spite of the weakening of traditional centripetal forces due to globalization and IT. The US team explored the US and Mexican software industries. The well-known software clusters such as Silicon Valley and Route 128 are now being challenged by oﬀshore programming platforms like those in India and Ireland. Four possible future scenarios for this industry were provided, taking into account the advantages of US production and proximity. The research indicated that the most likely scenario would be ‘nearshore’ operations in the same time zones as the US – such as Central American and Caribbean countries – which have gradually developed worker skills and IT-related infrastructure. In particular, the Mexican software industry has grown to be a threat to the US traditional software clusters. Overall, we found that, although industrial clusters can be divided into several distinct types, the essential features common to clusters are present in all the types. These common features include the presence of the allimportant knowledge spillover eﬀect, which creates positive externalities for ﬁrms; the ‘collective eﬃciency’ of collaboration between ﬁrms; the presence of increasing returns to scale; the value of face-to-face contacts among skilled workers and the exchange of ‘tacit knowledge’; the advantage of stable demand and the destabilizing eﬀect of demand shifts; the ‘centrifugal forces’ of transportation and communication costs; and the importance of government policy for small and medium-sized enterprises in clusters. This last feature – the importance of government policy – is especially important, as we have discovered that some countries’ rapid development has been due in large part to the intentional planning of ‘industrial parks’ and other special zones designed to attract foreign direct investment. This ﬁnding is particularly important for government oﬃcials seeking to harness the power of clustering to jump-start their nations’ economies. We sincerely hope that readers, using this volume, will be better equipped to understand contemporary patterns of spatial economics under globalization and the rapid progress of the Internet revolution.

REFERENCES Fujita, Masahisa and Jacques-François Thisse (2002), Economics of Agglomeration: Cities, Industrial Location, and Regional Growth, Cambridge: Cambridge University Press.

6

Industrial agglomeration and new technologies

Fujita, Masahisa, Paul Krugman and Anthony J. Venables (1999), The Spatial Economy: Cities, Regions, and International Trade, Cambridge, MA: MIT Press. Giovannetti, Emanuele, Mitsuhiro Kagami and Masatsugu Tsuji (eds) (2003), The Internet Revolution: A Global Perspective, Cambridge: Cambridge University Press. Kagami, Mitsuhiro and Masatsugu Tsuji (eds) (2000), Privatization, Deregulation, and Economic Eﬃciency: A Comparative Analysis of Asia, Europe, and the Americas, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Krugman, Paul (1995), Development, Geography, and Economic Theory, Cambridge, MA: MIT Press. Kuchiki, Akefumi and Masatsugu Tsuji (eds) (2005), Industrial Clusters in Asia: Analyses of their Competition and Cooperation, London: Palgrave Macmillan. Porter, Michael (1990), The Competitive Advantage of Nations, Basingstoke: Macmillan. Small- and Medium-scale Enterprise Agency (1994), ‘White Paper on Small- and Medium-scale Industries’ (in Japanese), Ministry of International Trade and Industry, Japan. Small- and Medium-scale Enterprise Agency (1996), ‘White Paper on Smalland Medium-scale Industries’ (in Japanese), Ministry of International Trade and Industry, Japan. Small- and Medium-scale Enterprise Agency (1997), ‘White Paper on Smalland Medium-scale Industries’ (in Japanese), Ministry of International Trade and Industry, Japan.

PART I

Agglomeration in Asia

2. The relationship between Toyota and its parts suppliers in the age of information and globalization: concentration versus dispersion Masatsugu Tsuji* 1.

INTRODUCTION

The automobile industry, a main force in economic development during the twentieth century, is examined here, with a focus on the economic basis of the Toyota production structure. The automobile industry has many related industries, such as automotive parts and machine tools, and also has the most advanced social division of labor. In particular, it is said that 30 000 to 40 000 parts are required to produce one automobile and this means that a vast number of ﬁrms are involved to some degree. Therefore, it is essential for the automobile assembler to integrate and eﬃciently manage those manufacturers who supply its parts. On this point, the Japanese automobile assemblers have established their own world-renowned production system. An analysis of how their method of production reached such a level of eﬃciency and how this industry developed interdependently with other ﬁrms, especially small- and medium-sized ﬁrms, will be made in this chapter. We also attempt to explain how the Toyota production structure is based on Japanese production culture. In the light of this analysis, the heavy concentration or localization of the automobile industry in Aichi Prefecture, particularly the Nishi-Mikawa region, will be discussed. Since the 1990s, the Japanese economy has undergone a transformation resulting from the development of an information society and globalization, and this chapter analyses how Toyota and its parts suppliers have been coping with these changes by making use of IT and international collaboration, so as to maintain vitality.

9

10

Agglomeration in Asia

2. PRODUCTION SYSTEM OF THE AUTOMOBILE INDUSTRY The production systems of Japanese automobile manufacturers are fundamentally the same. The general features of the system are examined, with a subsequent examination of Toyota’s in particular. Automobile manufacturers such as Honda, Nissan and Toyota assemble parts only, and do not produce all the required parts themselves. In contrast to American automobile manufacturers which produce about 60 per cent of their parts themselves, Japanese assemblers produce only 40 per cent within their own ﬁrms. This implies that most of their parts are supplied by related ﬁrms. Another aspect of the Japanese automobile industry is that the ﬁrms that are connected to the manufacturing of automobiles, such as those for automotive parts, are organized in a hierarchical structure. The ﬂow of automobile parts and the hierarchical structure of ﬁrms engaged in the production of automobiles are as shown in Figure 2.1. The ﬁgure illustrates how related ﬁrms of the automobile industry can be classiﬁed as follows: (a) primary parts manufacturers (primary subcontractors); (b) secondary parts manufacturers (secondary subcontractors); and (c) tertiary parts manufacturers (tertiary subcontractors). Primary Parts Manufacturers (Primary Subcontractors) The ﬁrms in this category supply parts directly to automobile assemblers. They comprise the following three types of ﬁrms: 1.

2.

Related parts manufacturers These ﬁrms supply complete parts such as tires, batteries or glass to the automobile assemblers. Such ﬁrms are as large as the automobile assemblers. The ratio of their products supplied to the assemblers is, generally speaking, less than 50 per cent. They are not dominated or controlled by the assemblers, and are therefore sometimes omitted from the category of automotive parts manufacturers, Specialized parts manufacturers Firms in this category are also primary parts manufacturers, and they supply complete parts such as pistons, clutches, brakes and shock absorbers, which are often called ‘primary parts’. They are characterized by specializing in the production of automotive parts, including high-tech parts such as those used in advanced electronics. They are medium to large in size, and they are at the core of Japanese automotive parts manufacturers. There are two classes among them; namely, subsidiary and independent. The former supplies only certain automobile assemblers, and the latter deals with non-speciﬁc assemblers. Since it is crucial to the

The relationship between Toyota and its parts suppliers

Automobile (complete)

(i) Primary parts manufacturers (primary subcontractors)

(a) Related parts manufacturers Related parts Tires, bearings

(b) Specialized parts manufacturers (c) Subcontractors

Primary parts (complete) Casting, forging, pressing

Pistons, clutches, brakes, transmission cases, instrument panels

Machine work, Simple plating parts

(ii) Secondary parts manufacturers (secondary subcontractors) Secondary parts Subcontracting work

Casting, forging, pressing

Machine work, plating

Secondary parts (complete)

General parts

Simple parts clutches, cylinders, radiators, brake linings, thermostats

Screws, cloths, cogwheels, springs

(iii) Tertiary parts manufacturers (tertiary subcontractors)

Tertiary parts subcontracting work Casting, forging, pressing, machine work, plating

Figure 2.1

11

Hierarchical structure of the automobile industry

12

3.

Agglomeration in Asia

quality of their automobiles that assemblers have subsidiary parts suppliers of superior technical ability, assemblers maintain close ties with them through the ownership of stocks or an interlocking directorate. This is accomplished through the synchronization of the production process, as well as joint R&D activities. Their ﬁrms thus expanded as the automobile assemblers grew. Some have become enterprises on a global scale in their speciﬁc ﬁeld of parts or level of technology. Subcontractors These supply simple and standard parts directly to the assemblers. They are engaged in labor-intensive processing, such as casting, forging, plating and pressing. They are relatively small in size. Firms which supply their products directly to the assemblers as primary parts suppliers or primary subcontractors are thus deﬁned. Since the most important ﬁrms in this category are specialized parts manufacturers, they are primary parts manufacturers.

Secondary Parts Manufacturers (Secondary Subcontractors) Manufacturers of this type supply secondary parts such as clutches, cylinders, brake linings and thermostats, to primary parts manufacturers. They supply these parts in the form of a unit. They also produce screws, cogwheels and springs. The ﬁrms are generally medium to small in size. Secondary parts manufacturers have strong ties with primary parts manufacturers. Although they are positioned below the primary parts manufacturers, secondary parts manufacturers stand at the top of their own hierarchical structure. Tertiary Parts Manufacturers (Tertiary Subcontractors) Firms in this category are of small size, and depend on family labor. As subcontractors of secondary parts manufacturers, their main business is of a processing nature such as that of casting and forging. These are labor intensive; thus their productivity is also low. The number of ﬁrms in this category is the largest. It is said that they are competitive in the market for obtaining orders from secondary parts manufacturers. Summary The Japanese automobile industry has a hierarchical structure with the assemblers positioned at the top, and the primary and secondary subcontractors below, on down to the very bottom where the tertiary parts manufacturers are situated. The higher the hierarchical structure, the stronger the connection between the contractors and the subcontractors in terms of

The relationship between Toyota and its parts suppliers

13

technology, equity and directorate. In other words, the complementary relationship between these two becomes even stronger and such a relationship tends to include a long-term, implicit contract. On the other hand, the lower the position in the hierarchy, the stronger is its relation through the market mechanism. In order to analyse the Japanese automobile industry, it is also important to grasp the qualitative diﬀerence between the upper and lower realms of the hierarchy.

3.

PRODUCTION STRUCTURE OF TOYOTA

Let us now examine the Toyota Motor Corporation in the context of Section 2. First, the primary parts manufacturers which have a direct relationship with Toyota will be examined. They can roughly be classiﬁed into two types. The ﬁrst type includes the 14 ﬁrms which belong to the ‘Toyota Group’: Toyota Tsusho, Aichi Seiko, Toyota Koki, Toyota Boshoku, Toyota Gosei, Toyota Shatai, Aishin Seiki, Nippon Denso, Toyota Jido Shokki, the Toyota Central Research Institute, Towa Real Estate, Daihatsu, Kanto Jidosha and Hino Motors. The ﬁrst 11 of these companies are similar in origin for they spun oﬀ from Toyota Jido Shokki. The last three were amalgamated by Toyota and joined the Toyota Group. Since they are assemblers of automobiles and their production plants are located in regions other than Tokai, they have been omitted from this analysis. The Toyota Central Research Institute and Towa Real Estate are not involved in the automobile industry; thus they have also been omitted here. Companies of the Toyota Group have very strong ties with Toyota, either traditionally or as a result of the synchronization of production known as the ‘Kanban method ’ initiated by Taicho Ohno (Ohno 1978) and also see Monden (1998), which will be explained in more detail below. Moreover, with regard to the interrelationship achieved through holding stocks and an interlocking directorate, it is important for Toyota to hold stocks of companies in the group or assign a director to a given company, though not necessarily vice versa. The companies also conduct joint research for the development of new products or technology. Another group of ﬁrms in this category are the ‘cooperative companies’. They form cooperative organizations such as Kyohokai and Kyoeikai, and have ties with Toyota that are as strong as those of the Toyota Group.1 Certain companies such as related parts manufacturers belong to Kyohokai and, as explained earlier, these ﬁrms are not necessarily Toyota’s subcontractors. There are 67 member companies, whose headquarters are located in Aichi Prefecture. Toyota owns 10 per cent or more of the shares of 41 of these companies. In order to synchronize the production process, they

14

Table 2.1

Agglomeration in Asia

Comparison of Toyota’s and GM’s systems

Toyota

GM

Low domestic production: 20%–25% 200 trade partners Toyota dominates parts suppliers Long-term commitment on quality and price Parts suppliers invest in specific equipment

High: 40%–50% Much larger Equal partner Market-based relationship General equipment

Source: Tsuji (2000).

successively adopted the Kanban method around the late 1970s. The size of the member ﬁrms varies from medium to large. Most of the member companies in Kyohokai also organize their own hierarchy of subcontractors and form cooperative systems. Toyota’s secondary parts manufacturers largely belong to this cooperative system. They are engaged in the production of simple parts or processing work. There are no recent data available, however, of the exact number of such secondary or tertiary parts manufacturers. The data provided by the ‘Survey of the Structure of Division of Labor’, conducted by the Medium and Small Business Agency in 1977, showed that at that time Toyota had 168 primary subcontractors, 5437 secondary subcontractors, and 41 703 tertiary subcontractors. This therefore implies that there was a total of approximately 36 000 (non-overlapping) subcontractors involved in the manufacturing of Toyota automobiles. On the other hand, in the United States the production structure of General Motors (GM) is non-hierarchical, and it has only 12 000 parts suppliers. This implies that the ratio of domestic production is much higher than that of Toyota. That is, in the United States, automobile assemblers produce a higher ratio of their parts within their own factories, and the number of parts supplied from outside ﬁrms is small. A comparison of the two systems is summarized in Table 2.1. As stated above, a large number of ﬁrms make up this social division of labor. We shall now examine the reason for the formation of this hierarchy of production.

4. HIERARCHICAL VERSUS NON-HIERARCHICAL SYSTEMS It is vitally important for assemblers to organize their huge number of subcontractors in a manner whereby overall production eﬃciency is achieved.

The relationship between Toyota and its parts suppliers

15

In the following subsection, a comparison will be made between the two production systems of the Japanese and US assemblers with regard to eﬃciency. The production system of the former is ‘hierarchical’, and that of the latter ‘non-hierarchical’. Economic Basis of the Hierarchical Production System The relationship of parts suppliers with Toyota can be explained by the long-term implicit contract (see Asanuma 1989, for example). When Toyota begins trading with a certain parts supplier, this implies that Toyota will make purchases from that parts supplier over an extended period. This long-term relationship also reduces transaction and information costs. Parts suppliers can invest in speciﬁc equipment for the sole production of Toyota parts. In addition to this, eﬃciency of the hierarchical production structure can be explained by the ‘principal–agent model’ – Toyota is the principal and the parts suppliers are the agents. It is not necessarily eﬃcient for Toyota itself to produce the various types of parts. It is more eﬃcient to hire certain ﬁrms as agents and to arrange a contract with them to set up production, since those ﬁrms have more information on manufacturingrelated parts than the principal. Subcontracting is commonly adopted in industries such as construction, since subcontracting improves the eﬃciency of a large organization. The Toyota production system is a multiplayer principal–agent relationship. Toyota is the single and ultimate principal of the whole system, but the primary parts suppliers are principals and the secondary parts suppliers are agents at the second stage; and secondary parts suppliers are the principal and tertiary parts suppliers are agents at the third stage. According to Coase (1937) and Williamson (1989), the optimal length of the stages in this context is determined by either the transaction costs or the information structure of the system to prevent opportunism and bounded rationality. Toyota can determine the optimal length of layers. Amount of Required Information Let us examine the merits and demerits of the hierarchical and nonhierarchical systems in more detail. From the viewpoint of contractors or assemblers, the amount of information necessary for the management of their subcontractors is less in the hierarchical than in the non-hierarchical system. The production system of the latter is so centralized that the assemblers require much more information, especially since there are a large number of parts suppliers. This increases the assemblers’ expenses for the management and organization of subcontractors. If there is any limitation

16

Agglomeration in Asia

of information ﬂow, the system fails to be organized eﬃciently. On the other hand, when the hierarchical system is highly decentralized, less information concerning the subcontractors is required. A disadvantage of the hierarchical system, however, is that since more agents take part in decision making, the assemblers’ wishes might not be successfully transmitted to the very bottom of the hierarchy. This implies that the general consistency of decision making may not be adequately maintained throughout the system. Pertaining to this point, the principal–agent model in economic theory can be applied. According to this theory, an optimal contract exists between the principal and the agent such that the best decision for the agent is also optimal to the principal. Thus, for the latter, it is preferable to refrain from direct management of the company and more beneﬁcial to select one ﬁrm as an agent and allow it to manage the other subcontractors. By so doing, the amount of information required by the principal decreases even further on the one hand, and eﬃciency is improved on the other. In the principal–agent theory lies the essence of how incentives should be given to the agent whose best decision is also considered to be best for the principal. In the case of Toyota, the incentive for subcontractors to work as an agent includes the following: (i) Toyota guarantees a certain amount of proﬁts to the agent (primary subcontractors). Price negotiations between Toyota and its parts manufacturers are recognized as being particularly stringent, but the margins of proﬁt for parts manufacturers are not subject to this negotiation and they have already been traditionally provided; (ii) there is a long-term implicit contract relationship (once an agent joins the system, the transaction is secure over a long period); (iii) there is a guarantee of growth (the agent can grow in step with Toyota); and (iv) there is a feeling of solidarity with Toyota. All of these are important factors of Japanese management. Production Eﬃciency Another characteristic of the hierarchical system of production is improved eﬃciency through the division of labor, since subcontractors specialize in their own speciﬁc production process by means of rationalization. An automobile consists of several tens of thousands of parts. There are a great number of simple and tiny parts which are required in small numbers. The assemblers decrease their cost of production by allowing subcontractors to produce these parts rather than doing so themselves. Due to diversiﬁcation of consumers’ taste in recent years, assemblers have had to produce an increased variety of cars in decreasing numbers for each type. This has resulted in increasing the number of parts even further.

The relationship between Toyota and its parts suppliers

17

Even if the eﬃciency of individual ﬁrms in the hierarchy is improved through specialization by the division of the production process, the eﬃciency of the entire system is not necessarily achieved. In the case of Toyota, it is the synchronization of production by the ‘just-in-time system’, or Kanban method, that makes this possible. Each subcontractor has to supply automotive parts of a certain quantity, at a certain time, and at a certain place, as has been decided beforehand by the assemblers. A single mistake by those suppliers would cause mass confusion in the production process. Each subcontractor is thus required to act in consideration of the entire system. Risk Sharing The Japanese industrial group is sometimes deﬁned by the diversiﬁcation of risk; that is, if various kinds of companies make up the industrial group, the total risk of the group is decreased. The same can be said of an industry with subcontractors, such as the automobile industry. Contractors can avoid investing in equipment if they consign production to subcontractors, or subcontractors can lessen their risk in long-term investment by making implicit contracts with the assemblers. Thus, both sides beneﬁt by lessening the risk factor. Growth Sharing Another reason why nearly 36 000 subcontractors support Toyota’s hierarchical production system is ‘growth sharing’, that is, when Toyota grows, the parts suppliers also grow. As a matter of fact, parts suppliers have grown in step with Toyota, and most Toyota Group ﬁrms are now global enterprises. Toyota, as illustrated above, has adopted a system of production making full use of the merits of the hierarchical structure. The core of its production system is found in the just-in-time system or Kanban method. This means that parts are supplied at a speciﬁc time and in a speciﬁed quantity. Moreover, this implies not only that the speed of each process in production at the assemblers’ factory has been synchronized to the speed of the assembly line, but also that the production process of all subcontractors situated even at the very bottom of the hierarchy has been synchronized. Furthermore, quality management (QM) or total quality management (TQM), which works to improve the quality of automobiles, is practiced systematically throughout the hierarchy from the assemblers at the top to the lowest level of subcontractors.2

18

Agglomeration in Asia

5. LOCALIZATION OF THE AUTOMOBILE INDUSTRY IN AICHI PREFECTURE Following the previous discussion on the general characteristics of the automobile industry, we shall now examine the localization or concentration of the automobile industry in Aichi Prefecture. Location of the Automobile Industry and its Concentration To begin with, we shall consider the location of the headquarters and factories of the Toyota Group and those of the major member companies of Kyohokai whose headquarters are located in Aichi Prefecture. Table 2.2 shows that most of the headquarters as well as the factories are located in Aichi Prefecture – speciﬁcally, for the most part they are located in the Nishi-Mikawa district, the eastern sector of the prefecture. During the bubble economy years, Toyota faced a severe labor shortage. Because of this, it established three assembling factories outside the Mikawa region for the ﬁrst time – at Hokkaido, Tohoku and Kyushu – which are all far from Toyota City. These plants started operation in 1992 and 1993. Other parts suppliers constructed their factories close to Toyota’s. Nearly 70 per cent of the parts are shipped to these factories from Nagoya Port. With regard to the location of factories of the member companies of Kyohokai whose headquarters are situated in Aichi Prefecture, 80 per cent are in the prefecture itself, and 55 per cent are located in the Nishi-Mikawa district. Moreover, nearly half are found in Toyota City. This contrasts greatly with the fact that factories of the Toyota Group are dispersed in the cities of Toyota and Kariya. The reason for this is that the members of Kyohokai are directly tied up with Toyota and have located close to its factories in Toyota City. Little can be said regarding secondary subcontractors due to insuﬃciency of data. However, from Table 2.3, it can be concluded that factories are located close to primary parts manufacturers. It is therefore apparent that there is a heavy concentration of factories connected with Toyota in Aichi Prefecture, and this is crucial to Toyota’s eﬃciency. The reason behind the largest concentration of automobile industries in the prefecture is path dependency, that is, Toyota based its headquarters there, and the Kanban method inﬂuenced subcontractors to locate close to Toyota plants so as to economize on money and the time it would take to deliver parts to Toyota factories. Another reason for Toyota’s decision to locate there is that the region has a long tradition of machinery for textiles and tools.

19

The relationship between Toyota and its parts suppliers

Table 2.2

Location of Toyota factories and its parts suppliers (firms)

Aichi Prefecture Owari region Nagoya Ohbu Inazawa Tokai Others Nishi-Mikawa region Toyota Kariya Anjo Nishio Hekinan Takahama Okazaki Miyoshi Koda Others Higashi-Mikawa region Gifu Prefecture Mie Prefecture Others

Toyota Motor Co.

Toyota Group

11

47 11 1 3 3 1 3 34 7 8 4 4 3 2 1 2 2 1 2 1 1

6

1

3

1

3

Factories of Kyohokai members (%) 116 (79.5) 33 (22.6) 14 (9.6) 1 (0.7)

18 (12.3) 80 (54.8) 33 (22.6) 5 (3.4) 7 (4.8) 5 (3.4) 3 (2.1) 4 6

(2.7) (4.1)

17 (11.6) 3 (3.4) 10 (6.8) 20 (13.7)

Note: Kanto Motor, Hino Motor, Daihatsu, Toyota Central Research Institute and Towa Real Estate have been omitted. The Toyota Group has been omitted from Kyohokai. Sources: Calculated from Japan Auto Parts Industries Association and Auto Trade Journal (2000) and IRC (2002).

Furthermore, the concentration of contractors and subcontractors of the assembling industry in one particular region led to even more eﬃciency through the Kanban method or joint investment in R&D.3 Moreover, even at the bottom of the hierarchy, or among the tertiary subcontractors, a specialized economy was created through outside orders for parts. It has been reported that even among factories employing less than ﬁve workers, which engaged in the manufacturing of simple parts or in simple tasks, an interdependence of production existed through outside orders.

20

6 6 1 5 – – – – – – – – – – – – 1 –

54 45 13 32 7 1 1 1 1 – – – 3 2 7 5 1 4

74

Toyota-Gosei Haruhi Kyowakai

3 – – 1 2 – 2 – – – – – – – – – – –

75 39 26 13 36 11 10 4 3 1 1 1 5 – 3 – 4 22

102

Toyota-Shatai Kariya Shatai Kyorokukai

2 1 – 1 1 – 1 – – – – – – – – – – –

74 52 20 32 22 4 7 1 2 1 – 1 6 – 3 4 – 22

129

Aichi-Seiko Tokai Hokokai

5 1 – 1 4 1 – – – – – – 3 – – – – –

34 30 12 18 4 4 – – – – – – 4 3 8 1 3 4

Tokai Rika Denki Ogichi Kyorokukai 52

Sources:

See Table 2.2.

Notes: a: name of primary contractor, b: location of headquarters, c: name of cooperative association, d: number of members. Numbers in bold are the number of factories of contractors, i.e., company in bold at the top of each column.

67 22 18 4 42 2 11 7 3 5 5 2 7 3 2 5 3 8

Aichi Prefecture Owari Nagoya Other Nishi-Mikawa Toyota Kariya Chiryu Anjo Nishio Hekinan Takahama Others Higashi-Mikawa Gifu Prefecture Mie Prefecture Shizuoka Prefecture Others

9 – – – 8 – 2 2 2 – – 1 1 1 – 1 – 1

84

Denso Kariya Hishoukai

d

a b c

Table 2.3 Location of factories of Toyota’s major secondary subcontractors (firms)

The relationship between Toyota and its parts suppliers

21

Growth of the Automotive Parts Industry in Aichi Prefecture We now focus on the Kyohokai parts manufacturers located in Aichi Prefecture, which began to supply parts to Toyota before the Second World War and have maintained a long and rewarding relationship with Toyota since then. Although the growth of these parts manufacturers is partially due to their special relationship with Toyota, it cannot be denied that it was accomplished through their mutual eﬀorts. During the early stages of the automobile industry, Toyota concentrated on nurturing the parts manufacturers; that is, it supported them in borrowing funds from banks which allowed them to purchase production equipment, provided instructions on new technology, and supplied materials. Later when they began to introduce the just-in-time system, Toyota sent specialists to make adjustments to their production system by redesigning the process and improving the assembly line. Moreover, they adopted QM and TQM, following the example of Toyota. While lending support to the parts manufacturers, Toyota also requested estimates for new parts from more than one parts manufacturer in order to decide which was best; this meant that competitive forces were at work among those manufacturers. When the price of parts is negotiated, instead of calling for bids, discussions based on the estimates are conducted until an agreement can be reached. Toyota does not force subcontractors to accept any particular price, and instead engages in discussions with them to the point where a price can be mutually agreed upon. After going through such a process, Toyota’s primary parts manufacturers have since achieved a fairly high level of technology of their own. They are now trying to diversify the assemblers so that their parts may be supplied by them and expand their business into new areas in preparation for the post-automobile society. Medium- and Small-sized Firms of the Automobile Industry Regarding the role of small- and medium-sized ﬁrms in Japanese industries, it suﬃces to say that the coexistence of large ﬁrms on one hand, and small- and medium-sized ﬁrms on the other, has been accepted as the dual characteristics of the Japanese economy. However, we shall focus here on the interdependence of these two types of ﬁrms in the context of the production structure hierarchy, especially the role of small- and medium-sized ﬁrms in Aichi Prefecture. It is said that there are two types of small- and medium-sized ﬁrms in the prefecture; namely, the ‘Owari group’ and the ‘Kariya group’. The former are those involved in the textile and machine tools industries located in the

22

Agglomeration in Asia

Owari district in the western sector of Aichi Prefecture. The latter are those of the automotive parts industry, particularly specializing in manufacturing parts. The biggest diﬀerence between them is that the former must diversify their business because the textile industry is in the midst of structural shifts, but the latter are enjoying the prosperity brought about by the automobile industry. Small- and medium-sized ﬁrms are mainly engaged in the production of simple parts or labor-intensive processes. They do not possess any specialized technology or management and are therefore strongly inﬂuenced by their contractors. They manufacture parts by using designs rented from contractors. In the case of Toyota subcontractors, a system of production, wages, training and even management similar to Toyota’s was adopted. It is said that their level of technology is superior to that of other subcontractors. To determine the price of parts, they are asked to submit their data on cost and even their ﬁnances, and it is then that the appropriate price is negotiated with the contractors. In addition, they are guaranteed a certain ratio of proﬁt margin, and wage levels are advantageous as they are tied up with Toyota. Small- and medium-sized ﬁrms in Aichi Prefecture have become the backbone of the regional economy because of the regional concentration of parts manufacturers; that is, the excellent management of small- and medium-sized ﬁrms originated in these ﬁrms themselves. Companies with a ‘ﬁrm backbone’ are founded on others of the same ilk. Nowadays, however, technological innovation is taking place so rapidly that it seems almost impossible for such ﬁrst-rate companies to be reproduced simply by the accumulation of small-sized ﬁrms. Now that automobile assemblers have began to shift their production overseas, small- and medium-sized ﬁrms should raise their technological level and develop their products in order to survive.

6. TOYOTA PARTS SUPPLIERS FACING GLOBALIZATION The Japanese economy has been faced with globalization. An increase in imports of foreign goods has forced local industries to face competition for the ﬁrst time. In addition, ﬁrms have been shifting towards production abroad, and Toyota and its parts suppliers are no exception. This section examines how globalization has aﬀected the location of Toyota and its parts suppliers.

The relationship between Toyota and its parts suppliers

23

Foreign Direct Investment by Toyota and its Parts Suppliers Motivation for foreign direct investment (FDI) is either natural resource or market oriented. The aim of the former is to make use of relatively inexpensive natural resources, and that of the latter is to promote the sale of products in the local markets.4 The automobile industry has chosen overseas locations according to the latter. In East Asia, including China, Asian newly industrialized economies (NIEs), and Association of South East Asian Nations (ASEAN) countries, the ratio of products sold in the local market has been increasing in accordance with the increase in income level. Recently, in addition to an abundance of natural resources, raising the technology level of the above regions has been intensively promoted. As a result, there has been a tendency for the regions to serve as the base of manufacturing for export to Japan or other countries; that is, the processing and assembling industry has been conforming widely to the network of the international division of labor in such a way that ﬁrms decide on the location of their plants by comprehensively taking the following points into account: namely, the cost of manufacturing parts, as well as that for assembling ﬁnal products, such as the exchange rate, wage rate, cost of materials, and level of technology in all international regions. Overseas production Toyota decided to commence production in the United States because there was trade friction involving Japanese automobiles in the US in the late 1970s and early 1980s. In 1982, Honda began manufacturing automobiles in the US, followed by Nissan, Toyota and others who began operating production plants there. In 1984, Toyota established NUMMI (New United Motor Manufacturing, Inc.) in California in the form of a joint venture with GM, and later started production in Kentucky (TMMK (Toyota Motor Manufacturing, Kentucky, Inc.)), and in Illinois (TMMI (Toyota Motor Manufacturing, Illinois, Inc.)) in 1998. In Europe, Toyota has factories in the UK (established 1992), France (2001) and Portugal (1968), and started production in the Czech Republic with Peugeot in 2005. In Asia, Toyota has plants in Thailand (1964), Malaysia (1968), Indonesia (1970), Taiwan (1986), the Philippines (1989), Vietnam (1996), India (1999) and so on. Toyota also announced start of production in China in 2002. Thus, by 2005 Toyota had 51 overseas plants in 26 countries. The recent trend of overseas production and sales are summarized in Figures 2.2 and 2.3. According to these ﬁgures, North America dominates the world market, but the increase in Asia is remarkable. In 2001, Toyota produced more than 40 million vehicles, which is the largest amount of overseas production. Thus, overseas production of Japanese automobiles is projected to expand

24

1993

1992

1998

1997

1996

1995

1994

Toyota’s overseas production (thousand vehicles)

Toyota Motor Corporation.

Figure 2.2

Source:

1200 1100 1000 900 800 700 600 500 400 300 200 100 0 2001

2000

1999

Asia

Africa

Europe

Latin America and the Caribbean

North America

25

1994

1993

1992

1997

1996

1995

Toyota’s overseas sales (thousand vehicles)

Toyota Motor Corporation.

Figure 2.3

Source:

0

200

400

600

800

1000

1200

1400

1600

1800

2000

2001

2000

1999

1998

Middle East and Southwest Asia

Oceania

Asia

Africa

Europe

Latin America and the Caribbean

North America

26

Agglomeration in Asia

greatly in future years as long as Toyota’s prevailing international competitiveness in the international economy continues. With an increase of overseas production, parts suppliers also established plants abroad. Table 2.4 indicates the number of plants of Toyota’s parts suppliers in North America. Local parts content As overseas production increases, so does the purchase of local parts by Japanese automobile manufacturers. In North American factories, nearly 50 per cent of the parts are purchased locally. An increase in this ﬁgure is always required in negotiations between Japan and the US, and Toyota has decided to produce engines and transmissions in Alabama. This will bring the ratio of local parts in the US to more than 50 per cent of the total. Toyota has been purchasing parts from overseas, which include among the major items tires, lamps, glass, catalysts, leather, carpet fabric and large electronic equipment. They are all of standard quality or diﬀer little from the quality of Japanese products, excluding large equipment. FDI by Parts Manufacturers As stated above, overseas production by Japanese auto manufacturers has continued to increase; therefore, automotive parts manufacturers have also intensiﬁed their activities in foreign countries. By the end of 2001, 304 automotive parts manufacturers in Aichi Prefecture were involved in foreign investment, in particular 86 in North America, 130 in Asia (for example, in Taiwan, China and Thailand) and 59 in Europe.5 The current situation of FDI in North America by automotive parts manufacturers located in Aichi Prefecture, is as shown in Table 2.4. It is reported that 12 manufacturers have invested in 17 plants. Their investments mainly aim to supply parts to the TMMK and TMMI plants. Therefore, they are located in neighboring states such as Kentucky, Illinois and Ohio. Because of the dispersion of these parts plants and the location of local parts manufacturers, the Kanban method has been practiced quite erratically by the collection of the necessary parts at the required time by trucks. Thus, production is not synchronized with the assembler as it is in Japan, so the assembler retains two days’ worth of parts in its inventory.6 The characteristics of parts manufactured in those plants are as follows: (i) the quality of Japanese parts is superior; (ii) there is a high freight fee compared to the price of parts such as air conditioners for automobiles, ﬂoor mats, weatherstrips and body parts; (iii) speciﬁcation of parts diﬀers from the American one; and (iv) parts produced by US automobile manufacturers for themselves are diﬃcult to purchase on the market.

27

Source:

1 5 1

Aichi Prefecture (2002).

Toyota Toyota Group Kyohokai

California 1 3 2

Indiana 1 6 6

Kentucky – 3 2

Missouri – – 2

Ohio – 1 1

Illinois

Table 2.4 Toyota plants and its parts suppliers in North America (as of 2001) (firms)

– 2 1

Tennessee

1 2 –

Canada

– 5 1

Others

28

Agglomeration in Asia

FDI and Dispersion of Location Now that we have reviewed foreign investment by parts manufacturers, we shall consider its eﬀect on the hierarchical production structure. First, in general terms, the strength of the connection between contractors and subcontractors in the processing and assembling industry must be taken into consideration. For assemblers, the merit of using subcontractors is to be able to purchase cheaper parts of high quality, and their interest lies in how they can obtain such parts. Basically, there is no marked diﬀerence whether they are made in Japan or not. This is the economic foundation of the international division of labor in the production process. The same can be said of automotive manufacturers. They can substitute domestic with foreign parts as long as the latter satisﬁes their requirements in price and quality. Then the issue is whether this expansion of the international division of labor will change the hierarchical production structure. One of the fundamental bases of the Japanese hierarchical structure is the long-term relationship between the contractors and subcontractors. The closeness of the relationship comes more from speciﬁc Japanese traits than for economic reasons. In the case of Toyota, the relationship between it and its subcontractors began prior to the Second World War and the historical cohesion of being a member of the Toyota Group seems to be even stronger today. On the other hand, decision making by ﬁrms is based on economic law. Contractors and subcontractors cannot continue their relationship without considering the beneﬁts. Whether they can do so depends upon the delicate balance of the centripetal and centrifugal forces in the hierarchy. Here is an example of how the relationship between Toyota and its parts suppliers began to change. Figure 2.5 shows the destinations of parts in 1988 and 2001, comparing various automobile assemblers that are typically supplied by Toyota’s parts suppliers. Toyota’s major primary parts suppliers, such as Aishin AW and Aisin Seiki, reduced their sales to Toyota. This was unimaginable ten years ago, since Toyota used to be reluctant to allow its parts suppliers to do business with other customers. Toyota was wary of a drain of technology and know-how. However, Toyota has since realized that allowing its suppliers to cultivate alternative customers allows those ﬁrms to secure suﬃcient quantities of production and sales, and to reduce costs by maintaining scale economies, both of which are ultimately beneﬁcial to Toyota itself. Toyota refers to this trend of supplying other assemblers as ‘wide extension’. From the current situation of the automobile and parts industries, the automobile industry is still expected to grow and the parts industry has not matured enough to establish its own production structure.

The relationship between Toyota and its parts suppliers

7.

29

CONCLUSION

The hierarchical production structure, which is crystallized in the form of the Japanese automobile industry, particularly in the Toyota production system, is as analysed here. One reason why this structure is widely found in other assembling and processing industries is that when these industries were ﬁrst introduced to Japan, the parts manufacturing industry had not yet been developed; therefore, it was the assemblers who fostered the growth of those industries. Despite such a history, in eﬀect, the Japanese hierarchical production structure (or Japanese management) promoted the growth of these industries into those of world-renowned stature within a short period of time, since this was based upon not only economic theory but also Japanese production culture. In addition to the hierarchical production structure which has these characteristics, the nature of the automobile such that it requires thousands of parts led the automobile industry, particularly the Toyota Group, to concentrate heavily in a particular, rather small region. The distribution of parts cannot be replaced by information technology (IT), but the supply chain management applied by IT can reduce the time and cost of shipment. In this sense, the just-in-time system is essential for the heavy concentration of Toyota parts suppliers in locations close to Toyota plants. Another rationale for the heavy concentration lies in the economies of scale. The Toyota Group as a whole can exploit scale economies by concentrating production within a limited number of companies. Since the 1990s, the transformation of the Japanese economy has resulted in IT and globalization penetrating into all Japanese industries, to the extent that the Japanese automotive industry can no longer exploit the other economies by concentration. The above transformation is quite new to the Japanese economy, and may not suit the Japanese economic system with its long-term contractual relationship among ﬁrms. IT, for instance, is a key technology for collecting real-time information, and it can reduce the time spent and costs involved in searching for the best partner in trade. In this sense, IT helps the market mechanism to work better. One example of this is ANX (Amerian Automobile Network Exchange), which interconnects all automobile assemblers and parts suppliers. This network functions like an e-marketplace, and helps to ﬁnd the best partner for trade instantly according to the price as well as quality desired. Toyota also tries to use IT in all segments of its activities, and the new Kanban method is referred to as ‘eKanban’. This is still at an experimental stage. Toyota is in the process of constructing ‘WARP’ (Worldwide Automotive Realtime Purchasing System), which is a global database for sharing information on the quality of parts, production plans and transactions related

30

Aishin AW Aishin Seiki Denso Toyota Gosei Futaba Industrial Takashimaya Nippatsu Tsuda Industries Sango Chuo Spring Hosei Brake Aisan Industry Arakawa Auto Body Kyosan Denki

82.0 64.0 55.4 58.7 46.4 81.2 53.7 66.5 46.4 83.0 66.6 98.6 15.1

1988

Toyota

64.1 11.0 – 55.1 55.0 81.4 – 75.4 40.0 86.7 77.0 – 8.0

2001 – – – – – – – – 1.9 – – – 0.8

1988

Nissan

– – – – – – – – 1.0 – – – 0.7

2001 – – – 4.2 4.3 – – 3.9 4.2 6.0 7.4 0.5 1.2

1988 0.3 – – 4.1 4.2 – 1.0 6.1 4.0 3.7 7.4 – –

2001

Daihatsu

Table 2.5 Automobile assemblers supplied by Toyota Group parts manufacturers (%)

– – – 1.3 – – – – 8.1 – 3.9 – 2.1

1988

Mazda

0.2 – – – – – – – 5.0 – – – 1.3

2001

31

Sources:

3.0 – – – 13.5 – – – 2.5 – 4.0 – 0.8

3.3 2.9 – – 7.4 – – – 3.0 – 2.1 – –

2001 – – – – 1.7 – – – 9.7 – – – –

1988

Honda

– – – – 8.5 – – – 11.8 – – – –

2001 – – – – – – – – 3.4 – 3.7 – –

1988

Japan Auto Parts Industries Association and Auto Trade Journal (2000) and IRC (2002).

Aishin AW Aishin Seiki Denso Toyota Gosei Futaba Industrial Takashimaya Nippatsu Tsuda Industries Sango Chuo Spring Hosei Brake Aisan Industry Arakawa Auto Body Kyosan Denki

1988

Mitsubishi

Suzuki

0.3 – – – 6.0 – 0.5 – – – – – –

2001

– – – 8.8 – – 33.1 18.3 8.0 – 7.1 – 69.5

1988

Others

– – – – – – – – – – – – –

2001

32

Agglomeration in Asia

to all aspects of the Toyota Motor Corporation. This system is intended to be available to all suppliers, including non-Toyota ones. Another new application of IT is ‘concurrent engineering’, which is the system of sharing information with diﬀerent sections of the automobile assembler. Chrysler made use of this information system in the development of Neon in the early 1990s by connecting all sections related to its development, especially the design, R&D and prototype sections. This made the time involved and the costs of development shorter and smaller. Its development period was approximately 31 months, which is said to be shorter than that of Toyota at that time.7 Toyota and its group companies still proceed with the development of new models in the traditional way. We analysed the globalization of Toyota and its parts suppliers. At the early stage of globalization in the 1980s, the general outlook was pessimistic, that is, it was believed that all Japanese automobile assemblers and parts suppliers would not be able to survive stiﬀ international competition. This expectation has partly been realized, since Nissan and Mazda both encountered diﬃculties and they restructured nearly one-third of their parts suppliers. On the other hand, Toyota still maintains its hierarchical structure almost intact, although some eﬀects, such as wide extension, were mentioned earlier. Toyota experienced ‘CCC21’ (Construction of Cost Competitiveness for the 21st Century), which aimed to reduce costs by 30 per cent for three years starting in 2001. This is based on an entirely diﬀerent philosophy in that the costs were required to be reduced by a certain percentage from the previous model. Whereas this traditional philosophy is based on relative price, CCC21 pays attention to absolute costs. That is, Toyota chooses 173 core parts and sets the absolute costs by considering those of global markets and parts suppliers, and the design, purchase and production sections are asked to collaborate from the early stage of development to the ﬁnal stage of production so as to meet the desired cost reduction. This also aims to protect its parts suppliers from global competitiveness.8

NOTES *

The author is indebted to the oﬃcials of the Aichi prefectural government, and to those of the Toyota Motor Corporation and its parts suppliers for their collaboration during the research. 1. Kyohokai consists of 211 parts supply companies (as of 2002), and there are also subgroups such as those for unit and body parts. The member companies of Kyoeikai mainly deal with tools and machine tools; it has four subgroups comprising body equipment, unit equipment, construction and distribution. The total number of member companies is 123 (as of 2002). 2. These were formerly referred to as quality control (QC) and total quality control (TQC), respectively.

The relationship between Toyota and its parts suppliers

33

3. Joint R&D activities with Toyota and other ﬁrms of the Toyota Group are no longer common. They tend to seek their own innovative management. In addition, sometimes they do not share the same aims and objectives. One such example can be found in the development of the car navigation system. Toyota was opposed to the R&D of Aishin A&W, since Toyota would have preferred Denso to proceed with R&D. However, Aishin A&W did not comply, and its R&D ultimately succeeded. 4. In North America, investment in the form of setting up oﬃces to promote sales used to be largest due to that country’s great purchasing power. In addition, investment in plants whose R&D is aimed at collecting information on innovation as well as new technology has been on the increase. 5. These ﬁgures are calculated from Aichi Prefecture (2002). 6. This is the same as the Honda plant in Guangzhou, China. 7. See Tsuji and Nishiwaki (1999b) and Tsuji (2003) for more detail. 8. See IRC (2002, pp. 170–71).

REFERENCES Aichi Prefecture (2002), ‘Overseas activities of Aichi ﬁrms’ (in Japanese), Aichi Prefectural Government, Nagoya. Asanuma, B. (1989), ‘Structure of transactions of Japanese industry: the case of the automobile industry’ (in Japanese), Economic Review (Kyoto University), 133 (3), 1–30. Coase, R.H. (1937), ‘The nature of the ﬁrm’, Economica, 4, 386–405. IRC (2002), Reality of the Toyota Group (in Japanese), Tokyo: IRC. Japan Auto Parts Industries Association and Auto Trade Journal (2000), The Japanese Automotive Parts Industry (in Japanese), Tokyo: Auto Trade Journal Co., Inc. Monden, Y. (1998), Toyota Production System: Integrated Approach to Just-inTime, 3rd edn, Norcross, GA: Engineering & Management Press. Ohno, T. (1978), Toyota Production System (in Japanese), Tokyo: Diamond Publication. Tsuji, M. (2000), ‘Envisioning the Japanese economic system in the 21st century in relation to economies of network’, in F. Schober, T. Kishida and Y. Arayama (eds), Restructuring the Economy of the 21st Century in Japan and Germany, Berlin: Duncker & Humblot, pp. 15–36. Tsuji, M. (2003), ‘Transformation of the Japanese system towards a network economy’, in E. Giovannetti, M. Kagami and M. Tsuji (eds), The Internet Revolution: A Global Perspective, Cambridge: Cambridge University Press, pp. 7–20. Tsuji, M. and T. Nishiwaki (1996), Nettowa-ku Nirai (Future of Networking) (in Japanese), Tokyo: Nihonhyoronsha. Williamson, O.E. (1989), ‘Transaction cost economics’, in R. Schmalensee and R. Willig (eds), Handbook of Industrial Organization, Vol. 1, Amsterdam: NorthHolland, pp. 135–82.

3. Iron town cluster: Yawata, its glory, decline and rebirth Mitsuhiro Kagami 1.

INTRODUCTION

Basic questions in this chapter are why a certain location is attractive to ﬁrms while others are not and why ﬁrms form industrial agglomerations. To ﬁnd answers to these questions traditional heavy industries are focused upon because these industries have long histories that possibly can provide some traceable factors to help explain industrial cluster formation. The iron and steel industry has a long history in developed countries because of its initial and substantial role in the industrialization process. In Japan one of the ﬁrst state-owned steel mills was established at Yawata in Kitakyushu City in 1901. Since then the city and the mill (Yawata Works) have shared their fortune, that is, rise, glory, fall and rebirth. By observing this process, common factors appertaining to industrial cluster formation can be considered. The US steel city of Pittsburgh, where the US Steel Corporation was established in the same year as Yawata Works, is brieﬂy examined in order to draw comparisons (see US Steel Corporation website). In comparing the two cities and steel companies some common as well as diﬀerent conclusions are realized. In this chapter, ﬁrst, Kitakyushu City, which has been home to both heavy and chemical industry clusters from the initial industrial development stage in Japan, is introduced. The city’s destiny has been closely interwoven with changes in the iron and steel industry, including redevelopment and revitalization programs that in recent decades have brought about a bright new future for the city. The more than 100-year history of Yawata Works, which highlights many important facets of Japan’s industrialization process, is examined in Section 3. Section 4 explains redevelopment projects and new businesses in Kitakyushu City after the steel industry declined. The comparison with Pittsburgh is undertaken in Section 5 and interesting similarities as well as diﬀerences with Kitakyushu City are highlighted. Finally, concluding remarks follow. 34

35

Iron town cluster: Yawata

2. KITAKYUSHU CITY: HEAVY INDUSTRY CLUSTER Kitakyushu City is recognized as a heavy and chemical industry cluster because of its foundation based on iron and coal. Iron and Coal Nexus Cluster Since the Edo era the Chikuho area has produced coal. Production at the Chikuho coalﬁelds peaked at 20.5 million tons in 1940 (see Table 3.1). A state-owned iron and steel mill was established at Yawata (now called Yawata Works) in 1901 mainly owing to its proximity to the coalﬁelds. Indeed, at ﬁrst Yawata Works had its own coalﬁelds at Futase. As a consequence of this coal and iron nexus many industries such as chemicals, metal, machinery, machine tools, cement, sheet glass, railway, shipping, power generation and banking developed. Although the central role of coal ended during the 1950s when it was replaced by petroleum, heavy and chemical industries remained in the Kokura and Moji areas, now called Kitakyushu City (see Figure 3.1; Kitakyushu City website). Table 3.1

Coal production at Chikuho coalﬁelds

Year

Japan (thousand tons)

Chikuho (thousand tons)

Its share (%)

1885 1890 1895 1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970

1293.6 2628.3 4772.7 7471.7 11 637.2 15 681.3 20 490.7 29 245.4 31 459.4 31 366.0 37 762.0 56 313.0 22 334.5 39 330.0 42.5 52.6 50.1 38.3

236.0 787.6 2136.6 4017.5 5804.1 7811.0 8769.6 11 689.7 12 746.8 11 467.5 14 988.0 20 490.0 7177.5 12 757.0 12.8 13.6 8.5 4.4

18.2 30.0 44.8 53.8 49.9 49.8 42.8 40.0 40.5 36.6 39.7 36.4 32.1 32.4 30.1 25.9 17.0 11.5

Source: Kitakyushu City (1998).

36

Agglomeration in Asia

Hokkaido

Honshu Kitakyushu City

Tokyo

Shikoku Kyushu Figure 3.1

Map of Japan

The city’s population peaked at 1 068 415 in 1979 but has since declined to about 1.02 million in 1995. The population of Fukuoka City, its neighbor and main rival, passed that of Kitakyushu City in 1980 and reached 1.29 million in 1995. The workforce living in Kitakyushu City peaked at 566 000 in 1965 but decreased to 501 000 in 1995. This decline in the workforce is closely related to the evolution of the iron and steel industry because Yawata Works is seen as the guardian of Kitakyushu City. Today, many well-known companies such as Yaskawa Electric Co. (robots and electronic products; see Yaskawa Electric Co. 2002), TOTO Ltd (bath, kitchen, washing and rest-room products; see TOTO Ltd 2002), Mitsui Hightech (plastic molds), Shabondama Soap (soap), Zenrin Co. (town maps and maps for car navigation; see Zenrin Co. 2002) and Sankyu Co. (transport) are located in this area. Moreover, Nissan Motor Co. and Hitachi Metals have their roots here in the Tobata Foundry established by Yoshisuke Aikawa, founder of the Nissan Konzern, in 1910.

Iron town cluster: Yawata

37

Why did Kitakyushu City form this industrial cluster? A simple explanation comes from the classic Marshallian trinity of external economies. First of all, the area was well endowed with natural resources such as coalﬁelds and harbors – coal being essential for making steel, which has strong forward linkages for other manufacturing industries as an input. Production of coal and steel was supported by steady demand due to government-led modernization and catch-up policies. Local governments were also keen to foster manufacturing industries and knowledge spillover worked as industries agglomerated. Moreover, large-scale capital goods industries such as the iron and steel industry have scale merits. This meant that the agglomerated area as a whole could beneﬁt from increasing returns to scale. Thus, this area enjoyed good industrial performance in terms of employment, especially, during the 1960s (see Table 3.2). Why has this area faced diﬃculties? From the 1950s onwards various factors conspired to change Kitakyushu’s status. Yawata Works decided to disperse operations by constructing giant new steel mills closer to other areas of demand (for example, Hikari, Sakai and Kimitsu, that is, distance matters) and many skilled workers were relocated away from Yawata Works to those new production centers. The decline of the coal industry due to substitution by petroleum also delivered a blow to the region. Moreover, external diseconomies such as pollution and environmental destruction also created centrifugal forces, particularly during the 1970s. Finally, overall declines in demand caused by the two oil shocks and severe competition in steel production from newly industrialized countries such as Korea and Brazil accelerated the process. Restructuring toward New Businesses As iron and steel production in Kitakyushu waned, mainly due to the construction of new mills elsewhere and severe pollution, the city had to ﬁnd a way to reform its industrial structure, particularly from the 1980s on. To this end, four routes to the future were outlined: (i) knowledgeintensive industries; (ii) pollution-free ecological approaches; (iii) international distribution center; and (iv) agglomeration of automotive industries. Not only the steel industry but also other industries in the Kitakyushu area had accumulated industrial and managerial knowledge from agglomeration that could be utilized for creating new ideas and businesses. Realizing that research and development (R&D) activities were essential, the city government emphasized R&D and technical assistance and invited universities as well as research institutes from private companies to locate there. Successful examples of this policy were the establishment of the Kitakyushu International Techno-Cooperative Association (KITA) in

38

Agglomeration in Asia

Table 3.2

Industrial census, Kitakyushu City

Year

Establishment (units)

Employment1 (persons)

Wage and salary2 (¥m)

Shipping amount3 (¥m)

1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

2195 2100 2075 2266 2197 2169 2450 2412 2365 2748 2681 2635 2819 2703 2646 2772 2647 2607 2082 2011 2849 1927 2689 1973 1880 2743 1847 2724 1929 1873 2668 1745 2502 1640 1731

135 241 132 757 129 605 128 676 128 586 126 104 130 292 127 520 124 546 124 552 123 409 120 197 119 270 112 626 105 969 102 803 100 547 95 150 94 890 92 699 93 138 89 205 89 140 88 530 84 051 82 789 81 908 83 238 84 484 82 393 82 676 76 515 75 932 72 539 72 862

276 105.6 304 406.3 306 414.4 311 083.6 339 066.5 357 976.6 389 268.1 407 565.7 435 560.6 450 637.4 470 819.0 484 789.3 496 579.3 471 214.2 460 093.6 440 469.1 439 968.5 433 825.6 436 300.5 421 262.3 421 873.5 432 443.1 427 218.9 432 829.9 407 578.8 410 029.0 415 218.7 397 218.4 406 684.8 394 051.5 391 705.9 388 969.1 385 514.7 407 608.7 389 907.9

931 217.2 1 120 039.5 1 102 192.2 1 149 882.6 1 363 010.5 1 455 917.3 1 579 173.6 1 816 969.6 1 958 642.4 1 843 852.4 1 880 770.7 2 052 461.3 2 013 763.8 2 002 179.5 2 162 477.1 2 100 592.1 2 154 458.4 2 311 633.7 2 372 680.2 2 353 587.3 2 193 276.1 2 300 296.8 2 435 895.1 2 253 570.8 2 085 577.7 2 226 352.1 2 257 812.9 2 298 857.3 2 455 126.8 2 353 984.3 2 216 965.2 2 125 055.7 2 185 270.6 2 235 219.2 2 315 964.9

Notes: 1. ‘Employment’, indicates permanent employees individual entrepreneurs with family employees. 2. ‘Wage and salary’, indicates wages and salaries at constant prices (CPI: 1995 100). 3. ‘Shipping amount’, at constant prices (WPI: 1995 100). Source:

Kitakyushu City Industrial Census, various issues.

Iron town cluster: Yawata

39

1980, Techno-parks and the Technocenter in 1990, and the plan for the development of the Kitakyushu Academic Research Promotion City in 1996. Yawata Works invested heavily in environmental protection, especially during the 1970s, and both Yawata Works and the city government promoted forestation of old factory areas (brownﬁelds) and the city itself. In addition, recycling of industrial waste, particularly, plastics and oﬃce automation (OA) machinery was intensely targeted, and the city allocated reclaimed land for this purpose (Corporation Recycle Tech 2002). The declaration of Kitakyushu City as an ‘eco-town’ or ecology town was announced in 1997 (Kitakyushu City 2002). This new approach extends value chains from normal production and sales to recycling and rebuilding activities, that is, including both ‘artery’ and ‘vein’ industries and hence increases new business opportunities. Owing to its geographical position, Kitakyushu has historically had contacts with Korea, China and East Asian countries. Taking advantage of its good position regarding sea routes, in 1996 the city launched a major container port plan, the Hibikinada Hub Port Initiative, in order to contribute to international logistics and beneﬁt from China’s emergence as a world economic power. Lastly, this area is now attracting the automotive industry. Nissan expanded its Kanda factory in the vicinity of Kitakyushu City and Toyota opened the new Kyushu plant near Fukuoka City during the 1990s. These moves inﬂuenced parts and component makers to relocate their factories to this area. These centripetal developments are explained by the existence of an accumulated technology base and a high industrial knowledge and skill-based workforce in the region. Good port facilities – easy delivery of parts and components and shipping out of ﬁnal products – are another advantage. Indeed, the very centrifugal forces that had convinced Yawata Works to downsize in Kitakyushu in favor of locating new mills in Honshu (central Japan) were viewed in reverse by the recent auto makers. Because Honshu (see again Figure 3.1) is congested in terms of transport cost and labor shortages, the auto industry decided to move into the Kyushu area. Recent information technology (IT) developments supported this tendency as information can easily be exchanged between the headquarters and local plants through Internet conferencing.

3.

HISTORY OF YAWATA WORKS

The modern histories of Kitakyushu City and Yawata Works are inseparable because the two have walked hand in hand for over 100 years. Indeed,

40

Agglomeration in Asia

the company’s evolution and decline has mapped out the city’s destiny so it is worthwhile here to sketch out the company’s history. Yawata Works In 1970 two giant steel makers, Yawata Iron and Steel Co. and Fuji Iron and Steel Co. merged to create Nippon Steel Corporation (NSC: see NSC website). As of April 2002, NSC had 25 229 employees, capital of ¥419 524 million yen (approximately US$3.5 billion), and sales of ¥1 681 406 million (US$14.0 billion). NSC has 10 plants in Japan: Muroran Works, Kamaishi Works, Tokyo Works, Kimitsu Works, Nagoya Works, Sakai Works, Hirohata Works, Hikari Works, Oita Works and Yawata Works (Yawata Works 2002). Except for Kamaishi, Yawata Works has the longest history because it started as a state-owned iron and steel company in 1901. In 2001, Yawata Works in Kitakyushu City produced 3.21 million tons of crude steel, 3.55 million tons of steel products, and had 3703 employees. Its products include hot rolled sheets, electrical sheets, cold rolled sheets, hot-dip-galvanized sheets, electro-galvanized sheets, alsheets, terne sheets, tinplate, tin-free steel, spiral pipes, rails, sheet piles, shapes and stainless steel plates. Yawata Works used 5257 thousand tons of imported iron ore with 3142 thousand tons of imported coking coal to produce crude steel. About 60 per cent of iron ore and 58 per cent of coal came from Australia. Yawata Works consists of two sites: Tobata and Yawata. At the Tobata site, one blast furnace is working to produce hot rolled sheets, cold rolled sheets, coated sheets and spiral pipes. Stainless heavy plates, rails, steel sheet piles and shapes, and electrical steel sheets are produced at the Yawata site. At its peak in the mid-1960s, Yawata Works produced 9 million tons of crude steel, and employed about 43 000 workers. Selection of the Factory Site When the Meiji government decided to construct a state-owned iron and steel enterprise, four factors were considered: (i) defense reasons; (ii) proximity to raw materials; (iii) proximity to a port and its facilities; and (iv) availability of laborers. Iron and steel production was very important for industrialization under the slogan ‘strengthen wealth and military power’. Thus, the factory should be safe from military attack. And, as steel production needs coal and iron ore, the factory should be near those supplies. If materials were to come from foreign countries, a good port was necessary. Moreover a port was also required to ship steel products for domestic as well as foreign demand. Lastly, production needs skillful workers from an abundant labor pool. Fulﬁlling these conditions, three

Iron town cluster: Yawata

41

candidates were chosen: Sakamura in Hiroshima Prefecture, Moji in Fukuoka Prefecture and Yawata in Fukuoka Prefecture (Nippon Steel Corporation 2001). In 1896 the government ﬁnally decided on Yawata as the mill site for the following reasons: (i) Yawata is situated inside Dokai Bay, which cuts deeply into the land and so is diﬃcult to attack; (ii) the proximity of the Chikuho coalﬁelds, one of the country’s largest at that time; (iii) the port facilities could be utilized and easily expanded; and (iv) there was a large available industrially trained workforce due to the nearness of the coalﬁelds and other existing businesses. At ﬁrst, the plan was to use iron ore from Akatani in Niigata Prefecture, but this was soon abandoned owing to delays in mining development, and instead ore from Daye along the Yangtze River in China was imported. With regard to coal, the Futase, Miike and Takashima mines in the Chikuho area supplied coking coals. The ﬁrst blast furnace at Higashida was blown in 1901, starting the ﬁrst integrated iron and steel works in Japan. Technologies, especially blast furnace technologies, came from Germany. Yawata Works invited German engineers and technicians to set up and kindle the ﬁrst blast furnace. It was said that the salary of one of the technical advisers was double that of Japan’s prime minister. Several Japanese workers were also sent to Oberhausen to learn steel-making technologies as well as maintenance skills.1 Because of the government’s push to industrialize the country and strengthen military power, demand for iron and steel was very strong. The Russo-Japanese War (1904–05) also gave impetus to this demand. Yawata Works had three consecutive expansion plans: ﬁrst, in 1906; then in 1911; and ﬁnally in 1917. Completion of the sixth blast furnace at Higashida in 1921 was one of the highlights of the third expansion plan (see Table 3.3). However, the Great Depression in 1929 seriously aﬀected world economies and slashed the demand for iron and steel. Private iron and steel companies faced severe diﬃculties and ﬁnally in 1934 Yawata Works and ﬁve other companies merged, creating the Japan Steel Corporation (see Figure 3.2). Yawata Works itself recorded the highest crude steel production of 250 000 tons per year at Higashida in 1935. The Second World War devastated the Japanese economy. The ﬁrst US air raid targeted Yawata in 1944 and severely damaged Yawata Works. Japan had to reconstruct its economy from the ashes of defeat in 1945. Success and Japan’s Economic Rise In accordance with the General Headquarters (GHQ) guidance, Japan’s large companies (zaibatsu) were split up or dissolved. Japan Steel Co. was

42

1941 1947 1948 1950

Reproduction started Production Five-year Plan Japan Steel Co. was dissolved into four private companies including Yawata Iron & Steel Co. Ltd Second blast furnace at Yawata (Kukioka) kindled Fourth blast furnace at Yawata (Higashida) kindled Modernization Plan for Production Facilities

Yawata village was oﬃcially chosen as a public steel mill site First blast furnace at Yawata (Higashida) started operation Second blast furnace at Yawata (Higashida) started operation First expansion plan Third blast furnace at Yawata (Higashida) started operation Second expansion plan (steel products 300 000 t/year) Fourth blast furnace at Yawata (Higashida) started operation Third expansion plan (steel products 650 000 t/year) Fifth blast furnace at Yawata (Higashida) started operation Sixth blast furnace at Yawata (Higashida) started operation Third blast furnace at Tobata started operation First blast furnace at Yawata (Kukioka) started operation Second blast furnace at Yawata (Kukioka) started operation Six steel companies including Yawata Works were merged to form Japan Steel Co. (up to 1950) Third blast furnace at Yawata (Kukioka) started operation Fourth blast furnace at Yawata (Kukioka) started operation. This furnace was domestically produced with capacity of 1000 t/day

1897 1901 1905 1906 1909 1911 1914 1917 1918 1921 1924 1930 1933 1934

1937 1938

Events at Yawata and Tobata sites

Year

Table 3.3 Brief history of Yawata Works

• Korean War 1950–53

• Paciﬁc War 1941–45 • Emphasizing production of coal and steel

• US Steel was established

Related issues

43

1979 1984 1985

1977 1978

1976

1973

1970 1972

1969

1960 1961 1962 1963 1966 1967 1968

1951 1955 1959

Accumulated pig iron production reached 200 million tons

Production Master Plan Yawata Works recorded 100 million tons of pig iron Yawata and Fuji merged to form Nippon Steel Corporation (NSC) First and sixth blast furnaces at Yawata (Higashida) extinguished Fourth blast furnace at Tobata kindled Tobata strip mill recorded 10 million tons of cold-rolled steel sheet in terms of accumulated ﬁgure from the start of operation Factory Forestation Agreement with Kitakyushu City concluded Thick plate on-line-system was introduced Japan’s crude steel production recorded 120 020 000 tons Reconsideration of the Production Master Plan including production reduction and concentration into the Tobata area Seamless pipe production started All blast furnaces at the Yawata area extinguished. Two blast furnace system (ﬁrst and fourth at Tobata) established

Yawata Works recorded 9.2 million tons of crude iron Pollution Control Committee formed

New ﬁrst blast furnace (integrated iron and steel production) at Tobata kindled, scrapping two old ones Second blast furnace (1500 t/day) at Tobata kindled IBM computers 7070 and 1401 were introduced Third blast furnace (2000 t/day) at Tobata kindled

First blast furnace at Yawata (Higashida) kindled

Sakai Works began operation First blast furnace kindled at Usiminas (Brazil) Kitakyushu City founded Kasumigaseki building (36 ﬂoors) built Malayawata (Malaysia)’s blast furnace kindled Kimitsu’s ﬁrst blast furnace kindled VERs of steel against the US announced Kimitsu’s second blast furnace kindled

• Plaza Accord (Yen appreciation)

• Agreed to assist the construction of Shanghai Baoshan Steel Co. • Second oil shock

• First oil shock • First blast furnace of Pohang Steel (Korea) kindled

• • • • • • • •

• Hikari Works began operation

44

Source:

2002

1999 2000 2001

1998

1995 1997

Compiled from Nippon Steel Corporation (2001).

Lowest production of crude steel (25.5 million tons) by NSC recorded Mid-term Management Plan (In the plan, 24 million t/y by NSC against 90 million tons of all Japan’s crude steel production. One blast furnace at Yawata, two at Nagoya, three at Kimitsu, and two at Oita in operation; Sakai and Hirohata close down) Fourth blast furnace extinguished only one remained at Tobata New strip mill producing cold-rolled steel sheet started operation Theme park, ‘Space World’, opened All products got ISO 9000 series Accumulated seamless pipe production reached 10 million tons Accumulated pig iron production reached 250 million tons Nishi-nihon (West Japan) PET Bottle Recycle Co. started operation First blast furnace at Tobata closed and the fourth operation restarted after renovation Obtained ISO 14001 Nishi-nihon (West Japan) Auto Recycle Co. started operation New cold rolling mill for electrical steel sheets began operation Withdrew from seamless pipe production Joint-venture with Shanghai Baoshan Steel announced, producing steel sheets for automobiles in China Waste plastics recycling facility began operation

1986 1987

1988 1990

Events at Yawata and Tobata sites

(continued)

Year

Table 3.3

• Alliance with Sumitomo Metals and Nisshin Steel for stainless steel • NSC with Sumitomo Metals and KOBELCO in terms of stock and management

• City started ‘Kitakyushu Eco-town project’

Related issues

45

Kamaishi Mining Tanaka Steel Mill 1887

Fuji Steel Co. 1917

Mitsubishi Steel Co. 1917

Tanaka Mining 1917

Hokkaido Steel Wanishi Steel Mill 1917

Kyushu Steel Co. 1917

Japan Steel Co. Wanishi 1919

1918

Kamaishi Mining 1924

Tokai Iron & Steel Co. 1958

Fuji Iron & Steel Co. 1950

Yawata Iron & Steel Co. 1950

Japan Special Steel Pipe Co. 1935

Wanishi Steel Co. 1931

1967

Yawata Steel Pipe Co. 1960

Japan Steel Co. 1934

1934

1968

Nippon Steel Corporation 1970

Evolution of major steel companies in Japan

See Table 3.3.

Figure 3.2

Source:

Note: In addition to Nippon Steel Corporation, there are also four other major steel companies in Japan in 2002: Kobe Steel Ltd (KOBELCO), Sumitomo Metal Industries Ltd, JFE (Kawasaki Steel Co. and NKK Corporation) and Nisshin Steel Co.

State-owned Kamaishi Mining Co. 1874

Hokkaido Coal Mining Shipping Co. 1909

State-owned Steel Co. at Yawata 1901

Toyo Steel Co. 1917

Tobata Steel Co. 1918

46

Agglomeration in Asia

also divided into four private companies. Thus, Yawata Iron & Steel Co. was reborn in 1950. The government adopted a short-cut policy to quickly reconstruct its economy due to severe shortages in coal: the so-called ‘slope production’ or ‘two-sector priority production’ method. That is, the strengthening of the coal and steel industries was given priority. All available economic resources were earmarked for the two industries. Imported petroleum was preferentially forwarded to the iron and steel industry whose subsequent output was channeled into the coal industry. Next, the increased production of coal was preferentially re-circulated back to iron and steel. This priority treatment of the two industries resulted in output surplus that later went to other industries (Kagami 2001). The Korean War in 1950–53 revitalized Japan’s economy through special war procurements and maintenance, especially for iron and steel. Because of its proximity to the Korean Peninsula, Kitakyushu acted as a base for this war-related demand. This economic buoyancy accelerated the expansion and modernization of Yawata Works’ facilities. The Production Modernization Plan started in 1950, when old facilities were scrapped and new ones installed. This plan continued until 1955, extending the original plan for two more years. In 1957, construction of a brand new integrated iron and steel mill started at the Tobata site on reclaimed land with modern port facilities. The ﬁrst new blast furnace at Tobata started operation in 1959, replacing two old ones. During the 1950s, the government pushed ‘targeted’ industries by providing tax and lending incentives. These were called ‘industrial policies’. Such industries as iron and steel, coal, shipbuilding, synthetic ﬁber and chemical fertilizer were chosen as targeted industries. Prime Minister Hayato Ikeda launched the ‘Doubling National Income Plan 1961–70’ in 1960 and the government set an objective for trade liberalization as shown in the ‘Outline Program for Liberalization of Trade and Foreign Exchange’ approved in the same year. The acceptance of obligations under Article 8 of the International Monetary Fund and entry into the Organization for Economic Cooperation and Development in 1964 ﬁnally committed Japan to observing an open international economic system.2 Japan experienced handsome growth rates in this period. The average annual rate of growth of real GDP recorded 9 per cent between 1961 and 1972, while industrial structure widened as well as deepened and exports expanded. For example, in 1964, the Tokaido Shin-kansen (bullet train) started operation; the Tokyo Metropolitan Highway opened; and the Tokyo Olympic Games were held. The ﬁrst high-rise oﬃce building, the Kasumigaseki Building (36 stories) was constructed in the central part of Tokyo in 1968, for which Yawata Works provided H-shaped bar steel that eﬀectively absorbed earthquake tremors.

Iron town cluster: Yawata

47

Yawata Works recorded crude steel shipments at 100 million tons in 1965. In order to respond to increased demand for iron and steel, especially in the largest demand area, that is, Tokyo Metropolitan area and its vicinity, Yawata Works decided to construct a mammoth steel mill at Kimitsu in Chiba Prefecture in 1967. Two blast furnaces were constructed by 1969, making Kimitsu Works the most modern-equipped iron and steel mill in the world. Almost half of the workers at Kimitsu came from Yawata Works. In 1970 the NSC was founded, combining two giant companies, Yawata Iron & Steel Co. and Fuji Iron & Steel Co. albeit facing some opposition, particularly from the Fair Trade Commission (see again Figure 3.2). The iron and steel industry enjoyed its most prosperous period between 1965 and 1980 with global supremacy in terms of cost and quality, but some negative characteristics had already started to appear to cloud its glory. Decline and its Factors Three factors negatively aﬀected iron and steel production: (i) demand shift caused by oil-price hikes, so-called oil shocks; (ii) trade friction due to Japan’s disproportionate presence in international markets; and (iii) environmental destruction because of rapid and large-scale industrialization. The ﬁrst oil shock in 1973 was a memorable year for Japan. The economy fell into minus growth in 1974 for the ﬁrst time in postwar history. Such energy-intensive industries as iron and steel, nonferrous metal, chemicals and paper and pulp declined to the point of depression while the automotive and electronics industries barely survived due to their relatively light energy requirements and in-house eﬀorts to save/substitute energy. The second oil shock in 1979 further accelerated energy saving and energy substitution technologies. For example, the automotive and electronics industries required lighter materials such as aluminum and plastics instead of iron and steel. Demand was irrevocably changed due to these oil shocks. Japan’s crude steel production peaked in 1973, reaching 120 million tons. Production never again exceeded this peak although it remained around the 100 million ton mark up to 2000 (see Table 3.4). Another negative factor was Japan’s participation in the world economy. Capital liberalization, which began in 1967, ﬁnally reached full liberalization in July 1972. Coupled with the yen appreciation that followed the collapse of the Smithonian monetary system in 1971 (the so-called ‘Nixon shock’), Japan’s foreign direct investment expanded four times between 1971 and 1973, reaching US$3.5 billion. Japan’s participation in the international economy in terms of both trade and investment created new problems: trade friction and/or trade imbalances with other countries. For

48

Agglomeration in Asia

Table 3.4 Crude steel production (thousand tons) and employees at Yawata Works Year

Japan

Nippon Steel Co.

Yawata Works

Employees at Yawata Works

1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988

5298 6782 6912 8033 7875 9791 11 678 12 309 12 773 18 247 23 161 29 399 27 250 34 080 40 532 41 296 51 898 63 777 68 987 87 026 92 406 88 441 102 972 120 017 114 035 101 613 108 326 100 646 105 059 113 010 107 386 103 029 96 299 100 200 106 470 103 758 96 379 101 877 105 656

2335 3136 3165 3474 3492 4247 4785 4946 5372 7341 8889 11 141 10 064 12 091 14 381 14 917 18 678 22 734 24 400 31 098 32 982 29 971 35 369 40 989 36 899 32 293 34 394 31 655 31 994 33 582 31 683 29 970 27 050 27 727 29 596 27 980 25 566 27 142 28 217

1466 1816 1815 1999 1929 2361 2673 2822 3064 4336 5197 6271 5602 6523 7689 6889 7943 9166 8587 8794 8651 7496 7757 8301 7490 6887 7047 5869 5470 6152 5967 5585 5104 5374 5835 5161 4917 5259 4608

35 038 37 087 36 729 35 431 34 578 33 697 33 237 33 524 33 988 37 027 37 326 39 893 42 220 43 666 39 677 37 705 36 235 34 577 32 486 30 030 27 624 26 364 24 917 23 757 22 847 21 575 20 554 19 932 19 116 18 207 17 202 16 404 16 267 15 885 14 905 14 692 13 797 13 547 12 261

49

Iron town cluster: Yawata

Table 3.4

(continued)

Year

Japan

Nippon Steel Co.

Yawata Works

Employees at Yawata Works

1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

108 139 111 710 105 853 98 937 97 095 101 363 100 023 100 793 102 800 90 979 97 999 106 901

28 361 28 992 27 686 25 319 25 122 26 564 26 172 25 705 26 618 23 200 25 620 27 838

2738 3852 3361 3351 3311 3383 3307 2911 3191 3132 3570 3799

11 373 10 472 9958 9838 9586 9199 8195 7292 6180 4900 4351 3873

Source: See Table 3.3.

example, the US claimed that the Japanese textile and iron and steel industries dumped products below fair market value. Japanese steel exports to the US increased rapidly from 4.5 million tons in 1967 to nearly 7 million tons in 1968. In response to US criticism, Japan carried out voluntary export restraints (VERs) from 1969 to 1974. Later, the US imposed the ‘trigger-price’ system3 on steel imports from Japan in 1978. Negative aspects of industrialization were also taken into account. Rapid industrialization during the 1960s led to increased levels of pollution and environmental destruction. By the late 1960s, problems such as smog and water pollution (for instance, the so-called ‘Minamata disease’ caused by mercury poisoning at Minamata Bay in Kumamoto Prefecture) forced the government to acknowledge and address these external diseconomies. Accordingly, the government announced the Basic Act for the Prevention of Public Nuisances in 1967 to ﬁght against pollution and other negative eﬀects of industrialization. Related laws and ordinances were soon passed and addressed problems such as water pollution (1966), noise pollution (1968) and air pollution (1968). Response of Yawata Works As far as Yawata Works is concerned, crude steel production had a declining tendency after the 1960s. Ten-year average crude steel production of Yawata Works increased from 2.4 million tons in the 1950s to 7.3 million tons in the 1960s but decreased to 7.1 million tons in the

50

Agglomeration in Asia

1970s, 5.1 million tons in the 1980s, and went even further down to 3.3 million tons in the 1990s4 (see Figure 3.3). The maximum production was recorded in 1967 at 9166 thousand tons, and since then it has gradually decreased. The lowest point was 2738 thousand tons in 1989. The last couple of years of the century showed a slight upward trend to 3799 thousand tons in 2000. Several reasons are given for this decrease in Yawata’s position. First, NSC diversiﬁed production sites, opening new mills near marketing areas. For example, Hikari Works near Hiroshima was opened in 1955, and between 1954 and 1969 about 1270 personnel were transferred to that site from Yawata Works. In Sakai Works near Osaka (1961), some 2940 personnel were moved between 1960 and 1969. However, the largest move took place at Kimitsu Works near Tokyo (1965), accounting for 4173 personnel being transferred between 1964 and 1975. Therefore, the share of Yawata Works in total NSC production declined from more than 55 per cent during the 1950s to about 13 per cent during the 1990s. Employment ﬁgures show more dramatic drops. In its peak year (1963) Yawata Works employed 43 666, but since that time the number has decreased drastically, falling to 3873 in 2000 (see Figure 3.4). This radical shift in the fortunes of Yawata is explained by changes in production, site location as mentioned above and the company’s eﬀorts to increase labor productivity, but a general decline in demand for iron and steel also played a part. However, yet another reason was that NSC’s competitors both in and outside of Japan gained ground. Iron and steel makers such as Kawasaki Steel Corporation, Kobe Steel Ltd, NKK Corporation, Sumitomo Metal Industries Ltd and Nisshin Steel Co. gradually expanded production capacity and encroached on the traditional leader. NSC now accounts for about 26 per cent of crude steel production in Japan as compared with more than 45 per cent in the glory years. It is also a little ironic that NSC assisted in the establishment of iron and steel mills in developing countries in the past, as some of the ﬂedgling ﬁrms that NSC helped to get started are now serious competitors in the world market. The Japanese government set up a joint venture to establish USIMINAS (Minas Gerais Steel Mill) in Brazil and NSC played a major role in its establishment in terms of technology and management at the beginning of the 1960s. In the cases of Malayawata in Malaysia, Pohang Steel in Korea and Shanghai Baoshan Steel in China, NSC also assisted in setting up blast furnaces and training workers. As a result, Pohang Steel in particular has become one of the world’s largest iron and steel producers. Consequently, nearly all the blast furnaces in the Yawata area were shut down. Yawata Works had to concentrate its iron- and steel-making divisions in the Tobata area, keeping two blast furnaces in operation in 1978.

51

Crude steel production of Yawata Works, 1950–2000 (thousand tons)

See Table 3.3.

Figure 3.3

Source:

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

52

Number of employees at Yawata Works, 1950–2000

See Table 3.3.

Figure 3.4

Source:

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Iron town cluster: Yawata

53

By 1988 only one blast furnace remained in operation, symbolizing the near-death of a town that now lives in the shadow of its past glory. Environmental protection During the 1950s and 1960s, black smoke and dirty drainage were symbols of an iron town. People suﬀered from asthma because of photochemical smog, and insomnia or hearing diﬃculties because of noise. Owing to national as well as city government regulations against such pollution and environmental destruction, Yawata was forced to adopt several measures. For example, it installed smoke extraction apparatus for desulfurization of soot and smoke. The company was required to spend heavily on environmental protection due to its noise and industrial waste emissions. In the peak year (1974) it spent ¥21 260 million (approximately US$177 million) for such protective measures as clean air (¥18 320 million), anti-noise devices (¥2390 million), and clean water (¥550 million). The accumulated amount between 1963 and 2000 reached ¥92 999 million (approximately US$775 million) (air ¥69 463 million or 74.7 per cent; water ¥16 314 million or 17.5 per cent; and noise ¥7222 million or 7.8 per cent) (NSC 2001). In 1973, Yawata Works concluded a Factory Forestation Agreement with Kitakyushu City. Since then Yawata Works has planted trees and ﬂowers in factory yards and expanded park areas to promote a clean and comfortable atmosphere.

4.

REVIVAL OF THE CITY

After the bubble economy burst, Japan suﬀered a lengthy recession in the 1990s. Many local governments faced diﬃculties, with severe ﬁscal deﬁcits. Municipal governments also had to tackle these diﬃculties and create new development plans for future prosperity. Although total demand for iron and steel did not decrease, Yawata Works faced severe complications in order to survive and new ideas and breakthroughs were desperately needed. Role of the City Government The Kitakyushu City government created new ideas and plans to challenge these diﬃculties under Mayor Koichi Sueyoshi (1987–present). They are summarized as follows (Kitakyushu City 1998): ● ● ●

Kitakyushu Technoparks and Technocenter (1990); new Kitakyushu International Airport (construction began in 1994 and ﬁnished in 2005); plan for Kitakyushu Academic Research Promotion City (1996);

54

Agglomeration in Asia ● ● ● ●

Hibikinada Hub Port Initiative (1996); Asia-Paciﬁc Import Mart (AIM) and Kitakyushu International Distribution (KID) Center (1996); Kitakyushu Eco-town Project (1997); and Special Free Zone for International Distribution (2002).

Of these, the Hibikinada Hub Port Initiative and the Kitakyushu Eco-town Project are very important. The port of Kitakyushu (Moji, Shin-Moji, Wakamatsu and so on) handled 86 million tons of cargo including both foreign and domestic trade (export import) in 2001 according to the Kitakyushu Port and Harbor Bureau (2002), ranking it seventh in Japan (after Chiba, Nagoya, Yokohama, Kawasaki, Mizushima and Osaka). Container cargo volume was 387 thousand TEUs (twenty-foot equivalent units) in 2001. Principal trading partners for container cargo were China (42 per cent), Taiwan (20 per cent), Korea (15 per cent), Hong Kong (5 per cent) and Thailand (3 per cent). Because of the heavy trade with neighboring countries the city decided in 1996 to construct a large-scale container port, Hibikinada, which has a 43-hectare (ha) container yard on reclaimed land (total 2000 ha). Port facilities opened in 2005 including two berths with a depth of 15 meters and two berths with a depth of 10 meters, accommodating handling capacity of 500 000 TEUs. Kitakyushu Eco-town Project5 (Kitakyushu City 2002) started in 1997 is a recycling cluster which includes the recycle, reuse and rebuilding of PET (Polyethylene Terephthalate) bottles, electronic machines (personal computers, printers, copy machines, electric domestic appliances and so on), and automobiles on reclaimed land. This is described as ‘vein’ as against ‘artery’ industries which are used in normal industrial activities. At present, the project covers a 41 ha area where about 30 ﬁrms and research institutions participate. For example, the capacity of the auto recycle shop is 1500 cars per month and that for electrical domestic appliances 4000 goods per day. The second development phase expands the area to 100 ha. This recycling idea is very interesting, as producers have begun to consider whole supply chains including goods in the post-consuming phase. This extended value chain enriches our notion of industry and produces new business opportunities. From now on, consumer products will be designed for reuse and/or recycling from the outset6 (see Figure 3.5). New Businesses Set Up by Yawata Works In the above-mentioned new projects, Yawata Works played a vital role by establishing wholly owned companies or joint ventures. The ﬁrst Yawata

55

Iron town cluster: Yawata

Artery industries

Vein industries

R&D

New R&D

Design

Redesign

Parts procurement

Reuse

Production

Rebuilding

Sales

Distribution

After-care service Figure 3.5

Recycling

Recycling: extended value chains

Works project that diversiﬁed from steel making was a theme park. Yawata Works decided to transfer all iron- and steel-making divisions in the Higashida area to the Tobata area in 1978. The idle Higashida area was redeveloped for new urban recreation centers during the 1990s. Yawata Works entered into an agreement with NASA (the US National Aeronautics and Space Administration) to construct a theme park on science and space, resulting in the birth of ‘Space World’ in 1990. New businesses were set up one after another by Yawata Works, including six engineering companies, two technical support companies, two vocational

56

Agglomeration in Asia

schools, two medical facilities, one supermarket, two travel service agencies, four telecommunications and broadcasting companies, one printing company, three estate companies, and six environmental and recycling companies, including the West Japan PET Bottle Recycle Co. (established in 1998) and West Japan Auto Recycle Co. (in 2000). In addition, Yawata Works sells electricity to Kyushu Electric Power Co., using gas generated from the blast furnace at Tobata. Automotive Cluster Island Kyushu Island has gradually emerged as an automotive agglomeration because major auto makers established their assembly plants in or near Kyushu. Nissan ﬁrst invested in Kanda Town, adjacent to Kitakyushu City, in 1976 and then expanded its assembly factory in 1992. Honda came to Otsu Town in Kumamoto Prefecture in 1976 assembling motorcycles. Mazda constructed its ﬁrst assembly factory at Bofu in Yamaguchi Prefecture in 1982 and opened a second plant in 1992. Toyota came to Miyata Town in Fukuoka Prefecture in 1992. Daihatsu started operations in Nakatsu in Oita Prefecture in 2004. These developments created the incentive for parts and component makers to move into the area and an automotive cluster came into being. It is said that the Nissan Kyushu Plant has 66 parts supply makers within Kyushu (Nissan Kyushu Plant 2002) while Toyota Kyushu Plant has about 50 suppliers west of Hiroshima. Honda Kyushu has 53 suppliers in Kyushu and Mazda’s Bofu plant has 28 suppliers in Yamaguchi Prefecture. Accordingly, Yawata Works faced increased demand for high-tension thin plate steel from the automotive industry. As a result, Yawata Works set up a new cold strip mill and started operation in 1990.

5.

COMPARISONS AND NEW DIRECTIONS

There are some similarities between Kitakyushu City and Pittsburgh in the US.7 Both cities thrived and declined with the iron and steel industry. Kitakyushu City has Yawata Works while Pittsburgh has the US Steel Corporation. Both companies were established because of the proximity of materials, especially coal: the Chikuho coalﬁelds for Yawata Works and the Appalachia coalﬁelds for US Steel. Both cities suﬀered from heavy pollution and demand deterioration due to economic changes in the steel industry. Both cities are in the process of restructuring. Thus, it is interesting to compare these two cities, once called ‘smoke towns’.

Iron town cluster: Yawata

57

Pittsburgh The city is situated on a sandbank where the Allegheny and Monongahela rivers meet and form the Ohio River in the State of Pennsylvania. After the completion of the Pennsylvania Canal (1843) and the Pennsylvania Railway (1852), the city played the role of gateway to the West. Many immigrants from Europe came to the city because the Civil War (1861–65) brought an instant boom as a result of special war demands for weapons and munitions. Between 1870 and 1910, the coal and iron and steel industries thrived. The US Steel Corporation was founded in 1901 (the same year as Yawata Works). In the 1920s, Pittsburgh was called the ‘capital of the world’ since it was recognized as a leading world manufacturing center with beautiful Victorian architecture and bustling streets. The two world wars also stimulated the production of iron and steel and brought huge wealth to the steel industry. However, since the 1950s, the city has endured severe air and water pollution, resulting in it being known all over the US as ‘smoke town’. Many mills and related plants became bankrupt and closed due to the upsurge of steel imports from Europe and Asia since the 1970s. It was said that half the employees of the iron and steel industry were ﬁred and many left the city.8 Confronting these problems, Pittsburgh City government adopted several measures, including the ‘Renaissance Project’ (ﬁrst period, 1945–69; and second, 1978–88). The ﬁrst period emphasized three controls: control of smoke and smog, ﬂood control and sewage control. The second period put emphasis on redevelopment of old factory sites. Because the steel industry declined, many empty mill and plant sites and railway yards were abandoned. Moreover, large tracts of land, polluted by industrial waste including slag, remained. The city undertook a policy of redeveloping these areas, known as ‘brownﬁelds’ as against ‘greenﬁelds’. The redevelopment of these brownﬁeld sites included housing projects, construction of parks and recreation facilities, hotels, shopping malls as well as business and cultural centers. Brownﬁeld sites faced special diﬃculties because often the land was contaminated by industrial waste such as PCB (polychlorinated biphenyl) and asbestos. To eradicate poisonous materials or seal up the waste took time and money. The city, private companies and community groups tenaciously undertook to remedy this process. Gradually, the city has recovered and revitalized business activities, particularly in high-tech industries such as information and communication technologies (ICTs), biotechnologies and robotics. Now, Pittsburgh has more than 250 software companies employing about 25 000 and is recognized as one of the R&D centers in the US (CLAIR 2002).

58

Agglomeration in Asia

Changing Actors in Redevelopment In the case of Pittsburgh, new approaches were adopted for redevelopment and revitalization programs. In particular, non-proﬁt organizations (NPOs) played a vital role in the process. In the center of Pittsburgh, about 60 NPOs are operating in the ‘NPO Tower’. Such NPOs as Sustainable Pittsburgh (environmental urban planning) and PPND (housing and community planning; see the PPND website) enthusiastically participated in the redevelopment projects. From the beginning they took part in discussions and contributed ideas for urban planning and housing design to the municipal government. Recently NPOs have become more professional. Moreover, Pittsburgh has wealthy foundations such as the Carnegie Foundation, the Heinz Endowment and the Richard King Foundation, which provide money for redevelopment. There are about 20 such foundations in Pittsburgh. Combined with the city government (Urban Redevelopment Agency of Pittsburgh: URAP), NPOs and community groups, Pittsburgh has succeeded in revitalizing its industries to transform itself into a ‘high-tech town’ or ‘smart town’. Diﬀerences between Kitakyushu City and Pittsburgh There are three main diﬀerences between the two cities. First, Kitakyushu City is situated along the coastline while Pittsburgh is inland. In the redevelopment stage this diﬀerence created the growth diﬀerence. Kitakyushu City can expand its land area by reclaiming the sea. Hence, Kitakyushu City has more options to develop other industries such as container ports, recycling towns, R&D centers and even a new airport. Pittsburgh had no spare land for new businesses to expand into, and as a result its population has shrunk to almost half that of the 1950s. Growth potential was limited by its location. Second, Pittsburgh tended to foster more ICT-related high-tech industries together with universities such as Carnegie-Mellon and Pittsburgh after the fall of the steel industry while Kitakyushu City stuck with manufacturing goods (monodukuri or making goods), by introducing recycling, reuse and rebuilding goods. Pittsburgh changed to become more service oriented while Kitakyushu City, in a sense, extended monodukuri value chains, including recycling, along the lines of manufacturing. In comparison, Pittsburgh became a high-tech town while Kitakyushu City became an ecology town. Third, a new actor emerged in the redevelopment stage in Pittsburgh, that is, NPOs. On the other hand, Kitakyushu City’s redevelopment processes did not have such participation maybe because of underdevelopment of

Iron town cluster: Yawata

59

NPO and non-governmental organization activities in Japan. Such activities with rich participatory foundations cannot be found in Japan because of reasons mentioned above and taxation problems.9

6.

CONCLUDING REMARKS

We raised the questions as to why certain locations attract ﬁrms, and why ﬁrms agglomerate. In this chapter we focused on heavy industries as against high-tech or service industries. Traditional heavy industries in Japanese are called juko-chodai (literally meaning heavy, thick, long and large) and are now generally considered to be dinosaurs. However, the truth often lies in history, and careful observation of these traditional industries shows us some simple explanations as to why cluster forming does or does not occur. Several factors that attract ﬁrms, especially the case of iron town clusters, are summarized as follows: ● ●

● ●

●

●

●

Existence of raw materials is essential. The iron and steel industry needs coal and iron ore, so mills are built near these resources. Well-organized infrastructure is needed such as railways, canals, highways and ports. As the iron and steel industry sometimes needs to import foreign materials, a good port is necessary. This is also true for the shipping of products. In addition, stable supplies of electricity, gas and water are also indispensable. A large pool of quality labor is another factor to attract ﬁrms. A coal and iron nexus attracts related industries such as chemicals, metals and machinery industries so that a heavy industry agglomeration is formed. Steel products and chemical materials are used as an input for other industries; that is, forward linkages are strong while consumer durables such as the automotive industry have solid backward linkages. Both strong forward and backward linkages in the input–out (IO) sense produce an industrial agglomeration.10 Important crossing points in terms of roads, railways, telecommunications and banking services are also incentive factors for encouraging other industries to come. Yawata Works activities shrank mainly due to the relocation or new mills constructed in areas other than Kitakyushu, in addition to demand shifts due to the two oil shocks and environmental destruction. New businesses emerged based on accumulated knowledge of steel making and pollution controls. The inclusion of recycling changed

60

Agglomeration in Asia

●

●

concepts of production (matching vein with artery industries) and created new businesses. The help of the city government is also essential. The municipal government provides tax and ﬁnancial incentives and infrastructure supplies. In particular, the city government can create initiatives to redevelop and revitalize industries and the city itself after traditional industries have declined. It is interesting that automotive industries are moving into the area and this development will help to form another type of agglomeration diﬀerent from the old one based on the iron and steel industry.

Two factors are worth mentioning. One is the notion of recycling. Kitakyushu City is emphasizing recycling, reuse and rebuilding of used or abandoned industrial products such as plastics, OA machines, electrical domestic appliances and automotives. This is an extended value to the supply chain. Enterprises have to start producing products taking into account the recycling phase. This will change production styles from the beginning, such as design and production processes using reused parts and components. This represents a new type of industrial agglomeration based on vein industries. The other important factor is the participation of NPOs in the decisionmaking processes of local governments. In this context Japan lags far behind, partly because of the amateur status of NPOs and partly because of a lack of understanding by public oﬃcials. Japan has to foster professional NPOs as well as enhance public oﬃcers’ recognition of NPOs. Moreover, contributions and donations should be ﬂexibly treated in terms of taxation in order to promote foundation activities. In sum, two forces have worked to foster the Kitakyushu City industrial cluster. The iron and steel industry is a classic case of the Marshallian trinity when the city expanded, that is, linkages, thick markets and knowledge spillovers. The industry itself had scale merits. Thus, the city’s agglomeration had both externalities and increasing returns. In the declining phase, external diseconomies such as environmental destruction worked in addition to the relocation of steel mills (transportation cost) and declines in demand. Decisions taken by head oﬃces on the location of new plants to be constructed, both in the case of the steel makers and the auto makers, were inﬂuenced by these two forces, that is, centripetal and centrifugal, however the conclusions reached could be said to be diametrically opposed. The NSC built new mills closer to the market-demand area during the 1960s (distance matters), whereas the contemporary auto makers have chosen to agglomerate in the Kyushu area mainly due to the IT revolution (‘distance is dead’).

Iron town cluster: Yawata

61

Lastly, city government initiatives to reclaim land and foster ‘vein industries’ are highly appreciated in the redevelopment stage because they will attract new industries.

NOTES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Those with excellent skills were later nominated as Shukurou, a life-long position to teach skills to other workers. Only seven were appointed to such a position in the factory’s history. Except for free movements of capital which was prolonged until 1972 for outﬂows and until 1992 for inﬂows. This system set a minimum price for Japanese steel products, and when import prices went below the set price, the US Treasury Department would initiate an investigation into the matter. Ten-year average crude steel production of NSC ﬂuctuated: 4.2 million tons in the 1950s; 16.8 million tons in the 1960s; 34.0 million tons in the 1970s; 28.3 million tons in the 1980s; and 26.1 million tons in the 1990s, also showing a declining trend after the 1970s. This area was nominated by the Japanese government as a special recycling estate. Sixteen areas were nominated all over Japan. The Government passed the Automotive Recycling Act in 2004. In fact, in 1999 both cities entered into an agreement called the ‘Kitakyushu/Pittsburgh Business Partnership Tie’. The population of the city of Pittsburgh was 670 000 in 1950 and 335 000 in 2001. Contributions and donations have partial or limited exemption in the tax calculation in Japan. This type of cluster is sometimes called a jokamachi, literally an industrial castle town where a lord attracts many people to live surrounding his castle.

REFERENCES CLAIR (Council of Local Authorities for International Relations) (2002), Pittsburgh, PA, www.clair.nippon-net.ne.jp/, accessed March 2003. Corporation Recycle Tech (2002), Company pamphlet (in Japanese), Kitakyushu City. Kagami, Mitsuhiro (2001), ‘Japan’s development model: success or failure?’, Paper presented to the academic forum on ‘El Modelo de Desarrollo Asiático: Relevancia para México’, organized by the Paciﬁc Study Center, University of Guadalajara, Mexico, May 7–8. Kitakyushu City (1998), Industrial History of Kitakyushu City (in Japanese). Kitakyushu City (2002), Kitakyushu Eco-town Project (in Japanese). Kitakyushu City, www.city.kitakyushu.jp/, accessed March 2003. Kitakyushu Port and Harbor Bureau (2002), Statistics of the Port of Kitakyushu 2001 (in Japanese). Nippon Steel Corporation (NSC) (2001), Beyond the Century: One Hundred Years of the Yawata Works (in Japanese), Tokyo and its CD-ROM. Nippon Steel Corporation, Tokyo, www.nsc.co.jp/, accessed March 2003. Nissan Kyushu Plant (2002), Plant Guide (in Japanese), Kanda Town, Fukuoka Prefecture.

62

Agglomeration in Asia

PPND (Pittsburgh Partnership for Neighborhood Development), Pittsburgh, PA, www.ppnd.org/, accessed March 2003. TOTO Ltd (2002), Annual Report 2002, Kitakyushu City. US Steel Corporation, Pittsburgh, PA, www.uss.com/, accessed March 2003. Yaskawa Electric Co. (2002), Company Guide 2002 (in Japanese), Kitakyushu City. Yawata Works (2002), Progress (Company pamphlet), Kitakyushu City. Zenrin Co., Ltd (2002), Company Guide, Kitakyushu City.

4. Information technology and economic growth: discovering the informational role of density Takuo Imagawa 1.

INTRODUCTION

The information technology (IT) revolution is a distinctive phenomenon in the present world. The impact of this new technology is powerful and far-reaching, aﬀecting almost every aspect of society. The convergence of computer and communications (C&C) allows us to handle all kinds of information very easily and cheaply, and it has high potential to realize a more eﬃcient economy. In the US, many people were convinced of the arrival of the so-called ‘New Economy’, since a prolonged economic boom in the 1990s driven by the IT revolution was sustained without inﬂation for a historically long period. Using the analogy of the booming US economy, it was expected that the Japanese version of the IT revolution would also be the major driving force to revive the stagnant Japanese economy. So far it seems to have been unsuccessful; however, other factors such us the bad-loan problem in the ﬁnancial sector have compounded the issue. In the meantime, the IT boom has lost its momentum, and we are even facing the collapse of the ‘IT bubble’. This tells us that we should not expect too much from a single technical innovation such as IT. Apart from the short-run ﬂuctuation of stock prices or company proﬁts, it is necessary for us to make an analysis, based on reliable logic and suﬃcient veriﬁcation, of what inﬂuences IT will have on our economy and society in the long run. The substantial changes that have occurred with the IT revolution have led to comparisons with the Industrial Revolution. In fact, the economic growth rate of Britain from the late eighteenth to the late nineteenth century is only 1.2 per cent per year,1 even though Britain enjoyed great prosperity throughout this period. Nevertheless, it is well known that the economy and society changed drastically through the British Industrial Revolution. Mass production by machinery and factories became widespread through the 63

64

Agglomeration in Asia

application of power to drive machinery, which until then had been operated by hand. It became easier to circulate inexpensive consumer goods in large quantities through the transportation network that included railroads and steamships. A big change occurred not only in the economy and industry but also in people’s lifestyle and urban structure. I believe that the same applies to the IT revolution. The penetration rates of Internet access and cellular phone subscriptions exceeded 40 and 50 per cent, respectively, by the end of 2000, and the domestic shipment of personal computers exceeded 10 million, surpassing that of color televisions.2 The production and distribution of information were popularized through the spread of communication networks and information apparatus, and a new lifestyle depending on e-mails and cellular phones prevailed. Recently, the subscribers to DSL (digital subscriber line), which realizes high-speed connection, expanded due to severe price competition, showing a rapid shift toward ‘broadband’. The technological trend sprawls out to the Internet, mobile and broadband, and IT will continue to spread steadily driven by endless innovation. This brings a big change to our life and culture as well as to politics and the economy, which can be more than anticipated driven by the ‘dog-year’ pace of innovation. It is more diﬃcult at the present stage to accurately foresee the economic and social impacts of IT on our society, compared with the managerial impacts on ﬁrms. The impact on cities is one of them. How would the urban structure be inﬂuenced by the spread of IT? This is also related to the major concern of Prime Minister Junichiro Koizumi’s structural reform plan; that is, reconstructing a new urban–rural relationship. In this chapter, I shall focus on this issue in the following three sections. The next section introduces two interesting paradoxes, which brought about the underlying motivation of my research. Sections 3 and 4 are presented to understand these paradoxes. Section 3 looks at the increasing role of cities as an information base under the knowledge economy, while Section 4 considers the impact of IT on this informational aspect of cities. Section 5 concludes with a brief discussion.

2.

TWO PARADOXES UNDER THE IT REVOLUTION

Productivity Paradox When ﬁrms make an IT investment, managers often introduce into corporate management ‘three-letter’ concepts such as SCM (supply chain management), CRM (customer relationship management) and ERP (enterprise resource planning). These concepts required for ‘knowledge management’

Information technology and economic growth

65

are so popular that it is usually expected that management improvement through an increase in eﬃciency or productivity should be realized if investment in information systems is carried out along with instruction in those concepts. In fact, the amount of IT investment in Japan showed a remarkable growth in the late 1990s even under the sluggish economy, and this trend has expanded from the manufacturing to the non-manufacturing sectors.3 The problem is, however, is IT investment really connected with an improvement in corporate management? In the US, Robert Solow, the Nobel Laureate, pointed out in 1987: ‘You can see the computer age everywhere these days, except in the productivity statistics’. Since then, the absence of productivity increase in the statistics has been called the ‘productivity paradox’ or the ‘Solow paradox’, and has caused a long-lasting controversy for over a decade. The paradox implies that spending on IT investment is just meaningless. For the US, which enjoyed overwhelming economic performance driven by the Internet, represented by high-tech companies such as Microsoft, Cisco Systems, Sun Microsystems and Yahoo, it seems to be beyond doubt that the increasingly accumulated IT investment was the major engine of US economic growth in the 1990s. Therefore, it is truly surprising that this linkage cannot be veriﬁed in US statistics. Although the productivity paradox continues to puzzle economists, a slight diﬀerence by industry came out of the trend of productivity after the second half of the 1990s. Figure 4.1 shows transition of labor productivity and the amount of IT investment in the US. The IT investment of the private sector began to increase from the second half of the 1970s, and soared in the 1990s, which is consistent with the ﬂourishing IT investment by US companies in those years. On the other hand, the labor productivity of the private sector (non-farm business) has shown little divergence from the long-run trend as stated by the productivity paradox. However, when we narrow down to the manufacturing sector, its productivity in the late 1990s clearly increased away from the trend at a pace that might pull up the whole economy’s productivity. If we interpret the manufacturing sector as ‘IT-producing industry’ and the non-manufacturing sector as ‘IT-using industry’, the former has been enjoying the beneﬁt of the IT revolution through the supply of IT-related products, while the latter has not even though it committed a considerable amount of IT investment. Therefore, it is the IT-using industry that is being trapped by the paradox. Although there are various interpretations of this phenomenon, a common view is that managerial eﬀorts such as a reform of corporate management or organization, improvements in employees’ skills and so on, are complementary to IT investment in order to raise productivity. For example, Brynjolfsson and Hitt (1998) analysed the micro data across

66

55/Q1

60/Q1

65/Q1

Real private IT investment (right)

70/Q1

Labor productivity for manufacturing (left)

Labor productivity for non-farm business (left)

Year/quarter

75/Q1

80/Q1

85/Q1

90/Q1

95/Q1

0 00/Q1

100

200

300

400

500

600

Billions of chained (1992) dollars

Figure 4.1

Labor productivity and IT investment in the US

Sources: Labor productivities for ‘manufacturing’ and ‘non-farm business’ are from ‘output per hour of all persons’ (US Bureau of Labor Statistics). ‘Real private IT investment’ is from ‘information processing equipment and software’ in ‘private ﬁxed investment’ (Bureau of Economic Analysis).

0 1950/Q1

20

40

60

80

100

120

140

160

Index (1992 = 100)

Information technology and economic growth

67

companies, and concluded that among those which advanced IT investment, the winners that gained a productivity rise are companies which simultaneously carried out management reform such as ﬂattening organization and simplifying decision-making processes, or made eﬀorts to increase employees’ skill by training or team activity. The productivity of the computer itself has been progressing very rapidly due to ‘Moore’s Law’ with improvements in the processing device and continued reduction in price. Certainly the potential ability of whitecollar workers has risen by the utilization of e-mail and Internet reference, while at the same time the capability of customer and inventory management has increased by establishing information networks. IT is just a highly eﬃcient ‘tool’ and IT itself will not do anything until a well-reﬁned purpose and speciﬁc plan are given. The problem is how we use it, not how much we spend on it. Therefore, we can conclude that IT investment is not an easy task, and we never get anything from IT without wisdom and eﬀort. Accumulation Paradox Now, I shall turn to another paradox. One of the changes that IT introduction may bring about is ‘decentralization’. There are many examples: the spread of the distributed networks represented by Gnutella, and ‘bluetooth’; popularization of information dispatch by individuals using homepages or mailing list services; decentralization of economic functions observed in the appearance of ‘eco-money’ and NPOs (non-proﬁt organizations); a shift in the industrial structure toward a horizontal relationship by reorganizing vertical keiretsu and complex distribution systems; ﬂattening the management organization by removing middle-class managers and the traditional seniority system; revitalizing local economies accompanied by the increase in telecommuting and SOHO (small oﬃces, home oﬃces); and so on. It is expected that our society will move toward a more democratic and distributed network-based system from a traditional centralized one. It sounds convincing that IT will realize a decentralized society, since the spread of new information and communication media such as the Internet and mobile devices should release us from restrictions of distance, place and time. In a ubiquitous society, which means that anyone can receive and transmit information anytime, anywhere, we shall be completely free to move. With a single personal computer associated with a telephone line we can easily live in the suburbs; working, shopping, eating and even property management can be done without moving. It is no longer necessary to live in a crowded city where we suﬀer from extremely expensive land prices and jam-packed commuter trains.

68

Agglomeration in Asia

However, at least for the moment, the potential of decentralization powered by IT is connected neither with revitalization of local areas nor with decentralization of the population or economic activities. Instead, a reverse phenomenon can actually be observed. Figure 4.2, constructed from the Population Census 2000, looks at the trend of population in Japan. Facing the arrival of the aging society and falling birth rate, the growth rate of population hit its lowest level after the Second World War. Nevertheless, the increases in population of the three major metropolitan areas (Tokyo, Nagoya and Osaka) achieved a total of 1.22 million persons in the 1995–2000 period, and accounts for 90 per cent of the population increase of the whole country. In particular, four prefectures of Tokyo metropolitan area (Tokyo, Kanagawa, Chiba and Saitama) achieved a more than 2 per cent increase, and the population of the Tokyo center area (23 wards) increased steadily from 1997. Centralization and concentration in cities has actually accelerated since the end of the twentieth century, when the increasing familiarity with IT should have removed the restrictions of distance and location.4 Other evidence can be seen in Table 4.1, where I simply measured shares of the Tokyo metropolitan area both in general economic and IT-related activities. At a glance, IT-related activities are conﬁrmed to be generating more concentration in all fundamental economic indicators including persons, oﬃces and outputs. Actually, the high-tech IT frontrunners which make most use of the Internet and mobile tools are known to gather in the same place; symbolic models are ‘Silicon Valley’ in the US or ‘Shibuya Bit Valley’ in Tokyo. According to the investigation of IT-related industries performed by the Ministry of Land, Infrastructure and Transport (Figure 4.3), a considerable number of IT-related ﬁrms accumulate within walking distance of terminal railroad stations in Tokyo, Nagoya and Osaka, such as Akihabara and Shin-Osaka. Overwhelmingly, almost 30 per cent are located within Tokyo’s 23 wards. It is therefore inevitable to recognize such an accumulation tendency as a signiﬁcant feature of IT-related economic activity. In addition, there are signs indicating the accelerating tendency of ‘back to the city center’. For instance, the construction of skyscraper buildings for oﬃces and residences, and the establishment of graduate schools in the heart of Tokyo have been conspicuous in recent years. In this context, there might be a circular impact of IT on cities; that is, concentration of IT-related activities may stimulate agglomeration of other economic activities, which fosters further accumulation of IT-related activities, generating an autonomous growth process. Taking all the above into account, it is hardly successful to discover data which support the story that the progress of IT cancels overconcentration in Tokyo or promotes decentralization in local areas, and this is why it is called the accumulation paradox.

69

1.36mil. m 1.36

Tokyo Metropolitan 0.84 m 62.1%

Population change in Japan, 1995–2000

Ministry of Internal Aﬀairs and Communications (2001).

Figure 4.2

Source:

Osaka Metropolitan 0.18 m 13.5%

Nagoya Metropolitan 0.20 m 14.6%

Miscellaneous 9.8%

Population increase

Below –1% –1 to 0% 0 to 1% 1 to 2% Above 2%

Population growth rate by prefecture

70

Agglomeration in Asia

Table 4.1

Tokyo Metropolitan’s share in IT-related activities General economic activities

IT-related activities

Consumers

26.1% (population, 2000)

30.1% (cellularPHS subscribers, 2000)

Labor force

27.5% (persons engaged (private), 1999)

57.5% (persons engaged (private) in information service industries, 1999)

Establishments

23.8% (establishments (private), 1999)

43.5% (establishments (private) in information service industries, 1999)

Output

30.7% (gross prefectural domestic product, 1998)

66.1% (annual sales in information service industries, 1999)

Note: Numbers are shares of the Tokyo metropolitan area (Tokyo, Kanagawa, Chiba and Saitama) in all prefectures in Japan; PHS denotes the personal handy-phone system. Sources: Basic Resident Registers (2000); Ministry of Post and Telecommunications (March 2000); Establishment and Enterprise Census (1999); Ministry of International Trade and Industry (1999); Prefectural Accounts (1998).

Tokyo’s 23 wards 27.6% Other cities 46.6%

Osaka City 7.5%

Nagoya City 3.5% Yokohama City 3.2% Kyoto City 1.0% Fukuoka City 2.7% Kawasaki City 1.1% Sapporo City 2.7% Kobe City 1.2% Hiroshima City 1.4% Sendai City 1.5% Source: Ministry of Land, Infrastructure and Transport (2001).

Figure 4.3

Location of IT-related ﬁrms

Information technology and economic growth

71

In the next two sections, I shall consider two factors in order to understand this accumulation paradox. First, I propose to reconsider the role of the cities in the knowledge economy.

3. INCREASING ROLE OF DENSITY UNDER THE KNOWLEDGE ECONOMY Information Spillovers to Bridge Endogenous Growth and Agglomeration In the literature of economics it is known that the accumulation of knowledge incorporating high externalities serves as the driving force of economic growth. This is the contribution of a new theory called the ‘endogenous growth theory’, which established a rigorous theoretical link between growth and knowledge. A variety of empirical studies stimulated by this theory ﬂourished, and conﬁrmed using international or local data that the initial conditions of knowledge accumulation represented by, for example, the educational level, signiﬁcantly inﬂuence economic growth later on. The development of this new theory inspired people’s interest in knowledge spillover. Along with a journalistic attention to industrial clustering such as Silicon Valley in California or Austin in Texas, economists’ major concern shifted toward agglomeration economies. It is a process in which proximity within cities fosters more knowledge spillover, and seemingly unrelated existing ideas mix up randomly to generate an important idea under the high-density urban spot, thereby enhancing knowledge production. In order to understand this externality, it is useful to see two sides of information ﬂow. When we consider the information diﬀusion process, it is necessary to classify information into ‘disembodied’ and ‘embodied’ information. The former describes information such as a document or a video, which is easily saved and shared. The latter, on the other hand, takes stock of an individual’s experience, know-how and so on, which is very diﬃcult to transfer and share. In principle, disembodied information is, once disclosed, available to anyone as homogeneous or uniform information since it is described. It is therefore useful to utilize IT in the form of mass media and the Internet to share the same information among a large number of people. Conversely, embodied information is highly relation speciﬁc and diﬀerentiated or heterogeneous since it is diﬃcult to describe. Therefore, face-to-face contact is indispensable to transfer this type of information. We all know that neither experience nor know-how can be easily inherited. Truly critical information will be undisclosed, and instead transmitted conﬁdentially depending on human relations.

72

Agglomeration in Asia

A city is a place where companies and residents interact closely, and it functions as an information base that oﬀers face-to-face exchange very cheaply. Although disembodied information may be shared over distance by use of IT, embodied information requires meetings to be transferred and diﬀused. In fact, it is the embodied information that plays the most important role in creating knowledge, and serves as the origin of competitiveness. Cities internalize this process whereby embodied information is automatically shared through proximity, and the probability of receiving beneﬁts from externalities will be increased by being located in an urban center. Although there is little research that measures information ﬂows within a city, a few works have tried to show empirically that advanced and technical information such as patents hardly transfers over an area.5 In the next subsection, I shall introduce the essence of my previous empirical work, which showed that urban residents make use of more meetings and twoway communication tools per capita. Indeed, 60 per cent of mail transactions and 80 per cent of telephone calls, both of which should eliminate constraints by distance, surprisingly terminate within the originating prefecture. It is still not easy in rural areas to acquire important information appropriately, even though IT has progressed and become so easy to use. Moreover, cities incorporate a variety of companies and talented people gathering in the neighborhood, and its diversity brings a meaningful eﬀect to the accumulation of knowledge. For example, since the urban labor market has a large scale, specialization progresses, and the mismatch of skills is easily solved, leading to more eﬃcient employment. Urban workers may accumulate their specialized experience through the diversity and the large scale of business available in the metropolitan job market. The existence of various companies and industries also provides ‘insurance’ against risks such as layoﬀ or economic downturn. In sum, since cities possess predominance at various points in terms of knowledge accumulation, it is highly possible that cities function as a knowledge base and consequently as the engine of economic growth as the endogenous growth theory suggests. Such a mechanism becomes increasingly important under the knowledge economy enhanced by IT, and we are facing an age where city planning may condition the possibility of economic growth. Empirical Evidence of Cities Behaving as an Information Base Estimation strategy In this subsection, I provide empirical evidence of the role of urban cities as an information base stimulating information ﬂows. For this purpose, I employ a nationwide data set of communication usage in Japan that provides direct measures of the utilization of various communication tools

73

Information technology and economic growth

Table 4.2

Communication tools analysed in this study

Technology

Tools

Novelty

Properties

Information content

One-way communication

Newspaper

Traditional (since 1869*) Relatively new (since 1952) Modern (since 1973)

Transcribed, time-lag Electronic, simultaneous Electronic, more variety, regional

Homogeneous

Satellite TV

Modern (since 1989)

Electronic, more variety, regional

Mail

Traditional (since 1871*) Relatively new (since 1890) Modern (since 1979)

Transcribed, time-lag Electronic, simultaneous Electronic, simultaneous, portable Very ﬂexible (with visuals, etc.)

TV Cable

Two-way communication

Telephone Cellular Phone Meeting (face to face)

Traditional

Homogeneous Homogeneous (some specialization) Homogeneous (some specialization) Heterogeneous Heterogeneous Heterogeneous

Highly heterogeneous

Note: * Started oﬃcially as approved by the government. These were available unoﬃcially even before the year shown.

across 47 prefectures. The variety of communication tools in the data allows us to diﬀerentiate the types of communication technology. In particular, I classify communication methods into one- and two-way technologies as described in Table 4.2. One-way communication tools (such as televisions) allow people to extract information in a way that its ﬂow goes in only one direction, while two-way communication tools (such as telephones) facilitate the interactive exchange of information. The presence or absence of interaction in these two types of technologies yields two major distinctions. First, interaction in two-way technologies embodies positive externalities. A person will receive utility gains when someone else joins the communication network (that is, network externalities), and when someone makes contacts at his/her expense (that is, call externalities). Eﬃciency gains from ﬂexible information exchange may also be achieved, since two-way interactive contacts should reduce communication loss.6 These externalities are potentially important in urban location decisions, as large city size and better information infrastructure will help the urban community to beneﬁt more from two-way communication.

74

Agglomeration in Asia

Second, presence of interaction aﬀects how information is transmitted. One-way technologies allow mass distribution of homogeneous information, but two-way technologies require peer-to-peer networking since the information transmission will be private or relation speciﬁc. The information transmitted through the two-way system is then diﬀerentiated by place and person of origin. This heterogeneity in information content is likely to require search costs to obtain the designated information, and the matching of communication partners is conditioned by spatial proximity or preexisting ties. Therefore, resulting informational transactions through the two-way system can be geographically localized in high-density regions. Considering these distinctions between one- and two-way technologies, leads us to the following hypothesis: The use of two-way communication is associated with cities, while that of one-way communication is not. Using communication data from Japan, I investigate this prediction on the relationship between communication technology and cities. Data summary Table 4.3 describes the data summary of demographic variables, and oneand two-way communication tools. All data are based on the aggregate prefecture measures for the 1988–94 period,7 and include penetration measures (subscribers) and/or utilization measures (volume of contacts). The eighth column shows the Herﬁndahl concentration indices across prefectures. Compared to the distribution of population and households, cable subscribers are more unequally distributed, while newspaper, television and satellite television subscribers are evenly distributed.8 Unfortunately, no exact information about the utilization of these one-way communication tools is available. The only available measure is the average hours per day spent on watching TV, listening to the radio and reading newspapers or magazines. This is derived from a survey conducted every ﬁve years by the government, and should be interpreted with caution although it is used in my estimation for the purpose of a rough comparison. For two-way communication, I have utilization measures for all tools derived from the origin–destination (OD) matrix data across prefectures. As a crude proxy for face-to-face meetings, I use the number of passengers carried by railways, motor vehicles, vessels and aircraft, to be meant as the number of trips. The Herﬁndahl indices show that both subscribers and usage of all two-way communication tools are highly concentrated compared to population, which is clearly distinguished from that of one-way tools. The use of mail is heavily concentrated, and shows asymmetric

75

Newspaper (88–94)

One-way communication

Two-way communication

Population (88–94) Households (88–94)

Demographic variables

7710 4139

Subscribers

55 714

54 363

19 405 793

33 940

51 733

41 822

Households

Subscriber (circulation copies) Subscriber (public channels) Subscribers

123 991

Quantity (1000 or 1000hr) average 88–94

Population

Unit

Number of letters (accepted) OD Number of matrix3 letters (a (88, 91, 94) sample day) Telephone Subscribers (88–94) (local network)

Mail (88–94)

Cable (88–94) Satellite TV (88–94)

TV (88–94)

Variable and sample period

Category

Table 4.3 Data summary

2.95

2.81

40.38

10.04

1.08

0.65

1.46

0.30

Annual growth (%) annual average

0.479

0.053

0.082

0.280

0.420

Penetration rate (per capita) 1994

0.491

0.445

Utilization per capita per day1 1994

0.049

0.075 (0.125)

0.128

0.037

0.067

0.039

0.047

0.044

0.039

Herﬁndahl index of concentration2 average 88–94

58.29

Ratio of intra-prefecture contacts (%) average 88–94

76

Unit

OD matrix Number of (88–94) calls Minutes of calls Cellular Subscribers Phone (92–94) OD matrix Number of (92–94) calls (92–93) Minutes of calls Meeting Passengers4 (OD matrix) (carried) (88–94)

Variable and sample period

(continued)

45.62 2.18

61 539 79 259 293

61.64

1593

44.31

4.14

3 557 185

2 531 317

3.35

Annual growth (%) annual average

76 199 976

Quantity (1000 or 1000hr) average 88–94

0.035

Penetration rate (per capita) 1994

0.092 (2.642) 8.257 (238.3) 1.818

1.824 (3.806) 310.8 (648.3)

Utilization per capita per day1 1994

Notes: 1. Unit for duration of calls is seconds. Utilization per subscriber per day in 1994 is in parentheses. 2. Concentration indices from outgoing measures, while those from incoming measures in parentheses. 3. OD matrix refers to the data from the origin–destination matrix across 47 prefectures. 4. Passengers carried by railways, motor vehicles, vessels and aircraft.

Category

Table 4.3

0.068 (0.074) 0.076 (0.081) 0.055 (0.055)

0.055 (0.055) 0.056 (0.056) 0.080

Herﬁndahl index of concentration2 average 88–94

90.75

72.20

78.85

76.48

81.59

Ratio of intra-prefecture contacts (%) average 88–94

Information technology and economic growth

77

patterns of concentration between outgoing and incoming mail. Telephone subscribers are marginally more concentrated than households, while the use of telephone is much more concentrated. Cellular phone subscribers are heavily concentrated, but its usage is slightly more dispersed than its subscriber base. Face-to-face meetings are distributed similarly to telephone use. In telephone and face-to-face meetings, outgoing and incoming are symmetric, while a little asymmetry is observed in the use of cellular phones. The last column shows the ratio of intra-prefecture contacts for each twoway tool. Among all the contacts, about 58 per cent by mail, 82 per cent (76 per cent in minutes) by telephone call, 79 per cent (72 per cent in minutes) by cellular call, and 91 per cent by face-to-face meetings are made within the same prefecture. This is somewhat surprising. While it is usually believed that communication technologies eliminate the necessity for geographic proximity, about 80 per cent of telephone calls do not cross the prefecture border. This geographic localization of two-way communication contacts may be interpreted as supporting my previous argument that twoway contacts are conditioned by spatial proximity and pre-existing ties. This simple data summary itself provides suggestive evidence. With the exception of cables, all one-way tool subscribers are evenly distributed given the distribution of population across cities, but all two-way tool subscribers are unequally distributed. Usage of all two-way communication tools is even more unevenly distributed. Thus, we can predict that two-way communication is more associated with cities than one-way communication is, supporting our hypothesis. Empirical examination Next, I test more formally the linkage between communication use and cities using prefecture-level density. After some theoretical consideration,9 I estimated the following linear equation: C ln Nit 1 ln Xit 2 ln Dit 3 ln Li i t it,

(4.1)

it

where Cit is total communication use; Nit is population; Xit is a vector of explanatory variables; Dit is population density; Li is a vector of prefecture location variables; i is a prefecture random eﬀect; 1 is a year dummy; it is a random error term, and i and t denote prefecture and year, respectively. As a dependent variable, penetration measures (subscribers divided by population) and/or utilization measures (contacts per capita) are used. As explanatory variables, I use average prefecture income per capita and business activity ratio, since communication for business purpose can be

78

Table 4.4

Agglomeration in Asia

Communication use and population density

Independent variables

One-way communication Penetration measures

Utilization measures

Newspaper TV Cable Satellite subscribers subscribers subscribers subscribers

Intercept Real income per capita Percent of business Density Density in habitable land Location Proximity to Tokyo Proximity to Osaka Year dummies R2 No. of obs. N T Speciﬁcation test

NewspaperTV Radio hours spent per day Weekdays

Weekend

0.931** (0.025) 0.139** (0.044) –0.001 (0.002)

5.854** (0.113) 0.004 (0.012) 0.002** (0.001)

4.882** (1.313) –0.188 (0.253) 0.008 (0.014)

1.912** (0.975) 0.147 (0.202) 0.001 (0.011)

0.883** (0.391) 0.091 (0.107) 0.004 (0.003)

0.720 (0.524) 0.132 (0.143) 0.001 (0.004)

0.003 (0.025)

0.078** (0.018)

0.278 (0.174)

0.127 (0.097)

0.002 (0.020)

0.013 (0.027)

0.017* (0.010) 0.018* (0.010) Yes 0.020 329 47 7 38.77 (0.000)

0.052 (0.080) 0.117 (0.077) Yes 0.382 329 47 7 10.17 (0.337)

0.009 (0.052) 0.003 (0.050) Yes 0.831 282 47 6 33.9 (0.000)

0.014 (0.012) 0.005 (0.010) – 0.231 47 47 1 – (OLS)

0.004 (0.016) 0.013 (0.013) – 0.251 47 47 1 – (OLS)

0.039** (0.013) 0.042** (0.013) Yes 0.578 329 47 7 9.81 (0.366)

Notes: Estimates are based on the random eﬀects GLS, unless otherwise noted. All variables except percentages and dummies are expressed in natural logarithms. All dependent variables are per capita measures. Standard errors are in parentheses. * Signiﬁcant at the 10% level. ** Signiﬁcant at the 5% level. Speciﬁcation test is Hausman’s (1978) speciﬁcation test, reporting chi-square values (P-value in parentheses).

diﬀerent from that for residential use.10 As prefecture location variables to describe an advantage from living close to cities, I introduced proximities (that is, distance–1) to the two-largest prefectures, Tokyo and Osaka. Estimation is performed by the feasible generalized least squares (GLS) procedure (the random eﬀects model). Table 4.4 presents the basic results on communication use and population density. The results ﬁrmly support our hypothesis. Both in penetration

Information technology and economic growth

79

Two-way communication Penetration measures

Utilization measures

Telephone Cellular subscribers subscribers

Mail Telephone Cellular Meeting letters passengers accepted Number Minutes Number Minutes outgoing outgoing outgoing outgoing outgoing

5.808** (0.147) 0.085** (0.016) 0.004** (0.001)

3.275** (1.496) 1.406** (0.397) 0.031** (0.012)

0.087** (0.023)

0.281** (0.087)

0.007 (0.013) 0.005 (0.013) Yes 0.497 329 47 7 4.15 (0.901)

0.062 (0.050) 0.065 (0.043) Yes 0.857 141 47 3 3.63 (0.604)

3.868** (0.661) 0.196 (0.127) 0.010 (0.007)

0.251** 0.181** (0.076) (0.036)

0.029 (0.040) 0.013 (0.039) Yes 0.557 329 47 7 13.55 (0.139)

6.214** (1.709) 1.711** (0.448) 0.003 (0.014)

5.391** (0.422) 0.249** (0.109) 0.004 (0.004)

0.212** 0.303** 0.224** (0.038) (0.103) (0.099)

0.102** (0.028)

5.243** 3.033** (0.355) (0.409) 0.063 0.152* (0.075) (0.091) 0.023** 0.014** (0.004) (0.004)

0.046** 0.006 (0.019) (0.021) 0.039** 0.012 (0.018) (0.020) Yes Yes 0.565 0.596 329 329 47 47 7 7 0.32 21.15 (1.000) (0.012)

1.947 (1.689) 1.218** (0.440) 0.018 (0.014)

0.059 0.027 0.020 (0.059) (0.058) (0.016) 0.102** 0.113** 0.012 (0.052) (0.050) (0.014) Yes Yes Yes 0.798 0.774 0.669 141 94 329 47 47 47 3 2 7 17.37 6.32 1.72 (0.004) (0.176) (0.995)

and utilization measures,11 all of the density coeﬃcients for two-way tools are signiﬁcantly positive even at the 1 per cent level, while they are not signiﬁcant or negative for one-way tools.12 This striking distinction survives the alternative use of density measures.13 Among the two-way tools, the cellular phone is the most ‘urbanized’ tool. Income elasticities are signiﬁcantly positive for the cellular phone, exceeding unity, and also the density coeﬃcients are the largest for this tool. Location variables showing proximity to the two biggest prefectures are not inﬂuential in most of the cases. However, those coeﬃcients are signiﬁcantly positive for two of the one-way tools, that is, newspaper and TV. This implies concentration of communication use in the largest cities, but it is associated with centralization rather than with population density itself.

80

Agglomeration in Asia

So far, the right-hand-side variables are treated as exogenous. Of all the right-hand-side variables, this assumption is most likely to fail for the density variable. Individuals or ﬁrms which are more inclined to interact may decide to cluster in cities. Cities with random factors that make them more interactive (for example, climate or geographic features suited to communication, or location choice by some inﬂuential ﬁrms and so on) may attract incomers. To allow for this endogeneity, I use two sets of instruments and examine how exogenous changes in density alter the communication usage: 1.

2.

Agglomeration in the nineteenth century: (i) population; (ii) population density; and (iii) the ratio of population living in cities with more than 30 000 population, all in 1890. Transportation infrastructure in the nineteenth century: (iv) a dummy that takes a value of one if a railroad was present in 1890; (v) a dummy that takes a value of one if a port for international trade was present in 1890; and (vi) a dummy that takes a value of one if the prefecture is on the main island.

(i) through (v) are derived from the Résumé of Statistics of the Japanese Empire (Bureau of Statistics 1892). These instruments are selected based on the identifying assumptions that the earlier sources of agglomeration and the status of transportation infrastructure in the nineteenth century have a remaining inﬂuence on the preference of people regarding where to locate, and that they are not related to modern diﬀerences in communication use not explained by our model. The last instrument is added as an exogenous geographic determinant to express potential advantages in terms of transportation. Table 4.5 presents the estimation using the instruments for the averages over the sample period (between regressions) and for the whole panel (random eﬀects). These experiments do not change our conclusion. Most of the density coeﬃcients for two-way technologies are signiﬁcantly positive, while those for one-way technologies are insigniﬁcant or negative.14 Therefore, the data strongly support our hypothesis on the positive linkage between two-way communication and cities. Implications I found that two-way communication technologies, but not one-way communication technologies, are strongly associated with urban agglomeration. In particular, the data show a robust positive linkage between the per capita usage of two-way communication and population density, suggesting a more than proportional amount of information exchange within cities.

Information technology and economic growth

81

Also, two-way technologies are positively connected to various service industries as well as to communication-intensive employment such as sales or services. The empirical evidence highlights the urban role in fostering information ﬂows in an ‘interactive’ form. Although a communication device is generally believed to be an eﬀective tool to overcome distance, our evidence suggests that the development of ‘interactive’ communication technologies may make cities more attractive.

4.

IT AND INFORMATION SPILLOVERS

Face-to-face Interactions Enhanced by IT The second factor to understand this ‘accumulation paradox’ is the interdependence between meeting face-to-face (FTF) and electronic exchange using IT. We communicate with others either by telecommunications or by visiting them. Thanks to technological innovation, we have obtained a variety of communication tools one after another, such as telephone, fax, cellular phone, e-mail and Web. In general, the traditional view (substitution) will be intuitive and appealing; that is, meetings will become unnecessary if we use communication media to realize much easier and more eﬃcient interaction. On the other hand, another view (complementarity) is also supported; that is, human interaction will always need human contact regardless of how electronic communication devices may progress. Again, there is very little empirical research in this ﬁeld, mainly due to the lack of quantitative data. However, in the next subsection, I shall introduce my empirical work, which utilized reliable data on telecommunications and transportation. It can be shown that even though substitution of visits by phone increases with the distance between communication partners, complementary eﬀects always exceed substitution eﬀects so that phone usage and transportation are net complements. In other words, if telecommunications usage becomes active along with the progress of IT, FTF interactions would also increase. In fact, several times I have found that browsing a homepage and sending an e-mail has created a brand new communication relationship, which has led to mutual visits and the launching of a joint research project. We should recognize that electronic exchange by IT is very likely to induce, complement and strengthen faceto-face interaction rather than make it unnecessary. Contact is essential to human relationships, especially when creative work is involved. There are always real people at both ends, and virtual relationships without any real connection is often unproductive or even dangerous. A creative idea is born from discussions with a peer who shares an interest. The same applies to

82

Agglomeration in Asia

Table 4.5 Density coeﬃcients: estimates with exogenous and endogenous density Estimation method

One-way communication Penetration measures Newspaper subscribers

TV subscribers

Density treated as exogenous a. OLS 0.012 0.009 (between) (0.028) (0.024) b. GLS 0.003 0.078** (random (0.025) (0.018) eﬀects) Density treated as endogenous c. IV-1 0.025 (between) (0.036) d. IV-2 0.024 (between) (0.035) e. GLS/IV-1 0.019 (random (0.070) eﬀects) f. GLS/IV-2 0.020 (random (0.064) eﬀects) No. of obs. a, c, d b, e, f

47 47 7

Cable subscribers

Utilization measures Satellite subscribers

NewspaperTV Radio hours spent per day Weekdays

Weekend

0.204 (0.166) 0.278 (0.174)

0.194** (0.096) 0.127 (0.097)

0.002 (0.020)

0.013 (0.027)

0.026 (0.031) 0.020 (0.031) 0.043 (0.060)

0.029 (0.217) 0.071 (0.212) 0.015 (0.467)

0.174 (0.124) 0.220* (0.122) 0.091 (0.741)

0.004 (0.038) 0.005 (0.025)

0.030 (0.004) 0.027 (0.035)

0.017 (0.053)

0.096 (0.426)

0.314 (0.640)

47 47 7

47 47 7

47 47 6

47

47

Notes: Only the density coeﬃcients are reported. Equations for GLS are the same as in Table 2.4, while those for OLS and IV are without year dummies. Instrumental variables are: (1) 1n (population in 1890); (2) 1n (population density in 1890); (3) the ratio of population in cities with more than 30 000 population in 1890; (4) a dummy for presence of a railroad in 1890; (5) a dummy for presence of a port for international trade in 1890; (6) a dummy for the prefectures in the main island. (1)–(3) are used for IV-1 and GLS/IV-1, and (1)–(6) are used for IV-2 and GLS/IV-2. Standard errors are in parentheses. Signiﬁcant at the 15% level, * Signiﬁcant at the 10% level, ** Signiﬁcant at the 5% level.

‘trust’ which is indispensable in business dealings. The foundation of human interactions is eventually based on a real relationship, and a virtual relationship is most useful when it is used as a ‘tool’ to supplement the real one.

83

Information technology and economic growth

Two-way communication Penetration measures

Utilization measures

Telephone Cellular subscribers subscribers

Mail Telephone Cellular Meeting letters passengers accepted Number Minutes Number Minutes outgoing outgoing outgoing outgoing outgoing

0.098** (0.032) 0.087** (0.023)

0.253** (0.086) 0.281** (0.087)

0.269** 0.178** (0.084) (0.039) 0.251** 0.181** (0.076) (0.036)

0.151** 0.262 0.171** (0.040) (0.100) (0.098) 0.212* 0.303** 0.224** (0.038) (0.103) (0.099)

0.097** (0.030) 0.102** (0.028)

0.101** (0.042) 0.094** (0.041) 0.097** (0.044)

0.189* (0.112) 0.199* (0.110) 0.202* (0.119)

0.360** 0.187** (0.110) (0.051) 0.343** 0.183** (0.107) (0.050) 0.400** 0.178** (0.127) (0.050)

0.141** 0.192 (0.052) (0.130) 0.136** 0.214* (0.051) (0.127) 0.120 0.189 (0.131) (0.172)

0.031 (0.130) 0.057 (0.126) 0.100 (0.147)

0.090** (0.038) 0.092** (0.038) 0.089** (0.041)

0.077* (0.040)

0.223* (0.115)

0.351** 0.174** (0.118) (0.047)

0.154 (0.121)

0.242 (0.165)

0.150 (0.142)

0.096** (0.040)

47 47 7

47 47 3

47 47 7

47 47 3

47 47 2

47 47 7

47 47 7

47 47 7

Empirical Evidence of Complementarity between Face to Face and IT Estimation strategy In this subsection, I attempt to test the controversial relationship between telecommunications and transportation. First, although telecommunications usage is usually assumed to be a substitute for FTF interactions, it may also function as a complement. A simpliﬁed theoretical model of the utilitymaximization problem is able to demonstrate that the net impact of telecommunications technology on transportation demand may be either positive or negative, depending on the relative frequency of substitutable and complementary forms of telecommunications usage in an economy.15 Here I merely present a simulation result from the model in Figure 4.4, which is very intuitive. The substitutable (complementary) form increases (decreases) with

84

0

200

400

600

800

1000

1200

0

1

2

3 4 5 Transportation cost (increases with distance)

F

6

T TS

7

TC

8

Figure 4.4

Simulation results of the interdependence of telecommunications and transportation

Note: F: Traﬃc volume of face-to-face contacts (transportation); T: Traﬃc volume of telecommunications contacts (T TS TC); TS: Traﬃc volume of a substitutable form of telecommunications usage; TC: Traﬃc volume of a complementary form of telecommunications usage

Traffic volume of telecommunications or transportation

Information technology and economic growth

85

distance, since transportation costs relative to telecommunications costs rise with distance. Therefore, telecommunications and transportation are likely to be net complements for short distances, and net substitutes for long distances. For empirical purposes, I employ a dataset that is similar to the one used in Section 3. Telecommunications (telephone and cellular phone), transportation (as a crude proxy for the number of trips approximating the number of FTF contacts), and reference variables (mail, commodities and freight) are available based on the OD matrix across 47 prefectures. Treating each entry of the OD matrix as an independent observation, we obtain 47 47 observations to express the volume of communication ﬂow across prefectures.16 The data on telephone and transportation are available for 1988–94, while other data are available for limited periods. Prefecture characteristics variables such as population and income are obtained from a panel of prefectures over the 1988–94 period. In the rest of this chapter, I use the values averaged over the years for which necessary data are available. Data summary Table 4.6 reports the distribution of communication contacts by the distance between prefectures.17 On average over the sample period, about 82 per cent of telephone calls, 79 per cent of cellular calls (both in the number of calls) and 91 per cent of trips terminated within the same prefecture. Although telecommunications is generally believed to eliminate the necessity for geographic proximity, about 80 per cent of phone messages do not cross the prefecture border. Inter-prefecture contacts account for only 20 per cent of phone calls and 10 per cent of trips, and the share of each distance segment declines as the distance rises. ‘Resistance’ by distance will be stronger for FTF interactions because their distribution is more localized than that of phone contacts. Phones cover greater areas than visits do. It is consistent with the model in the sense that more substitution of trips by telecommunications takes place for long distances. Table 4.6 also compares the distribution of communication contacts with that of mail, commodities and freight. Interestingly, both phone messages and trips are geographically more localized than the other three ﬂows are. For example, 89 per cent of telephone calls, 90 per cent of cellular calls and 97 per cent of trips, compared with 69 per cent of mail, 85 per cent of commodities18 and 82 per cent of freight terminate within 50 kil-ometers. This is counterintuitive. Phones allow electronic and instantaneous transmission of voice messages, whereas mail, commodities and freight involve costly and delayed shipment of documents or goods over distance. In this sense, we expect that phone messages should be less conditioned by geographic proximity. I ﬁnd the opposite results in the Japanese data. This observation is, however, consistent with the model’s predictions. If phone

86 76.1 47

Freight (90)

Prefecture pairs*

69.0 (12.3) 84.8 (6.6) 81.8 (5.7) 38

97.4 (6.6)

90.4 (11.5) 86.5 (14.3)

88.9 (7.3) 86.1 (9.6)

–50km

73.0 (4.0) 88.1 (3.3) 85.9 (4.1) 100

98.7 (1.3)

93.7 (3.3) 90.7 (4.2)

91.5 (2.5) 89.0 (2.9)

50–100km

78.3 (5.3) 92.0 (3.9) 91.2 (5.3) 282

99.6 (0.9)

96.5 (2.8) 94.4 (3.7)

94.4 (2.9) 92.5 (3.5)

100–200km

86.8 (8.5) 95.4 (3.5) 95.6 (4.4) 558

99.8 (0.3)

98.2 (1.8) 97.1 (2.7)

97.1 (2.7) 95.9 (3.4)

200–400km

Distance between prefecture pairs

Notes: The values in the table are the ratios of cumulative distribution (incremental ratios in parentheses). * Prefecture pairs (i, j) and (j, i) are counted separately; 2,209 (47 47) prefecture pairs in total.

78.2

Commodities (90)

56.7

72.2

Minutes of calls (88–94)

Reference variables Mail (88, 91, 94)

78.9

2. Cellular phone Number of calls (88–94)

90.8

76.5

Minutes of calls (88–94)

Transportation variable 3. FTF meetings Number of passengers (88–94)

81.6

Intra-pref.

Telecommunication variables 1. Telephone Number of calls (88–94)

Variables

Table 4.6 Distribution of communication contacts by distance (%)

95.2 (8.5) 99.2 (3.8) 98.3 (2.7) 628

99.9 (0.1)

99.4 (1.2) 98.9 (1.9)

99.0 (2.0) 98.5 (2.6)

400–700km

100.0 (4.8) 100.0 (0.8) 100.0 (1.7) 556

100.0 (0.1)

100.0 (0.6) 100.0 (1.1)

100.0 (1.0) 100.0 (1.5)

700km–

Information technology and economic growth

87

contacts are linked with FTF interactions as the model describes, the electronic messages are a complement to visits at least for short distances, and therefore geographically localized. I also suggest that the length of the phone conversation provides useful information on the interdependence between phone messages and FTF contact. If visits are replaced by phone calls, one should talk longer to complete the intended information exchange. For example, I may spend an hour interviewing a famous economist over the phone, while I need only a few minutes to make an appointment if I decide to visit. In other words, lengthy (brief) phone conversations are substitutable with (complementary to) visits.19 Therefore, we can predict that the average conversation length over the phone should monotonically increase with distance since the relative frequency of substitutable phone usage goes up. This prediction is supported by the data. Table 4.7 reports the mean duration of telephone and cellular calls by distance. Mean duration is computed by dividing the total minutes of calls by the total number of calls for each pair of prefectures, and then averaged over all prefecture pairs for each distance segment. On average, an intra-prefecture call by telephone (cellular phone) lasts 153 (70) seconds, while a long-distance call over 700 kilometers lasts 272 (221) seconds, which is 78 per cent (215 per cent) longer. It is important to note that the mean duration gets longer for long distances, regardless of more expensive long-distance call rates. The last column of Table 4.7 shows a negative correlation between the average call length and per capita FTF contact20 for all prefecture pairs. It indicates that phone conversations are longer between communication partners who visit each other less frequently. Empirical examination Next, I perform more formal empirical tests on the interdependence between phone contacts and FTF interactions. Given that the data are available in the form of an OD matrix, the popular gravity model is a convenient formulation to examine the pattern of communication ﬂow across prefectures. The number of contacts between a pair of prefectures (i, j), Tij, is represented by: lnTij ln dij D

Xl l i j ij,

(4.2)

li, j

where dij is the distance between prefectures;21 D is a dummy for intraprefecture contacts; Xl is a vector of prefecture characteristics variables; i and j are prefecture random eﬀects; ij is a mean zero disturbance term. For the observable prefecture characteristics, I use the standard gravity components of population and income per capita, and then add business

88

215.2 (212.6) 113.9 (105.4) 138

152.7 (157.5) 70.1 (77.8) 47

122.7 (113.8) 282

228.1 (200.4) 138.6 (132.3) 558

245.7 (217.4) 161.2 (136.4) 628

262.2 (220.9)

221.0 (151.4) 556

272.2 (258.4)

Intra–100km 100–200km 200–400km 400–700km 700km– prefecture

Distance between prefecture pairs

160.7 (85.0) 2209

250.9 (168.1)

Total

–0.474 [0.000] 2209

–0.510 [0.000]

Correlation with per capita FTF contact*

Notes: The values in the table are unweighted for all prefecture pairs (traﬃc-weighted averages in parentheses). Unweighted averages are preferred to avoid the bias from the large volume of traﬃc between big cities. The signiﬁcance level of each correlation coeﬃcient is shown in square brackets. * Correlation coeﬃcients between log (mean duration of a call) and log (number of trips/population in the originating prefecture). ** Prefecture pairs (i, j) and (j, i) are counted separately.

Telecommunication variables 1. Telephone Mean duration of a call (average of 88–94, seconds) 2. Cellular phone Mean duration of a call (average of 92–93, seconds) Prefecture pairs**

Variables

Table 4.7 Mean duration of a call by distance (%)

89

Information technology and economic growth

activity ratio and population density. All other unobserved prefecture characteristics are captured by i and j. Estimation of this two-way randomeﬀects model follows the feasible GLS estimation procedure developed by Fuller and Battese (1974).22 After conﬁrming that this speciﬁcation ﬁts very well for all of our OD matrix variables,23 I include the FTF contact variable to the gravity regression to test the complementarity between phone and FTF contacts. I also rely on the instrumental variables (IV) approach to deal with the possible endogeneity bias caused by the interdependence. Recalling that the extent of substitution is predicted to increase as the distance between communication partners rises, I allow the FTF contact variable to be interacted with dummies for six distance segments: ln Tij

kSk ln Fij 1 ln dij 2 ln pij D Xl l k

i j ij.

li, j

(4.3)

Fij is the FTF contact variable, and Sk (k1, . . ., 6) is a dummy variable which takes a value of one for the kth distance segment. The six distance segments are: within 100 km (k1); 100 to 200 km (k2); 200 to 400 km (k3); 400 to 700 km (k4); 700 to 1000 km (k5); and over 1000 km (k 6). Since substitution eﬀects increase with distance, we expect a smaller

k for a greater k, and are interested if k becomes negative to suggest that the substitution eﬀects exceed the complementary eﬀects. Other variables are as deﬁned before.24 To allow for the possibility that phone and FTF contacts are simultaneously determined, I use the following three instruments for the FTF contact variable: (i) railway fares; (ii) a dummy for airline connections; and (iii) a dummy that takes a value of one if prefecture pairs are adjacent to each other. The railway fares as a crude proxy for transportation costs exogenously aﬀect the demand for FTF interactions,25 the presence of airline connection contributes to a large increase in long-distance trips, and adjacency helps neighborhood travel by motor vehicles or railways. I assume that these instruments have an inﬂuence in determining the volume of phone contacts only through their eﬀects on transportation traﬃc. In other words, railway fares, airline routes and adjacency will not be correlated with the random components of phone contacts not explained by our model. The direction of bias is unclear. An upward bias from the simultaneity and the attenuation bias from the measurement errors are likely, and the bias from the possible omitted variables can go in either direction. Table 4.8 reports the estimates by GLS and by GLS with instruments. The GLS estimates of k for telephone and cellular phone are signiﬁcantly

90

(4) IV/GLS

(3) GLS

(1) GLS

(2) IV/GLS

Minutes of calls

Number of calls

Telephone

Intraprefecture dummy Population Origin

Call rates

Distance

0.274** (0.013) 0.266** (0.011) 0.209** (0.010) 0.182** (0.009) 0.152** (0.010) 0.078** (0.014)

(5) GLS 0.375** (0.049) 0.360** (0.044) 0.366** (0.041) 0.325** (0.034) 0.399** (0.026) 0.234** (0.032)

(6) IV/GLS

Number of calls

0.256** (0.012) 0.242** (0.010) 0.184** (0.009) 0.159** (0.008) 0.131** (0.009) 0.070** (0.012)

(7) GLS 0.320** (0.042) 0.302** (0.038) 0.272** (0.035) 0.262** (0.029) 0.272** (0.022) 0.197** (0.027)

(8) IV/GLS

Minutes of calls

Cellular phone

0.187** (0.011) 0.191** (0.010) 0.156** (0.009) 0.144** (0.088) 0.122** (0.010) 0.084** (0.013)

(9) GLS 0.310** (0.045) 0.307** (0.040) 0.290** (0.040) 0.295** (0.032) 0.293** (0.025) 0.241** (0.031)

(10) IV/GLS

Mail

Development variableslog (aggregate ﬂow from prefecture i to j)

0.331** (0.045) 0.394** (0.039) 0.344** (0.036) 0.280 (0.033) 0.313** (0.038) 0.134** (0.052)

(11) GLS

0.662** (0.163) 0.704** (0.146) 0.692** (0.145) 0.636 (0.118) 0.660** (0.090) 0.452** (0.625)

(12) IV/GLS

Commodities

Reference variables

0.207** (0.031) 0.258** (0.027) 0.281** (0.025) 0.325** (0.023) 0.347** (0.026) 0.169** (0.036)

(13) GLS

0.303** (0.110) 0.357** (0.098) 0.397** (0.098) 0.447** (0.080) 0.499* (0.061) 0.328** (0.075)

(14) IV/GLS

Freight

1.097** (0.171)

0.843** (0.206)

1.150** (0.170)

0.940** (0.199)

0.949** (0.176)

0.764** (0.261)

0.935** (0.151)

0.803** (0.169)

0.920** 0.732** 1.690** (0.176) (0.202) (0.662)

1.235* (0.708)

1.094** (0.496)

0.936* (0.515)

0.741** 0.133 0.742** 0.268** 0.750** 0.404** 0.655** 0.425** 0.352** 0.014** 1.282** 0.266** 1.214** 0.914** (0.043) (0.158) (0.043) (0.152) (0.041) (0.147) (0.036) (0.126) (0.038) (0.143) (0.151) (0.524) (0.104) (0.353) 0.137 0.594** 0.130 0.574** 0.155 0.808** 0.116 0.705** – – – – – – (0.120) (0.158) (0.119) (0.152) (0.260) (0.329) (0.229) (0.283) – – – – – – 2.600** 1.220 2.256** 1.045** 1.132** 1.258** 0.954** 1.055** 2.277** 2.350** 0.475** 0.714** 0.711** 0.837** (0.213) (0.269) (0.212) (0.259) (0.105) (0.119) (0.093) (0.103) (0.095) (0.111) (0.374) (0.405) (0.258) (0.273)

Interaction terms (FTP contact) X (distance dummy) 1. –100km 0.251** 0.339** 0.211** 0.272** (0.014) (0.048) (0.014) (0.046) 2. 100– 0.249** 0.383** 0.218** 0.321** 200km (0.010) (0.043) (0.010) (0.041) 3. 200– 0.197** 0.375** 0.179 0.317** 400km (0.009) (0.043) (0.009) (0.041) 4. 400– 0.179** 0.370** 0.167** 0.324** 700km (0.009) (0.035) (0.009) (0.034) 5. 700– 0.157** 0.369** 0.150** 0.343** 1000km (0.010) (0.027) (0.010) (0.026) 6. 1000km– 0.105** 0.306** 0.099** 0.277** (0.013) (0.033) (0.013) (0.032)

Independent variables

Table 4.8 Complementarity between phone and FTP contacts by distance segment

91

2209

2209

2209

2209

2209

2209

2209

2209

2209

0.315 (0.884) 1.211 (1.069) 0.8438 2209

2209

0.001 1.257 (0.987) (3.323) 0.880 2.670 (1.194) (3.175) 0.8054 0.5814

1.083** 0.896** 1.917** (0.213) (0.242) (0.633)

2209

1.885 (3.449) 3.346 (3.297) 0.5492

1.466** (0.678)

2209

0.786 (2.487) 1.192 (2.204) 0.653

1.298** (0.439)

2209

1.074 (2.521) 1.494 (2.234) 0.6437

1.142** (0.459)

Notes: Estimates are based on the two-way random eﬀects GLS estimation. All variables except percentages and dummies are expressed in natural logarithms. Intercepts are not reported. Standard errors are in parentheses. For the regressions of IV/GLS, the FTF contact variable is treated as endogenous. Excluded instruments are railway fares, a dummy for airline connections, and a dummy for prefecture pairs which are adjacent to each other. * Signiﬁcant at the 10% level. ** Signiﬁcant at the 5% level.

No. of obs.

1.070** 0.817** 1.168** 0.960** 1.027** 0.843** 1.030** 0.900** (0.172) (0.207) (0.173) (0.202) (0.187) (0.213) (0.169) (0.188) Income per capita Origin 0.428 0.717 0.549 0.807 0.011 0.291 0.680 0.445 (0.857) (1.004) (0.853) (0.967) (0.881) (0.979) (0.759) (0.823) Destination 0.442 0.755 0.971 1.247 0.467 0.784 0.679 0.925 (0.861) (1.009) (0.866) (0.983) (0.938) (1.042) (0.848) (0.919) Adjusted R2 0.9053 0.870 0.8937 0.8634 0.9101 0.889 0.9139 0.8989

Destination

92

Agglomeration in Asia

positive and decline as k rises, implying the weaker complementarity for long distances.26 The coeﬃcient estimates by IV/GLS, however, do not decrease very much. All estimates of k take signiﬁcantly positive values, suggesting that phone messages and visits are net complements even for the longest-distance segment. Throughout all the empirical experiments in this study,27 I ﬁnd no evidence to suggest that they are net substitutes. Implications I found evidence to suggest that the relative frequency of substitutable phone usage increases monotonically with the distance separating communication partners. However, complementary eﬀects always equal or exceed substitution eﬀects, so that phone usage and transportation are net complements. These ﬁndings suggest that telecommunications may be a complement to cities, which are physical centers that facilitate FTF interactions. Cities stimulate both physical and electronic contacts and thereby foster the ﬂow of ideas. The advance of information technology may strengthen this urban function: technology induces more electronic communication, and therefore more FTF interactions at the same time, making cities more attractive rather than obsolete.

5. DISCUSSION In this chapter, I overviewed the current situation of urban concentration in Japan, and explained two paradoxes accompanied by the IT revolution. The implications are as follows: Even though IT should have a potential impact on productivity improvement and decentralization, it is like a fantasy to presume that spending on IT would necessarily realize them. Strategy, wisdom and continuous eﬀorts are indispensable to obtain desired outcomes. In the context of IT and cities, it is not the case that decentralization would progress just because a substantial amount of budget is used for IT investment in less-populated areas. With the exception of the Tokyo metropolitan area where investment is carried out almost autonomously, every region will have some diﬃculty stopping an outﬂow of population or economic activities, unless there is a well-planned strategy. There are already a number of superbly equipped IT facilities in local regions that are hardly used due to insuﬃcient contents or a mismatch of demand. As explained in Sections 3 and 4, which evaluated two factors of the accumulation paradox, the importance of FTF interactions facilitated by

Information technology and economic growth

93

urban environment may become more important under the shift toward the knowledge economy. Also, the IT revolution is likely to function as a complement to this shift. Even in local regions, it is necessary to re-evaluate the function of agglomeration rather than to aim at uniform development over the areas. Carrying out attractive IT projects intensively or unevenly in the urban center based on a creative plan may attract a highly skilled workforce. Otherwise, it may fail to survive competition among cities under the movement toward merger and acquisitions promoted by the central government. More concretely, the investment approach should progress from construction to knowledge oriented, and it is desirable to establish the urban center of excellence or brain magnet where intellectual high-skilled workers gather to generate knowledge spillovers. For that purpose, an exceptional environment of education, research, life and culture and extraordinary treatment to attract talent are needed. In the trend toward the knowledgebased economy, creation of new technologies and business models both in hardware and software are always called for. If continuous and endless investment in knowledge is neglected, competitiveness will be lost due to the sharp devaluation of the knowledge stock. As a typical prescription, for example, city planning which places educational facilities such as universities in the urban center would serve as a good experiment. Knowledge spreads through miscellaneous exchanges around the center, and this may work as a base for industry–academic cooperation or lifelong education. The population census 2000 suggests ‘mini’ concentration of population or high-tech ﬁrms in cities like Sapporo, Sendai and Fukuoka where a prestigious national university is located along with industries and government in the urban center. This implies that Japanese universities may also play a signiﬁcant role in creating local attractiveness. The IT revolution brought severe competition among ﬁrms and companies seeking knowledge, and cities may not be an exception, either.

NOTES 1. 2.

3. 4. 5.

See Itoh (2001). The diﬀusion rates of the Internet access and cellular phone subscriptions are, 37.1 and 50.3 per cent, respectively (at the end of 2000, Ministry of Internal Aﬀairs and Communications (2001a)). Annual domestic shipments of personal computers and color TVs (including LCD TVs) total 11.55 million and 10.30 million, respectively (Japan Electronics and Information Technology Industries Association 2000). See Economic Planning Agency (2000). There are, of course, other factors for centralization such as land prices falling after the bubble in the metropolitan area. See Jaﬀe et al. (1993), for instance.

94 6. 7.

8. 9. 10.

11. 12. 13.

14.

15. 16. 17. 18. 19. 20. 21. 22. 23. 24.

25.

Agglomeration in Asia See descriptions of ‘books’ versus ‘lectures’ by Gasper and Glaeser (1998). In this estimation, I did not use the data after 1995, as e-mail has become widely available in Japan since 1995. Telephone traﬃc includes so-called ‘dial-up’, by which we can be connected to the Internet via telephone lines. Therefore, it became more diﬃcult in the late 1990s to capture the locations of communication partners. The similar level of the concentration index does not necessarily imply similar distribution in general, but it is so in my data since the volume of communication usage is strongly linked with population. See Imagawa (2002). For example, the length of business calls is much shorter on average. In 1990, the mean duration of a call was 113 seconds for business use, and 261 seconds for residential use. Another example is that a business unit will be equipped with a large number of telephones, but with a small number of TV sets. For the utilization measures of the two-way tools in Table 4.4, regressions for the outgoing and incoming contacts produce similar coeﬃcients, and therefore the estimation results for the incoming contacts are omitted from the table. I also used commodities (annual sales in wholesale and retail trade) and freight (annual freight tonnage) as dependent variables. I found only insigniﬁcantly positive linkage between density and the per capita usage of these ‘non-communication goods’. In an earlier version of this study, I estimated the same regression using a crude population density measure (population divided by total prefecture areas including forests, grassland and lakes), with no substantive change in results. Compared with this earlier result, all the density coeﬃcients in Table 4.4 for two-way tools increased their magnitude substantially, with only slight changes for one-way tools. This is convincing since the positive connection is strengthened with the ‘accurate’ measure of density. Also, there was no substantive change when urbanization measures (percentage of urban population and so on) were used instead of density. The density coeﬃcients for the utilization measures of cellular phones may not be robust enough. The data for cellular phone utilization cover only two or three years after 1992. Taking into account the fact that cellular phones have become remarkably popular since 1994, I do not take this result as a serious problem. It is possible that there was a supplyside restraint for the provision of cellular services in non-metropolitan high-density areas during our sample period. See Imagawa (2002) for a description of the model. Prefecture pairs (i, j) and (j, i) are counted separately. Prefecture pair (i, i) is treated as a pair of prefectures with identical features. The distance between prefectures is deﬁned as physical distance between prefecture capitals. For mail and commodities, there is a strong centralization in their transactions. For example, Tokyo’s intra-prefecture transaction alone accounts for 15 per cent of total mail and 26 per cent of total commodities. The theoretical model also suggests that the mean duration of a call monotonically increases with distance. See Imagawa (2002). Per capita FTF contact is computed as the number of trips divided by the population in the originating prefecture. In order to compare the coeﬃcients for phone and FTF contacts with those for reference variables for which appropriate prices are not explicit, I do not include price in equation (4.2) and interpret that the distance also internalizes price eﬀects. See Imagawa (2002) for details. Estimation results are not reported here. Most of the adjusted R2’s are close to 0.9. As prefecture characteristics variables, I used only population and per capita income as the coeﬃcients on the other two characteristics were not signiﬁcant in the gravity model estimation and the reference variables were not estimated accurately in the presence of those variables. Railway fares are regulated and determined based on distance rather than on railway demand (at least in the framework of the static analysis of this study).

Information technology and economic growth 26. 27.

95

Note that the same pattern is found for mail, but not for commodities and freight. I showed that the mean duration of a call increases as the frequency of visits declines. I also tested whether this negative correlation survives after controlling for factors (gravity components and so on) which should aﬀect the length of the phone conversation. Then, the negative correlation turned out to be robust, but the magnitude of its eﬀect is relatively small. This can be consistent when substitution eﬀects do not increase enough to exceed complementary eﬀects even if frequent visits reduce the average conversation length.

REFERENCES Brynjolfsson, E. and L. Hitt (1998), ‘Beyond the productivity paradox’, Communications of the ACM, 41 (8), 49–55. Bureau of Statistics (1892), Nihon Teikoku Tokei Tekiyo (Résumé of Statistics of the Japanese Empire), Tokyo: Naikaku Tokeikyoku (Imperial Cabinet). Ciccone, A. and R.E. Hall (1996), ‘Productivity and the density of economic activity’, American Economic Review, 86 (March), 54–70. Economic Planning Agency (2000), The Present Condition of the Japan Economy, Cabinet Oﬃce, Tokyo. Fujita, M., P. Krugman and A.J. Venables (1999), The Spatial Economy, Cambridge, MA: MIT Press. Fuller, W. and G. Battese (1974), ‘Estimation of linear models with crossed-error structure’, Journal of Econometrics, 2, 67–78. Gasper, J. and E.L. Glaeser (1998), ‘Information technology and the future of cities’, Journal of Urban Economics, 43, 136–56. Glaeser, E.L. (1994), ‘Cities, information, and economic growth’, Cityscape, 1, 9–47. Glaeser, E.L. and D.C. Mare (1994), ‘Cities and skills’, Hoover Institute Working Paper, E-94-11, Stanford University. Greene, W.H. (1997), Econometric Analysis, 3rd edn, Englewood Cliﬀs, NJ: Prentice-Hall. Imagawa, T. (2001), ‘Are telecommunications and transportation substitutes? Implications for information technology and cities’, IPTP Monthly, 153, 55–69. Imagawa, T. (2002), Economic Analysis of Telecommunications, Technology, and Cities in Japan, Tokyo: Taga Shuppan. Itoh, M. (2001), ‘The Fantasy of the IT revolution’, Mainichi Shimbun, Evening edn, March 22, p. 3. Jaﬀe, A., M. Trajtenberg and R. Henderson (1993), ‘Geographic localization of knowledge spillovers as evidenced by patent citations’, Quarterly Journal of Economics, 108, 577–98. Japan Electronics and Information Technology Industries Association (2000), Domestic Shipments of Major Consumer Electronic Equipment. Krugman, P. (1991), ‘Increasing returns and economic geography’, Journal of Political Economy, 99, 483–99. Ministry of Internal Aﬀairs and Communications (2001a), 2001 White Paper: Information and Communications in Japan, Tokyo. Ministry of Internal Aﬀairs and Communications (2001b), Population Census 2000, Tokyo. Ministry of Land, Infrastructure and Transport (2001), Survey of the Softwareoriented IT Industries, Tokyo.

96

Agglomeration in Asia

Romer, P. (1986), ‘Increasing returns and long run growth’, Journal of Political Economy, 94, 1002–37. Solow, R. (1987), ‘We’d better watch out’, New York Times Book Review, July 12, p. 36. Tirole, J. (1988), The Theory of Industrial Organization, Cambridge, MA: MIT Press. US Bureau of Economic Analysis, National Income and Product Accounts Tables, http://www.bea.gov/bea/dn/nipaweb/index.asp US Bureau of Labor Statistics, Overview of BLS Productivity Statistics, http://stats. bls.gov/bls/productivity.htm

5. Agglomeration of exporting ﬁrms in industrial zones in northern Vietnam: players and institutions Akifumi Kuchiki 1.

INTRODUCTION

Foreign direct investment (FDI) has contributed to enhancing economic growth in many Asian countries by agglomerating ﬁrms in many cities (see Urata 2001). It is eﬀective in applying Asian experiences to countries in other regions to ﬁnd out how multinational corporations (MNCs) are agglomerated in cities. Development strategy for industrial agglomeration can be useful once we make clear what kind of players participate and which institutions are established for the strategy. The Poverty Reduction Strategy Papers (World Bank 2002) repeat the importance of participation in development processes. We must identify players and divide their roles in the processes to maximize the eﬀectiveness of strategy for industrial agglomeration. Krugman (1993) and others set out industrial location patterns. Fujita and Hamaguchi (2001) focused on the role that intermediate goods play in agglomerating ﬁrms. Mori and Nishikimi (2002) emphasized increasing returns in transportation, in particular, economies of transport density. These are theoretical approaches to industrial agglomeration. So, paying attention to the roles of intermediate goods and transportation, this chapter attempts to show what kinds of players and institutions were involved in agglomerating ﬁrms in cities in East Asia. Kennedy (1999), Weijland (1999) and Kawakami et al. (2003) analysed industrial clusters in India, Indonesia and Taiwan, respectively. These are case studies on industrial agglomeration. No paper has tried to illustrate the case of the Vietnamese experience by focusing on the role of industrial zones (IZs) together with infrastructure and institutions of capacity for development to agglomerate ﬁrms, and enhancing the macroeconomic growth. The purpose of this chapter is to prove that IZs as quasi-public goods enhance macroeconomic growth by using a theoretical macroeconomic 97

98

Agglomeration in Asia

model and to illustrate that transportation infrastructure such as highways and ports as well as institutions for IZs are eﬀective in agglomerating a leading ﬁrm and its related ﬁrms in northern Vietnam. The related ﬁrms provide the leading ﬁrm with parts and components for intermediate goods. The players necessary for success in industrial agglomeration were private companies establishing IZs, oﬃcial development assistance (ODA) to facilitate transportation infrastructure, and central and local governments to establish institutions. The aim here is to show the positive eﬀects of IZs on enhancing aggregate growth by employing a macroeconomic growth model with quasi-public goods from IZs. Moreover, we shall show how in northern Vietnam after 2001 growth was attained by industrial agglomeration. The theoretical model derives from the condition that establishment of IZs enhances aggregate growth. The condition is that average capital productivity is higher than the critical value of the model. The players for industrial agglomeration are as follows. The Vietnamese government gave preferential taxes to tenant ﬁrms of IZs. A Japanese trading corporation (Sogoshosha) and a security company in the private sector established IZs in northern Vietnam to seek proﬁts by managing the IZs and related services rather than selling land for IZs to tenants. ODA facilitated the economic infrastructure such as Haiphong port and highway Route 5. A Japanese high-tech company became an IZ tenant together with its related companies to anchor parts and components ﬁrms. All these players contributed to the success of industrial agglomeration in northern Vietnam. A combination of IZs, preferential taxes, a port and highway, and institutions could agglomerate ﬁrms in northern Vietnam by raising average capital productivity. This success story will help to formulate and apply the strategy for industrial agglomeration to other countries. The chapter is organized as follows. In Section 2 we build a theoretical growth model to explain that an IZ functions as quasi-public goods to attract FDI and enhance aggregate growth. Sections 3–5 make clear that economic agents or players are crucial to establishing IZs and spreading economic growth throughout a country. Section 3 explains that organizations of central and local governments in the quasi-public sector constructed IZs in Malaysia and Thailand in the 1980s and 1990s. Section 4 points out that Japanese trading corporations established IZs in East Asia including Thailand, Indonesia and the Philippines, and that the IZs helped Japanese manufacturing companies invest in the IZs. Section 5 adds the fact that not only Japanese trading corporations but also other Asian companies and governments established IZs in East Asia. Section 6 shows positive macroeconomic eﬀects of national highway Route 5 on the northern Vietnamese economy. Section 7 summarizes and concludes the chapter.

Agglomeration of exporting firms in northern Vietnam

2.

99

A THEORETICAL MODEL

Industrial zones have three distinctive characteristics. First, they facilitate infrastructure and institutions required for investors. The physical infrastructure includes electricity, roads, telecommunications, water supply and sewerage, and so on, while institutions include one-stop services and others. Second, preferential taxation is an important factor when foreign companies are making decisions on investment since the main motivation for investing overseas is to reduce their production costs. To be more precise, the exemption of equipment from import tariﬀs and reduced or exemption from corporation tax are crucial incentives for foreign investors. Third, the permitting of 100 per cent foreign subsidiaries is also eﬀective to induce foreign companies to move into IZs. Step 1: Kuchiki and Yamada (1997) showed that the public sectors of Thailand, Malaysia and other Asian countries undertook a policy to enhance economic growth nationwide by constructing IZs: Z Z(G), where Z is the construction of IZs and G is the government expenditure on IZ infrastructure. Here we can deﬁne IZs as quasi-public goods since they are excludable, rival, but indivisible. An IZ is a package of land, infrastructure and institutions and the package is indivisible, for example, a stadium. That is why an IZ can cluster ﬁrms within an area of 300 hectares as is usual in East Asia. But its tenants can use each lot of the IZ exclusively. The tenant is a rival with other investors since it occupies its own lot and others cannot use it. We assume for the sake of simplicity that government expenditure (G) is ﬁnanced by taxes (T), and proportional to GDP (y): T G y, where y is real GDP and is the tax rate. Here we also assume without loss of generality that the government expenditure is equal to tax reductions, costs to construct IZs, and other costs related to IZs. In short, Z y. Note that there are many cases when private companies, rather than the public sector, pay costs to construct IZs since IZs have the property of a quasi-public good.

100

Agglomeration in Asia

Step 2: The investment function of FDI can be expressed as follows: If If (Z; e, t), where e is the exchange rate of currency and t is the rate of tax on FDI, and these variables are assumed to be given and constant. So the exchange rate and the tax rate are part of the investment climate for foreign investors. Here we assume that governments implement all the incentive policies mentioned above to invite FDI. We build a model of surplus labor, that is, the existence of cheap labor. We specify the investment function of FDI as follows: Ij y,

(5.1)

where is constant. The value of expresses the situation of investment climate of a recipient country. That is, the investment function is proportional to Z and G. Step 3: First we build the growth model, and then derive the condition that a country can raise the rate of economic growth by constructing IZs and inviting FDI. We consider the problem of utility maximization of a representative consumer subject to the condition of market equilibrium. The usual utility function (U) is: U

0

C1 t e t 1 dt,

(5.2)

where t represents time from now on, is time preference, is a coeﬃcient of relative risk aversion and Ct is consumption. We follow the model of Barro and Sala-i-Martin (1995, ch. 4). The condition of the market equilibrium is described as: yt Ct It Xt Mt, where It is domestic investment, Xt is export and Mt is import. Excess import (Xt – Mt) is ﬁnanced by foreign investment: Mt Xt Ift. We assume a production function in the following: yt AKt,

Agglomeration of exporting firms in northern Vietnam

101

where Kt is capital stock and A is constant. Here the production function is a Leontief type and labor is redundant in the economy. So our discussion is focused on capital stock. Using equation (5.1), we can obtain: It Ki AKt Ct Ift A(1 )Kt Ct.

(5.3)

We shall compare the following two cases: (i) the government will not tax and not establish IZs and foreign investors do not invest in IZs; and (ii) the government establishes IZs by taxing the people. The ﬁrst case is that Ift is zero. The problem is to maximize equation (5.2) under a constraint of equation (5.3) by putting Ift at zero. The economic growth rate without tax and IZs (0) is: 0

A .

The economic growth rate () of the second case of establishing IZs and inviting FDI is

A(1 ) .

The beneﬁt of inviting FDI is the diﬀerence ( 0) multiplied by GDP, that is, A ( 0 ) . The net beneﬁt is the beneﬁt minus the cost that is taxed to the people, A A Net benefit . The result shows that the higher the tax rate and the GDP, the higher the net beneﬁt. Note that this model cannot take account of taxpayers’ complaints about higher taxes. We can obtain the condition that makes the net beneﬁt positive: A . If this condition is satisﬁed, and FDI is invited by construction of IZs ﬁnanced by taxes, then the national income increases. The government’s role is to raise A of capital productivity of the production function. Infrastructure for IZs is, of course, important. One of the most critical policy

102

Agglomeration in Asia

measures is to permit 100 per cent ownership by foreign capital since it is said that the most diﬃcult hurdle is ﬁnding a good partner for a joint venture. This system reduces risk and raises the stability of A.

3. IZ DEVELOPMENT BY THE QUASI-PUBLIC SECTOR This section shows that organizations in the public and quasi-public sectors played an important role in attracting FDI and particularly in establishing IZs. Thailand’s Board of Investment is responsible for attracting FDI, while the Industrial Estate Authority of Thailand handles construction of IZs for FDI. The Malaysia Industrial Development Authority attracts FDI, and the state economic development corporations handle construction of IZs. In this section, we shall focus our analysis particularly on cases from Malaysia and Thailand. State Economic Development Corporations of Malaysia In the case of Malaysia, for example, more than 160 IZs had been developed by 1992. These were mainly developed by state governments and proved to be particularly eﬀective in attracting foreign investment in the electronics/electrical industry. The high economic growth since 1988 initiated by FTZs (free trade zones which acted as export-processing zones (EPZs) in Malaysia) acting as pump-primers, spread to the whole western part of Peninsula Malaysia by the public sector establishing IZs. One characteristic of foreign investors moving into IZs in Malaysia in the 1990s was that the government allowed the investors to invest in not only the electronics/electrical industry but also many diﬀerent types of manufacturing industries. There were 10 FTZs and 120 IZs in 1987. The function of the FTZ is the same as the EPZ. The main states that employed large numbers of workers in IZs were Johor, Penang and Selangor. These IZs were developed by the state economic development corporations, most of which were established in the 1960s. In the cases of Selangor, Penang and Johor, they were established in 1964, 1965 and 1966, respectively. For example, the Penang State Economic Development Corporation constructed Bayan Lepas IZ and Prai IZ. In addition, it managed to enhance the level of technology by establishing the Institute of Precision Molding that belongs to the Penang Skills Development Center. Kedah State Economic Development Corporation developed Kulim IZ and established the Training Center for Human Resource Development. The above

103

Agglomeration of exporting firms in northern Vietnam

Table 5.1

Number of IZs and FTZs in Malaysia

States

Factories (end of 1989)

%

Employees (end of 1989)

%

IZs* FTZs* (1987) (1987)

Johor Kedah Kalantan Malaka Negari Semibilin Pahang Penang Perak Perlis Selangor Terengganu Kuala Lumpur Labuan Sabah Sarawak

748 228 107 147 135 164 551 552 14 1022 83 530 27 788 686

12.9 3.9 1.8 2.5 2.3 2.8 9.5 9.5 0.2 17.6 1.4 9.1 0.4 13.6 11.8

103 949 37 071 11 684 23 198 17 273 22 783 96 423 49 908 3581 137 699 7688 34 626 1710 27 090 23 895

17.3 6.1 1.9 3.8 2.8 3.8 16.1 8.3 1.0 23.0 1.2 5.7 0.5 4.5 3.9

14 6 6 9 10 8 6 14 2 14 13 2 1 7 8

– – – 2 – – 5 – – 3 – – – – –

Total

5782

100.0**

598 578

100.0**

120

10

Notes: * MIDA, Malaysia, Statistics on the Manufacturing Sector 1988. ** Totals rounded. Source: Torii (1994).

illustrate that the role of state economic development corporations was crucial to the development of IZs for industrialization throughout the country. The number of IZs increased from 120 in 1987 to 166 in 1992 (Tables 5.1 and 5.2). States with large areas sold as IZs are Selangor, Johor and Penang, followed by Terengganu and Perak (Table 5.2). The government planned to develop 94 IZs in 1992. Until that year, Japanese multinational corporations, mainly in the electric and electronics industry, had invested in the IZs. Other Japanese companies had invested in various manufacturing industries until 1996. The preferential treatment oﬀered by the LMW (licensed manufacturing warehouse) was the same as that by the FTZ. Malaysia succeeded in attracting FDI, creating employment and spreading the growth of FTZs over the development of IZs. In the next stage, both the public and the private sectors took part in construction of the IZs. But the purpose of the public sector, which was to diﬀuse growth throughout the country to combat income inequality, was diﬀerent from that of the private sector, which was the pursuit of proﬁt maximization. We can judge the diﬀerence between the two sectors from

104

Table 5.2

Agglomeration in Asia

IZs in Malaysia, January 1, 1992 (ha)

States

Developed areas

Areas for sale

2922.05 630.16 458.92 2090.78 208.23 1487.36 1658.95 839.29 76.94 934.48 458.99 2084.96 230.10 1168.57

2254.37 518.44 334.97 2036.24 198.81 1206.04 1316.89 714.79 63.09 1466.08 360.49 1453.92 143.98 719.86

15 249.78

12 787.97

Johor Malaka Negari Semibilin Selangor Labuan Perak Penang Kedah Perlis Pahang Kelantan Terengganu Sabah Sarawak Total

Sold areas

No. of IZs

No. of planned IZs

1438.18 501.44 311.84 1953.33 198.81 923.18 1301.23 688.62 38.79 564.84 295.46 1029.80 121.57 814.46

22 8 7 22 2 26 16 10 4 7 14 10 8 10

7 6 3 21 2 15 8 4 3 4 6 3 6 6

10 181.55

166

94

Source: Harian (1992).

the location of IZs as follows. One policy measure of Malaysia’s Five-Year Plan, which started in 1996, was to develop IZs. The development plan focused on Sabah and Sarawak, states that are located on the east coast far from Kuala Lumpur. IZs cover a total area of 3926 hectares, or 41.7 per cent of the planned area of the country. The economic growth in the states was slow from the latter half of the 1980s to the mid-1990s. So the purpose of the Five-Year Plan was to narrow the gap between the rich and poor states. Industrial Estate Authority of Thailand In Thailand, IZs were ﬁrst established around Bangkok. As of 1996, site sales took place at 16 major IZs, including Udon Thani IZ located at some distance from Bangkok. Local capital was active in developing IZs in Thailand. The Siam Cement Industrial Land and Thai Industrial Estate Corporation developed more than 15 IZs in 1996. The 1985 Plaza Accords between summit countries agreed to appreciate exchange rates of Asian currencies, including Japan and Korea. The FDI from Asian countries to Thailand increased in the latter half of 1986, and mainly invested in EPZs and IZs (general industrial zones) which were

Agglomeration of exporting firms in northern Vietnam

105

established by the public sector, and located in Bangchan, Lartkrabang, Bang Phlee, Bang Poo and Map Ta Phut near Bangkok. However, economic agents who were responsible for construction of the IZs changed from the public to the private sector in 1988. The private sector established IZs in Bang Khadi, Bang Pakon, Suranaree and M. Thai near Bangkok (Table 5.3). The investment promotion policies to attract FDI into Thailand were directed by the Board of Investment (BOI), which was formed in 1960. The purpose of the BOI was to implement the Industrial Investment Encouragement Law regulated in 1960. The law was amended in 1972, specifying that the role of the BOI was to give preferential treatment to FDI. The degree of preferential treatment depended on what category of industry a ﬁrm was classiﬁed as or where the ﬁrm located. First, the BOI intended to diﬀuse the economic growth to local areas where the per capita income was much lower than that of Bangkok. One of the characteristics of the establishment of IZs particularly in Thailand was joint projects carried out by both the public and the private sectors. The private sector applied to the Industrial Estate Authority of Thailand to develop an IZ, and the quasi-public sector constructed 90 per cent of the planned IZs. The Authority established an oﬃce after it completed investigations into whether an application satisﬁed the conditions of the law. The oﬃce acted as a one-stop service for ﬁrms who invested in the IZ. Given the fact that in 1996 IZ construction was scheduled to be in local areas, it is clear that the government’s intention was to diﬀuse the growth in/near Bangkok to local regions to narrow the income gap between Bangkok and local regions (Table 5.4).

4.

IZ DEVELOPMENT BY THE PRIVATE SECTOR

This section shows that Japanese trading corporations played an important role in attracting Japanese manufacturing ﬁrms to ASEAN (Association of South East Asian Nations) countries to invest by establishing IZs. EPZs were developed in Thailand, Malaysia and Indonesia in the 1980s under the leadership of the central government. Around 1990, however, the construction of IZs became more important than that of EPZs since the role of IZs was to manufacture products for both the export and the domestic markets. For example, the construction of IZs by the private sector in Indonesia was permitted, partly because of the inadequate funds available in the public sector. So the quasi-public sector was replaced by the private sector, which then provided quasi-public goods such as EPZs and IZs. Six leading Japanese trading corporations – Itochu Corporation, Marubeni Corporation, Mitsubishi Corporation, Mitsui & Co. Ltd, Nissho

106

7. Map Ta Phut For heavy & chemical industry For coastal area

3. Bang Poo I II III 4. Bang Phlee I II III (planning) 5. Northern Region 6. Laem Chabang

A. IEAT IZs 1. Bangchan 2. Lartkrabang I, II III

Names of IZs

Rayong

Rayong

IEAT

IEAT

Chonburi

IEAT/NHA IEAT/NHA IEAT/NHA IEAT

IEAT

Samut Prakam Samut Prakam Samut Prakam Lamphun

TIDC/IEAT TIDC/IEAT TIDC/IEAT

Samut Prakam Samut Prakam Samut Prakam

Bangkok Bangkok

IEAT IRD/IEAT

Bangkok

IEAT

Abbreviation Public Private Area

Economic agent

Table 5.3 IZs in Thailand (end of 1988)

200 km

200 km

40 km 40 km 40 km (25 km from Chiangmai) 130 km

34 km 34 km 34 km

35 km 35 km

30 km

Distance from Bangkok

Location

1989(I)

1989(I)

1990

1984 1989 1990 1985

1988 1989

1979 1989

1979(1972)

Completion

Included in the above

6000

3556

700 1762

456

900

3733

1290 1103

677

Area (rai)

60% sold out

All reserved

Sold out (1986) All reserved Soon for sale CIE: 10% sold, EPZ: 44% sold CIE: EPZ to accept reservation

75% sold out 95% sold out

Sold out (1987) CIE, EPZ are almost sold out

Sold out (1986)

Situation of sales

107

Nakhon Ratchasima Nakhon Ratchasima Samut Prakarn

Chacheongsao Chacheongsao

Patham Thani

Chonburi Chonburi

Patham Thani Patham Thani Patham Thani Patham Thani

Rayong

(7 km from Korat) (7 km from Korat) 40 km

80 km 80 km

40 km

110 km 110 km

45 km 45 km 45 km 45 km

200 km

Started in 1989 1989

1989

1989 Started in 1989

1977 Started in1989 1989

1984 1987 1989 1989

1989(I)

824

528

530

300 1400

1023

1202 More than 500

1600 1009 2029 Included in the above

Included in the above

Industrial Estate Authority of Thailand, Board of Investment and Japan External Trade Organization (Japan).

Rai 1600m2; CIE general industrial estate.

Sources:

Note:

6. Theparak M.Thai IEC

SIZC

SIZC

5. Suranaree I

II

BPIC BPIC

BPIC

SPIC SPIC

2. Sriracha I II (planning)

3. Bang Khadi 4. Bang Pakon I II (planning)

NNC NNC NNC NNC

IEAT

For supporting industry B. Private sector IZs 1. Nawa Nakhon I II III IV

Started to make reservation 60% reserved

57% reserved

All reserved Started to make reservation

Almost sold out

91% sold out 40% reserved

Sold out (1987) Sold out (1987) Almost reserved Almost reserved

10% sold out

108

Bangchan Bang Poo IE No.1.2 Bang Phlee IE 1.2.3 Nawa Nakhon IE Pojana Ind.Park 1 Sriracha IE Northern Region IE 1 Minburi No.1 Suranaree Ind Zone 1 Mah Bookhrong IE Bang Khadi IE M.Thai IE Pojana Ind.Park 2 Saha Rattananajhon Leam Chabang IE Northern Region IE 2 Minburi No.2 Larkrabang IE Bangkok Airport Well Crow IE Chonburi Bo-win 1

Name

1972 1977 1984 1987 1988 1988 1988 1988 1988 1988 1988 1988 1990 1990 1990 1990 1990 1990 1990 1991 1991

Completion

Table 5.4 IZs in Thailand (rai)

IEAT IEAT/Private IEAT Private Private Private IZ Private Private Private Private Private Private Private IEAT IEAT Private IEAT/Private Private IEAT/Private IEAT/Private

Economic agents Minburi, Bangkok Samut Prakam Samut Prakam Pathum Thani Ayuthaya Chonburi Lamphun Minburi, Bangkok Nakhon Rachasima Pathum Thani Pathum Thani Samut Prakam Ayuthaya Ayuthaya Chonburi Lamphun Minburi, Bangkok Bangkok Nonthaburi Chacheongsao Chonburi

Location 677 5930 1011 3900 820 1202 – – 1000 1410 1136 650 605 1700 3556 1760 – 2515 – 3000 1500

Total area 509 3811 796 2027 533 1202 356 300 530 1410 1136 650 393 650 1715 – 270 1184 680 1510 760

Area of IZs

– 272 – – – – – – – – – – – 267 804 812 – 683 – 530 516

Area of EPZs

109

Bang Pakon IE Lamphun IE Eastern (Map Ta Phut) Jong Stit Ind Park Nong Khae IE Hitech IE No.1 Samut Sakhon IE Saraburi(Kaebg Khoi) Gateway City Suranaree Ind Zone 2 Map Ta Phut IE Siam Cement Land Bang Paln IE Saha Group IP Lamphun Northern Region IE Khanthaburi IE Chiang Rai IE Nakhon Sawan IE Pattani IE Lower North IE No.1 Southern Region IE Upper Northeast Upper Northeast 1 Pha Daeng IE Southern IE Narathiwat IE

1991 1991 1991 1991 1991 1992 1992 1992 1993 1993 1993 1993 1994 1994 1994 1996 1996 1996 1996 1996 1996 1996 1997 1997 1998 2000

IEAT/Private Private IEAT Private IEAT/Private IEAT/Private Private IEAT/Private IEAT/Private Private IEAT Private IEAT/Private Private IEAT/Private IEAT/Private IEAT/Private IEAT IEAT/Private IEAT IEAT IEAT/Private IEAT IEAT IEAT IEAT/Private

Chonburi Lamphun Rayong Samut Sakhon Saraburi Ayuthaya Samut Sakhon Saraburi Chacheongsao Nakhon Rachasima Rayong Saraburi Ayuthaya Lamphun Nakhon Rachasima Chacheongsao Chiangrai Nakhon Rachasima Pattani Pichit Songkhla Udon Thani Khon Kaen Rayong Surat Thani Narathiwat

2315 – 1555 650 1420 1580 1429 1420 3450 2000 6520 1450 1080 1000 – 197 2375 1376 939 1125 1200 1000 1548 554 1900 622

1594 826 1401 487 900 430 1046 702 1824 1500 5030 1450 532 – 1600 146 – – 565 763 830 – 763 – – 382

– – – – 300 280 – 255 431 – – – 165 – – – – – – – 300 – – 550 – –

110

Under Construction Under Construction Planning Planning Planning Planning Planning

Saha Kabinburi IP

Source:

Suehiro (1996).

Hitech IE No.2 Buriram Phayao Anthani Phrachuap Steel IE Saha Ubon Nakhon Ind

Kabinburi IE

Completion

(continued)

Name

Table 5.4

IEAT/Private IEAT/Private IEAT/Private IEAT/Private Private

IEAT/Private

Private

Economic agents

Ayuthaya Buriran Phayao Prachuap Khirikhon Ubon Raychatha

Kabinburi

Kabinburi

Location

– – – 5310 –

–

–

Total area

800 – – – –

–

–

Area of IZs

– – – – –

–

–

Area of EPZs

Agglomeration of exporting firms in northern Vietnam

111

Iwai Corporation and Sumitomo Corporation – played a signiﬁcant role in the development of IZs in Thailand, Indonesia and the Philippines. The IZs developed by these trading corporations made it easy for Japanese manufacturing companies to invest in other Asian countries, particularly in the 1980s and the 1990s. Thailand Comparison between Thailand and Indonesia shows that IZs were developed in Thailand in the second half of the 1980s (Table 5.5a), prior to their emergence in Indonesia. In 1988, the Lat Krabang Industrial Estate in Thailand, developed by Marubeni and local companies, became available for purchase. The total area of development was 200 ha. It is noted that an EPZ had already been developed at Lat Krabang by the Thai government. Itochu Corporation started to market its Ban Pakong Industrial Park (950 ha) in four phases commencing in 1989. In the same year, Mitsui & Co. completed its sale of the 190-ha Bangkok Industrial Park. Indonesia The development of IZs in Indonesia began in the early 1990s, several years after Thailand (Table 5.5b). The ﬁrst half of the 1990s also saw the Table 5.5a

EPZs and IZs in Thailand

Name

Main developer(s)

Location

1. Ladkrabang EPZ/IZ 2. Bangpakong IZ 3. Bankkadi IZ

Private Thai companies (60%), Marubeni (40%) Itochu (22%)

30 km east of Central Bangkok 60 km southeast of Central Bangkok 40 km north of Central Bangkok 108 km southeast of Central Bangkok

4. Amata Rayon IZ

Mitsui (49%), Toshiba, Private Thai companies Itochu via Bangpakong

Land area (ha)

Starting year of sales

200

19881

950

19892

190

19893

442

19954

Notes: 1. Located in investment promotion zone (exemption from corporation tax for 3 years). 2. Located in investment promotion zone (exemption from corporation tax for 3–7 years). 3. Completely sold; located in investment promotion zone (exemption from corporation tax for 3 years). 4. Located in investment promotion zone (exemption from corporation tax for 8 years followed by 50% reduction for 5 years).

112

Main developer(s)

11 Japaneese companies, including Sumitomo and Bank of Tokyo Lippo Group in Indonesia

PT Surya Semesta Internusa (65%) PT Town & City Properties (35%) (Sumitomo acts as sales agent)

BFIE (private) (55%), Marubeni (45%)

BFIE (private) (40%), Marubeni (60%)

Salim Group (Indonesia) (51%), Taisei (46%), Mitsui (3%)

Sinar Mass Group (Indonesia) (50%), Itochu (50%)

PT Kawasan Industri Jababcka (Nissho Iwai acts as sales agent)

Name

1. East Jalaeta IZ

2. Surya Cipta IZ

3. MM2100 EPZ/IZ (Phases 1 and 2)

4. MM2100 EPZ/IZ (Phase 3)

5. Bukit Indah IZ

6. Karawan IZ

7. Cikarang IZ

Table 5.5b EPZs and IZs in Indonesia

40 km east of central Jakarta 65 km east of Jakarta Airport 55 km from port Tanjeng Prioku

50 km east of central Jakarta

50 minutes drive from central Jakarta 1.3 hours from Jakarta Airport

30 km east of central Jakarta 60 km east of Jakarta Airport 35 km from port Tanjeng Prioku

30 km east of central Jakarta 60 km east of Jakarta Airport 35 km from port Tanjeng Prioku

55 km east of central Jakarta 85 km east of Jakarta Airport 60 km from port Tanjeng Prioku

40 km east of central Jakarta 70 km east of Jakarta Airport 45 km from port Tanjeng Prioku

Location

790

1120

200

400

320

Phase 1: 128 Phase 2: 174

Phase 1: 210 Phase 2: 110

Land area (ha)

19924

1993

1996

1995

19913

19952

19911

Starting year of sales

113

50 km from central Surabaya 45 km from Surabaya Airport

Indonesian government (Min. of Finance) (50%)

East Java provincial government (25%) Surabaya municipal authority (25%) (Sumitomo acts as sales agent)

Phase 2: 324

Phase 1: 150

1995

Notes: 1. Completely sold; favorable result of revised Foreign Investment Act in 1994 (approval of 100% foreign subsidiaries); provides clean water to neighboring villages; creation of 30 000 jobs; total investment of US$2 billion. 2. Developed by Obayashi Corporation. 3. Completely sold. 4. Location of a Japanese bank (Daiwa Purdania Bank).

8. Surabaya PIER EPZ/IZ

114

Agglomeration in Asia

marketing of IZs by Marubeni, Nissho Iwai and Itochu. Marubeni developed a 320-ha site, MM2100, where both an EPZ and an IZ were located, in 1991. Meanwhile, Nissho Iwai began acting as the marketing agent for the 790-ha Cikarang IZ in 1992. Itochu also began marketing the massive 1120-ha Karawan IZ in 1993. Sumitomo was given permission to establish the East Jakarta Industrial Park in 1989 and sales commenced in 1991 with completion in 1995. Sumitomo’s share of investment in the joint venture was 60 per cent and the total area of development was 320 ha. The increase in sales of the IZ can be partially explained by the decision of the Indonesian government in June 1994 to approve 100 per cent foreign subsidiaries, which was eﬀective in marketing the IZ. The development of the IZ created some 30 000 new jobs and approximately US$2 billion was invested from abroad. Consequently, until 1996, ﬁve Japanese trading corporations of the six corporations participated in the development of IZs in Indonesia. The Philippines In the Philippines, ﬁve leading Japanese trading corporations (excluding Itochu) began marketing IZs in 1991 (Table 5.5c). All the IZs are located within a 60-km radius of Metro Manila, the capital of the Philippines. Mitsubishi developed the Laguna Techno Park (220 ha), Mitsui developed the Light Industry and Science Park (106 ha) and Marubeni developed the First Cavite IZ (155 ha) in which JAIDO (the Japanese International Development Organization Ltd) made an investment. Meanwhile, Sumitomo and Nissho Iwai acted as the marketing agent for the Gateway Business Park (160 ha) and the Carmel Ray IZ (280 ha), respectively. The master plan for Carmel Ray IZ was prepared by the Jurong Group of Singapore. All the above IZs were fully occupied mainly by Japanese manufacturing companies, clearly indicating the role played by Japanese trading corporations in assisting overseas investment by Japanese manufacturing companies.

5. IZ DEVELOPMENT BY QUASI-PUBLIC/PRIVATE SECTORS IN VIETNAM This section shows that not only Japanese trading corporations but also other Asian governments or companies established IZs in Asia. Here we illustrate the case of Vietnam.

115

Ayara Land, Mitsubishi, Kawasaki Steel

ICCP (FEBTC) (22%), ICCP (35%), NDC (3%), PALIC (15%), Mitsui (10%), Bechtel (9%)

DLRF (60%), FPL (20%), SMG (15%), OG (5%) (Sumitomo acts as sales agent)

NDC (60%), Marubeni (32%), JAIDO (8%)

Sumitomo (30%), Philippine private companies (70%)

(Nissho Iwai acts as sales agent) Alsons Land (private company) (60%), Marubeni (40%)

Itochu, Cojuangco, local banks

1. Laguna Techno Park

2. Light Industry & Science Park

3. Gateway Business Park

4. First Cavite IZ

5. First Philippine IZ

6. Carnelray IZ 7. Lima IZ

8. Ruicita IZ

Notes: 1. Completely sold. 2. Master Plan prepared by Juron Group in Singapore.

Main developer(s)

EPZs and IZs in the Philippines

Name

Table 5.5c

120 km north of central Manila

50 km south of central Manila 70 km south of central Manila

50 km east of central Jakarta 63 km east of Jakarta Airport 47 km from port Tanjeng

35 km south of central Manila

40 km from central Manila

40 km east of central Manila 40 km east of Jakarta Airport 50 km from Port Manila 80 km from Port Batangas

40 km east of central Manila 40 km from Manila Airport 50 km from port Manila 80 km from port Batangas

Location

Undecided

280 400

300

155

160

143

334

Land area (ha)

After 1997

19912 After 1997

1997

19911

1991

1991

1991

Starting year of sales

116

Agglomeration in Asia

The Indonesian government developed IZs at Batum and Bintan in cooperation with the Singapore government, which is very active in the development of IZs in various Asian countries. Apart from Indonesia, it assisted the development of IZs at Suchou and Wuxi in China and at Song Be in Vietnam. The spread of IZs throughout a country, witnessed in many East Asian countries, is also evident in Vietnam where ﬁve leading Japanese trading corporations (excluding Marubeni) either planned or actually established IZs in 1996 and played a large role in the development of IZs (Table 5.5d). Nomura Securities Co. Ltd also developed an IZ at Haiphong. The development of many IZs in Vietnam began in 1994 and IZs at Noi Bai and Lon Bing also were EPZs. The development site at Tan Tuan, for which Mitsui acted as a marketing agent, was an EPZ. Other sites at Tang Long (Sumitomo), Amata (Itochu) and VSIP (Mitsubishi) were constructed as IZs. It is noteworthy that both the private and the public sectors of several Asian countries took part in the construction of IZs in Vietnam. The VSIP project that Mitsubishi Corporation of Japan joined was a joint project of the Vietnamese government with Jurong Town Corporation of Singapore. Salim Group of Indonesia (a leading business group) also invested in the project. Amata Industrial Estate originated in Thailand. Noi Bai Industrial Estate was developed by VSSB of Malaysia. The Danang EPZ was also constructed by Masscorp of Malaysia, composed of shareholders of the largest 60 ﬁrms in Malaysia, and was established following a proposal by Prime Minister Mahatir. CT&D Group of Taiwan joined the Tan Thuan EPZ near Hochimin. CVEC, a Chinese state-owned enterprise, constructed the Linh Trung EPZ. In short, IZs and EPZs in Vietnam were developed by ASEAN, Japan and China. The same pattern of development as in Vietnam can be found in many Asian countries including China, Indonesia and the Philippines. The role of Mitsui Trading Corporation in the Tan Tuan EPZ was also to promote the sale of IZs, particularly to Japanese manufacturing companies. This means that Mitsui introduced Japanese ﬁrms as tenants to an IZ developed by a Taiwanese company. Other Asian countries pursued the same policy and growth pattern and a network of manufacturing industries in IZs in East Asia was formed. In a country like Vietnam or Myanmar where GDP per capita in the 1990s was US$200/300, it was not easy to attract foreign direct investment for an IZ designed to sell its products in the domestic market (Table 5.5e). Consequently, as the Mingaradon IZ in Myanmar to be developed by Mitsui tried to attract export-oriented businesses, it was an EPZ. The development of IZs by Japanese trading corporations had spread throughout East Asia from the coastal areas of China to Myanmar, as of

117

Bampacon (Thailand), Itochu (70%) Dong Nai Provincial Government (30%)

Nissho Iwai (60%), AGTEX (controlled by

Vietnamese Ministry of Defense) (40%)

4. Amanta IZ

5. Long Birth EPZ/IZ

Sembaan, Julon Town, Temasec, Salim Group, Mitsubishi (5%) City Authority (Mitsui assists sales)

Nomura Group (70%) Haiphong City Authority (30%)

3. Nomura Haiphong IZ

6. Vietnam Singapore Industrial Park (VSIP)

Bien Hoa City 30 km northeast of Ho Chi Minh City

Sumitomo (58%), DAMC (controlled by Vietnamese Ministry of Construction) (42%)

2. Thang Long IZ

Thanh Anh District of Son Be Region 17 km from central Ho Chi Minh City

Bien Hoa City 30 km northeast of Ho Chi Minh City

15 km west of central Haiphong 85 km east of central Hanoi

13 km east of central Manila 14 km east of Jakarta Airport

40 km east of central Manila Adjacent to Noi Bai Airport

VSSB (Malaysia), HICC (controlled by Hanoi City Authority) (Mitsui assists sales)

1. Now Bai EPZ/IZ

Location

Main developer(s)

Name

Table 5.5d EPZs and IZs in Vietnam

500

Phase 2:100

Phase 1:100

Phase 1:93

Phase 1:83 Phase 2:70

Phase 1:128 Phase 2:174

Phase 1:50 Phase 2:50

Land area (ha)

1996

1996

1995

1995

1997

1991

Starting year of sales

118

Agglomeration in Asia

Table 5.5e

IZs in China and Myanmar

Name

Main developer(s)

Location

OECF (40%), Japanese companies (40%), Chinese government (20%)

27 km northeast of central Dalian

217

1992

Suzhou Industrial City

SUDC (China) (35%), SSTD (Singapore) (65% of which Mitsubishi and Mitsui have 2% each)

80 km west of Shanghai

1.52

1996*

Qingtao Coastal Economic and Technology Development Zone

Management Committee of Qingtao Development Zone (Sumitomo acts as sales agent)

3 km from Qingtao International Airport 9 km from Port Qingtao 20 km to urban Qingtao

660

1995

5 minutes walk from 90 Mingaradong Station 10 minutes drive from the airport

1996

China China–Japan Joint Dalian IZ

Myanmar Mingaradong IZ Mitsui (60%), Housing Bureau (40%)

Land Starting year area (ha) of sales

Note: * Modeled after Julon in Singapore.

1996. For example, the 15 IZs developed by Sumitomo Corporation spread all over East Asia. In short, the development of IZs throughout East Asia contributed not only to the movement of FDI but also to trade in goods and services, and in turn created an East Asian development network for the development of IZs.

6.

INDUSTRIAL ZONES IN NORTHERN VIETNAM

Here we shall show the macroeconomic eﬀects of national highway Route 5 on the economy of northern Vietnam and that capital productivity A of the production function in Section 2 can be enhanced by completing infrastructure and building institutions. The Vietnamese economy grew by 8.2 per cent in 1997 when the Asian crisis occurred, but by about only 4 per cent in 1998 and 1999, partly due to the crisis. The information technology industry world boom helped the growth rate to recover to 6.1 per cent in 2000 in Vietnam as it did in other ASEAN countries. The growth rate in

Agglomeration of exporting firms in northern Vietnam

119

2001 was a little short of 6 per cent and was expected to be about 6 per cent in 2002 (Asian Development Bank 2002). We shall show the growth rates of industrial production from 1998, before completion of national highway Route 5, to 2002 after its completion (see Table 5.11). We look at Hanoi, Haiphong and Hungyen in northern Vietnam, Danang in central Vietnam, and Ho Chi Minh in southern Vietnam. The growth rates are higher than the national average and much higher than Ho Chi Minh and show the positive eﬀects of national highway Route 5 on Hanoi and Haiphong in 2002. The national average rates were stable at 11.5, 17.5, 14.1 and 13.9 per cent from 1998 (1994 ﬁxed price, General Statistical Oﬃce 2002). Those for Hanoi were 8.2, 14.8 and 11.1 per cent from 1998 but increased to 24.9 per cent in the ﬁrst half of 2002 (see Miura 2002). Those for Hungyen were high and volatile at 108.7, 21.6 and 18.2 per cent from 1998 since Hungyen is an agricultural area that industrialized partly due to the completion of highway Route 5. Those for Danang were high and stable at 18.9, 17.6 and 20.2 per cent from 1998 but decreased to 19.1 per cent in 2002. Those for Ho Chi Minh were 6.6, 15.4 and 16.1 per cent from 1998 and decreased to 10.4 per cent since Ho Chi Minh is the center of Vietnam and started to develop earlier than other cities. Binhthuan, next to Ho Chi Minh, recorded 33 per cent growth, showing that the growth of Hanoi has begun to diﬀuse over northern Vietnam. We shall show below that foreign investors established factories along highway Route 5 and contributed to high growth rates in northern Vietnam since the level of production was low. Inﬂows of FDI into China in 2001 were the highest and grew at 10.9 per cent, totaling US$69.2 billion. Inﬂows to India and Vietnam recorded high growth rates of 61.1 and 22.4 per cent, respectively. Vietnam’s inﬂows totaled US$2.47 billion. The reasons why the amount approved by the Vietnamese government increased the growth rate were that European companies invested in an electricity supply project and that Canon and its related Japanese companies established factories in EPZs in Hanoi (Nihon Keizai Shimbun, April 9, 2002). Thang Long Industrial Park and Nomura Haiphong Industrial Zone Highway Route 5 is 100 km long and links Hanoi in the west and Haiphong in the east (see Figure 5.1). Hanoi is the capital of Vietnam and Haiphong is a port city. Thang Long Industrial Park (TLIP) was established by Sumitomo Corporation and located in Hanoi. Nomura Haiphong Industrial Zone (NHIZ) was established by Nomura Security Company and located in Haiphong. In turn, Vietnamese companies agglomerated in the industrial zone in Hungyen along Route 5.

120

Agglomeration in Asia

The developed areas of TLIP, NHIZ and Hungyen are 121, 180 and 100 ha, respectively. In 2002 TLIP started to develop 77 ha in the second phase and plans to expand another 89 ha in the third phase. TLIP is near Noibai International Airport and plans to have a logistics center of 45 ha, an international school, an international hospital, housing for foreigners and a Hanoi new town of 100 ha, complete with schools, hospitals and housing. NHIZ has its own facilities for electricity supply of 50 MW, water supply, a sewage processing plant and 2,000 telephone circuits for those preparing to rent standard factories. We quote the incentives outlined by TLIP and NHIZ to tenant companies (interviews, September 11 and 12, 2002). TLIP has the following six: (i) administrative functions in the capital, Hanoi; (ii) Hanoi is located centrally between Ho Chi Minh, Bangkok, Kunming in China, and Hong Kong; (iii) the educational level of labor around Hanoi is high since 60 per cent of Vietnam’s national universities are located in the Hanoi region; (iv) unskilled labor is abundant and cheap; (v) urbanization projects are progressing and infrastructure is well advanced with further plans; and (vi) there are supporting industries to supply parts and components. NHIZ has the following seven: (i) the local government of Haiphong gives preferential treatment to tenant companies; (ii) Haiphong has a port to export products; (iii) the zone has its own electricity supply of 50 MW and other high-quality infrastructure; (iv) tenant companies quickly start production at the rental factory; (v) they can employ high-quality workers at low wages; (vi) the one-stop service provided by NHIZ makes it easy to obtain investment licenses; and (vii) the department of investment services supports aftercare for tenant companies. Both TLIP and NHIZ oﬀer as a common incentive good-quality infrastructure. TLIP has national highway Route 5 from Hanoi to the Haiphong port (105 minutes from TLIP to the port), an extension project at Noibai International Airport, national highway Route 18 linking Hanoi with Cairong port, projects for electricity supply such as thermal power generation and a transformer substation (we interviewed the ﬁrst group of foreign investors at Sumitomo Corporation on August 1, 2001). NHIZ analyses the positive eﬀect in a relatively short time due to completion of national highway Route 5 as follows: it takes 1 hour and 15 minutes by car from Hanoi to NHIP compared with 3 to 4 hours before completion of national highway Route 5. It takes 15 minutes from NHIP to Haiphong port compared with the previous 30 minutes. The amount of freight dealt with at Haiphong port was expected to be 4.7 million tons in 2000 but reached 7.5 million tons. It was 8.6 million tons in 2001 and the amount was about 9.5 million tons in 2002, which is twice as much as initially predicted. One noticeable problem of NHIZ in inviting Japanese companies is the lack of schools and hospitals.

Agglomeration of exporting firms in northern Vietnam

121

Decisive Factors of Foreign Investors One factor that inﬂuences foreign investor decisions on investment is capacity building within recipient countries. Capacity depends on the following four conditions: (i) human resources; (ii) infrastructure; (iii) living conditions; and (iv) institutions. We shall explain each of them below. The existence of a package of public goods will help to attract FDI and promote investment in industrial zones, and increase macroeconomic growth. We explained a theoretical model in Section 2, considering northern Vietnam as an economic unit. Human resources It is well known that in Vietnam wages are low and quality of labor is high. Table 5.6a shows that the minimum wage in Vietnam is US$37 per month while those of Thailand, Indonesia and the Philippines are US$96, US$64 and US$129, respectively. A survey also showed that wages in Haiphong are 10 per cent lower than in Hanoi (Table 5.6b). Many projects are aimed at developing human resources in Vietnam. The Hanoi Institute of Technology has a project to teach technology in machine processing, metal processing and electrical control. The Haiphong Hi-tech Skill Training School was established in December 2001. Its main subjects are information and graphic, electric and electronic engineering, polymers, welding, milling and so on. The students of both schools graduated for the ﬁrst time in 2003. Roize Robotech Inc., a Japanese company, contributed to fostering skilled labor in machine processing by donating industrial machines to a university in Haiphong. Infrastructure The Japanese ODA loans contributed to facilitating infrastructure in northern Vietnam. National highway Route 5 and the Haiphong port, which were constructed and redeveloped by the loans, are eﬀective in forming industrial agglomeration. Figure 5.1 shows that ¥21 billion was provided for the construction of highway Route 5 in 1993 and 1995 and that ¥17.3 billion was provided for the redevelopment of Haiphong port. Loans that contributed to making logistics in northern Vietnam more eﬃcient were as follows: ¥12.5 billion for improving the Hanoi transportation network, ¥2.5 billion for constructing New Duong Bridge, ¥23.5 billion for national highway Route 18, ¥10.3 billion for expanding Cairong port, ¥6.8 billion for constructing Bai Chay Bridge, ¥8 billion for constructing the Bihn Bridge, and ¥30.4 billion for improving highway Route 10. Remaining infrastructure problems in northern Vietnam are to ﬁnish the work on national Route 18 and Cairong port, which will complete the

122

Source:

64 13

Jakarta

Indonesia

37 8

129 11

Manila

43 8

100 11

Sebu

96 13

First zone (third zone is partly included)

Philippines

Center of Hanoi Ho Chi Minh City

Vietnam

Thang Long Industrial Park

Sumitomo Corporation 2002.

Legal minimum wage ($US) National Holidays (days)

Country

Legal minimum wage ($US) National holidays (days)

Country

Table 5.6a Comparison of wages, April 2002

51 12

Shanghai

83 13

69 12

Shenzhen

77 13

Other areas of the third zone

China

Second zone, principal cities of the third zone

Thailand

Agglomeration of exporting firms in northern Vietnam

Table 5.6b

Comparison of wages in Vietnamese cities

City 1. 2. 3. 4. 5.

123

Wages/month (VND)

Ho Chi Minh City Dong Nai Hanoi Da Nang Haiphong

1 361 000 1 111 000 745 000 738 000 660 000

Note: US$1 VND 15 000. Source:

Vietnam Investment Review, January 21, 2002.

logistic system between Hanoi and Cairong port, and infrastructure to link the triangle of Hanoi, Haiphong and Cairong. Northern Vietnam will also have close industrial linkage with southern China, including Hong Kong. Living conditions We shall limit the discussion to living conditions for the Japanese, since it is Japanese companies that mainly invest in TLIP and NHIZ. Apartments, supermarkets, restaurants, hotels, direct air ﬂights from host countries to recipient countries, schools, hospitals and amusement facilities are key to attracting foreign investment. We shall explain below that the Hanoi area recently fullﬁlled the key condition. There are such apartments for foreigners as Daewoo, V Tower, Sedona, Somerset Hanoi Tower, MayFair Apartments, Hanoi Garden Lodge, the Hanoi Club, Flower Villages and Somerset Lakeside. Rental fees have dropped because of the increased supply of apartments in 2002. The Japanese supermarket chain Seiyu opened in 1998, selling Japanese food produce. By 2002 there were also many mini marts. There were 15 Japanese restaurants at the peak and 10 in September 2002. It is possible to buy fresh ﬁsh along Route 5. The Vietnamese food restaurants the Emperor, Nam Phuong, Soho and Cha Ca La Vong target foreigners as customers. In 1997 only the Hanoi Hotel and Soﬁtel Hotel existed as four- or five-star establishments while other smaller hotels were converted private houses. Since then, however, the Soﬁtel Plaza, Hilton, Daewoo, Media and Nikko hotels have been constructed. As a consequence, hotel rates in Hanoi have become lower. Direct air ﬂights between Tokyo and Hanoi started in July 2002. Hanoi is hoping to attract Japanese companies’ investment since there is a Japanese school and a high-quality hospital. However, there are some problems with living conditions. For example, there is no large shopping mall and no bookstore or suitable convenience store.

124

Hanoi

Haiphong

Redevelopment of Route 18 (1993, 1999 / US$235 million)

Infrastructure of Japanese logistics in northern Vietnam

Japan Bank for International Cooperation (2003).

Figure 5.1

Source:

To Ho Chi Minh City

Reform of Route 5 (1993,1995 /US$210 million)

Reform of traffic network in Hanoi (1998 / US$125 million)

Infrastructure maintenance in northern Hanoi (1996 /US$114 million)

Construction of New Duong Bridge (1993, 1995, 1996, 1998, 2000 / US$25 million, including Route 1)

Redevelopment of Route 10 (1997, 1999 / US$304 million)

Construction of Bihn Bridge (1995, 1999 / US$80 million)

Redevelopment of Haiphong Port (1995, 1999 / US$173 million)

Construction of Bai Chay Bridge (2001 / US$68 million)

Cairong

To China

Expansion of Cairong Port (1995 / US$103 million)

Agglomeration of exporting firms in northern Vietnam

125

Institutions One-stop service plays a large role in streamlining investment procedures. This means that at the NHIZ oﬃce, tenant companies can obtain all the required approval from ministries related to investment licenses, factory operation on export procedures and so on. For example, since December 2001 the NHIZ oﬃce has been able to obtain investment licenses within three days. The oﬃce is responsible for hiring employees for tenant companies by asking the Haiphong EPZ agency to recruit them. Streamlined customs clearance helps reduce tenant companies’ costs. We can illustrate this in the case of TLIP. Dragon Logistics Center is a private Japanese logistics company located in TLIP, whose functions are as follows: to reduce transportation costs by using containers and a trailer terminal, and realize smooth positioning of containers; to list bonded cargo by name of vendor; to list bonded cargo by name of exportprocessing companies; to control cargo to stock by a system of warehouse management; to make customs clearance eﬃcient by using customs oﬃcers based at Dragon Logistics Center; and to allow customers within TLIP and near Hanoi to specify a given time for Dragon to deliver small packages. Two Japanese companies invested 52 per cent of the company’s capital of US$ 4 million and three of the 146 employees at Dragon Logistics Center are Japanese. Industrial zones in Asia oﬀer preferential tax treatment. Corporate tax in Vietnam is exempted after a company makes a proﬁt for four years. After four years the tax rate is 5 per cent for a further four years and 10 per cent thereafter (see Table 5.7). Corporate tax in China is exempted after a company makes a proﬁt for two years. After two years the tax rate is 7.5 per cent for three years, then 15 per cent. It is worth noting that the corporate tax rate in Indonesia is 30 per cent or three times higher than in Vietnam. Tax treatment is a crucial incentive for FDI. The Japan Bank for International Cooperation (2003) survey showed that institutional reform is key to developing the private sector in Vietnam (see Figure 5.2). Japan’s Minister of Finance in April 1999, Kiichi Miyazawa, pledged to provide ¥20 billion to support programs for developing the private sector as requested by Vietnam’s Prime Minister Phan Van Khai. The loan was agreed and implemented in September 1999. The projects are divided into the following three categories: (i) ﬁnancial climate, (ii) business climate, and (iii) organizational treatment. With regard to the ﬁnancial climate, the Vietnamese government established a development support fund, a two-step loan fund, and an export support fund. Concerning the business climate, training was organized based on the law for promotion of domestic investment, controls on industrial property rights were reformed and strengthened, an ordinance

126

EPZ

China

Source:

15–24

5% of incomes

30

2 years 31⁄2 years*

4–8 years

No

8 years 51⁄2 years

5 years

3 years

4 years 41⁄2 years

Tax exemption

Sumitomo Corporation, 2002.

* EPZs and IZs. The minimum rate is 10%.

PEZA

Philippines

Note:

EPTE (KB)

Indonesia

30

30

Second zone

Third zone

30

First zone

Thailand

10

EPE

Vietnam

Tax rates (%)

Corporate tax

Table 5.7 Comparison of tax systems

Supporting industry: 8 years

Supporting industry: 8 years

High-tech company: 8 years

Other tax exemption

Exempted

10

10

10

10

10

3

Tax on proﬁt remittance (%)

Exempted

Exempted

Exempted (allotment tax)

Exempted

Exempted

Exempted

Exempted

Tariﬀ on imports of materials

Exempted

Exempted

10

10

10

Exempted

Value added tax on imports (%)

Exempted Exempted in the case of a high-tech company

Exempted

Exempted

Exempted

Exempted

Exempted

Exempted

Tariﬀ on imports of facilities

127

Institutional reforms

Japan Bank for International Cooperation (2003).

Data collected from private companies only.

Figure 5.2

Source:

Note: 0

50

100

150

200

250

300

0. Abolished the approval system for new business (changed to a registration system) 1. Abolished sub-licenses 2. Improved the collateral and access to banks 3. Liberalized trade

Enterprises

4. Expanded funding policies (Development Assistance Fund, Export Assistance Fund and other) 5. Promoted the establishment of business associations 6. Improved bidding regulations 7. Simplified government procedures and improved government-related procedures: Onestop services 8. Promoted reforms of the state-owned enterprises

9. Others

128

Agglomeration in Asia

on implementation of corporate law was proclaimed, and laws and regulations were made transparent. In order to improve the investment climate for foreign investors, a request for the use of local content was reduced, a request for procuring foreign currency was deleted, Visa application was streamlined, an institution for control of 100 per cent ownership by foreign investment was clariﬁed, and dual prices for telephone and water fees were abolished. With regard to the organizational treatment, the Authority for Small- and Medium-sized Industries and a Private Sector Development Committee were established, and job training schools were strengthened. According to the survey, private companies positively rated the abolition of the industries restricted or prohibited by some ministries, change from an approval to a registration system to establish companies, and trade liberalization. The number of restricted or prohibited industries was reduced from 400 to 250. The change in establishing companies has streamlined administrative procedures and reduced the average time required to less than one month compared with as long as three months before 2000. All of the private companies were permitted to export and import without licenses to guarantee free trade. As a result, the institutional reforms have been highly rated. Conversely, low interest loans and new loans to banks were negatively rated. Private companies found procedures unfavorable since low interest loans of the two-step loan fund and export support fund were not wellinformed, were favorable to state-run enterprises, and had complicated procedures for borrowing. The amount of loans provided to private companies did not increase, though the upper limit on interest rates for lending was abolished. However, we can conclude from the survey that institutional reforms for streamlining the procedures are eﬀective in developing the private sector and promoting investment. Industrial Clusters in Northern Vietnam We can explain economic development in Hanoi by dividing it into periods 1 and 2, according to a Sumitomo Corporation Hanoi member of staﬀ whom we interviewed on September 13, 2002. Toyota, Honda and Yamaha established factories in Vinh Phuc industrial zone near Noibai Airport in period 1 until 2000. Canon, Sumitomo Bakelite and Toto established factories in TLIP after national Route 5 was constructed and Haiphong port was redeveloped in period 2 starting from 2002. A staﬀ member of TLIP said that the companies chose TLIP to avert risk in China though they had established factories there.

Agglomeration of exporting firms in northern Vietnam

129

Hanoi is centrally positioned from Ho Chi Minh, Bangkok, Kunming and Guangzhou in China. It is located 1100 km from Ho Chi Minh, 950 km from Bangkok, 600 km from Kunming and 850 km from Guangzhou. Therefore we can expect northern Vietnam to be integrated into southern China from the supply chain management viewpoint in the future. The following subsections explain industrial clusters in Hanoi and Haiphong that are located west and east of national Route 5. Industrial cluster in Hanoi Table 5.8 shows that Canon anchored an industrial cluster in Hanoi. Sumitomo Corporation began selling TLIP land lots in June 2000. Two companies in 2000, six in 2001 and 11 in 2002 signed up. Canon started production in April 2001, but Parker Processing VN Co., whose products are paint and surface treatments for metal parts, had moved into TLIP in August 2000 to provide parts to Canon. Volex Cable Assembly started producing power supply cords and interconnectors in 2001. The Singaporean company began to provide products to Canon, though it was not its intention to sell solely to Canon. Companies that provided parts to Canon decided to move into TLIP, particularly in 2002. One such company is Sumitomo Coil Center, which produces parts for printers, another is a Japanese company producing dyecasting products, and a third is a Malaysian company, Santomas VN Co., which produces precision plastic injection molding. Thus Canon is an anchor company that leads other companies to provide parts and components. It is characteristic of 13 of the 42 Japanese companies in Hanoi in June 2002 that they are from the automobile and motorcycle industries and related parts and components (see Table 5.8). The motorcycle industry in China had overcapacity. Vietnam imported 2 million motorcycles in 2001 mainly from China and even prohibited imports temporarily. It is generally thought that Vietnam will be unable to form a cluster for the motorcycle parts and components industry, since development of a highway network making transportation between Vietnam and China convenient and the close relationship between the northern Vietnam economy and the economy of southern China would seem to work against it. Industrial cluster in Haiphong Table 5.9 shows that the number of NHIZ’s tenant companies increased from one or two between 1996 and 1999, to ﬁve in 2001 and 11 in September 2002. Four of the 11 have Hong Kong capital and three of the four are textile companies. The four are Oﬃce Xpress Manufacturing producing stationery, Vietphong Garment & Textile Co. producing ladies’ knitwear, R&T Manufacturing VN Co. producing textile fabrics and yarn, and BT

130

Japan Japan Japan Japan Malaysia Japan Vietnam Japan

April 11, 2001 October 4, 2001 November 5, 2001 November 5, 2001 January 9, 2002 March 22, 2002 n.a. June 13, 2002 July 18, 2002 March 1, 2002 April 29, 2002

Mitsubishi Pencil VN Co., Ltd Vina KDC Wiring Industries Ltd Parker Processing VN Co., Ltd Volex Cable Assembly (VN) Co., Ltd C5 Canon VN Co.,Ltd 6 Sumitomo Bakelite VN Co., Ltd 7 Denso Manufacturing VN Co., Ltd 8 TOA VN Co., Ltd C 9 Santomas VN Co., Ltd C10 Abe Asian Tech Hanoi Ltd C11 Dragon Logistics Co., Ltd C12 Matsuo Industries VN Inc.

C13 Ohara Plastic VN Co., Ltd 14 TOTO VN Co., Ltd 15 Sakurai VN Ltd

Japan Japan Japan

Japan Japan Japan Singapore

November 29, 2000 January 15, 2001 August 8, 2000 August 9, 2001

1 2 C3 C4

Nationality

Date of establishment

Thang Long Industrial Park: list of tenants

Company

Table 5.8

0.52 (rental factory) 7.2 1

1 0.50 (rental factory) Rental oﬃce 5 1.26

20 6.55 3

3.8 0.48 (rental factory) 2.31 0.47 (rental factory)

Land rental

Security camera Precision plastic injection molding Film & manuals Logistic services Plastic molding parts & steel processing Parts for automobiles & others Plastic molding products Sanitary ware Parts of machine tools, machines, laser beam machines, semiconductor equipment

Ink jet printers Flexible printed circuit boards Parts for automobiles

Writing implements Wire harness & power supply cord Paint & surface treatment for metal parts Power supply cord, interconnectors

Products

131

Fujikin VN Co., Ltd

Source:

Note:

Total

Japan Japan

August 30, 2002

Japan

July 3, 2002

July 3, 2002

Sumitomo Corporation, June 2002.

C3, C4 and C9–13 provide parts and components to Canon.

17 Yabashi VN CAD Technology Corporation 18 Seed VN Co., Ltd

16

54.2 ha

0.60 (rental factory)

Rental oﬃce

0.52 (rental factory)

Super precision ﬂow control systems, equipment and parts Designs, design processing and software products Manufacturing and sale of stationery products

132 N-7–N-12 (6 ha) F-2, F-3, F-4 (3 ha)

Korean

Japanese

Japanese Japanese Japanese Japanese

Japanese

Japanese Japanese

Hop thinh Co., Ltd As’ty Vietnam Inc. HI-Lex Vietnam Inc. Ortec Chemical Co., Ltd

Estelle Vietnam Inc.

Meicorp Vietnam Inc. PV Haiphong Inc.

Standard Factory C – 3rd ﬂoor F-7a (0.5 ha) Standard Factory C – 1st ﬂoor Standard Factory A – 1st ﬂoor

E-4 (1 ha) A-5 (1 ha) C-8 (1 ha) J-1 (0.5 ha)

Standard Factory B – 4th ﬂoor

Taiwanese

Chinese

F-12b (0.5 ha)

Hong Kong Hong Kong

Hong Kong

Standard Factory B – 2nd & 3rd ﬂoors Standard Factory A – 2nd ﬂoor Na-16 (1 ha) Standard Factory A – 4th ﬂoor Standard Factory B – 1st ﬂoor

Hong Kong

Location

Vietphong Garment & Textile Co., Ltd R&T Manufacturing VN Co., Ltd BT Garment Co., Ltd Oﬃce Express Manufacturing Co., Ltd Taiwan Fong Tai Paper Co., Ltd Vietnam Hoa Nguyen Garment Co., Ltd Shin Yong Chemical Vietnam Co., Ltd Rorze Robotech Inc.

Capital

Nomura Haiphong Industrial Zone: list of tenants

Company

Table 5.9

Ring materials Assembly of gas appliances

Manufacturing equipment of semiconductors Sewing Bags Control cables for motorcycles Treatment chemicals of industrial water Jewelry processing

Plastic products

Corrugated cartons, packaging materials Clothing, sewing

Textiles, sewing, knitwear Staples (ring ﬁle)

Textiles, sewing, knitwear

Ladies’ knitwear

Products

February 2, 2001 September 3, 2002 February 23, 2001 March 6, 2001

October 5, 1996 July 28, 1997 March 16, 1999 October 28, 1999

October 2, 1996

September, 2002

September 8, 2002

February 1, 2002

August 23, 2002 January 8, 2002

April 16, 2002

March 28, 2002

Date of investment permission

133

Nomura Haiphong Industrial Zone.

Japanese Japanese

Vietnam Fuji Mold Co., Ltd Nishishiba Vietnam Co., Ltd

Source:

A-10 (1 ha) A-9 (1 ha) Standard Factory E

Japanese Japanese Japanese F-8a (0.6 ha) Power Plant Oﬃce

C-5, C-6 (2 ha) L-1, L-2, L-3, L-7, L-8 (5 ha)

Japanese Japanese

Nichias Haiphong Co., Ltd Yazaki Haiphong Vietnam Co., Ltd Hiroshige VN Corporation Maiko Haiphong Co., Ltd Vina-Bingo Co., Ltd Electronics parts Microscopic bearings Precision parts processing for semiconductors Plastic molding products Power plant and maintenance

Sealing materials Wiring harness for automakers

July 26, 2002 October 8, 1996

March 14, 2002 April 15, 2002 June 27, 2002

July 17, 2001 September 17, 2001

134

Agglomeration in Asia

Garment Co. producing knitwear and textile fabrics. The companies’ investment in NHIZ means that these Hong Kong companies chose to locate in northern Vietnam instead of southern China due to increased costs. One characteristic of NHIZ in 2002 is that Taiwanese, Korean and Chinese companies invested in NHIZ, a Japanese industrial zone, to save costs.

7.

SUMMARY AND CONCLUSIONS

We obtained the following three results. First, we built a model to theoretically show that IZs as quasi-public goods enhance aggregate growth under given conditions. Second, we illustrated that there are two types of players to provide IZs for quasi-public goods. The players were in both the private and the quasi-public sectors in East Asia in the latter half of the 1980s and the ﬁrst half of the 1990s. One agent for them was a Japanese trading corporation (Sogoshosha) in the private sector. The other was the Industrial Estate Authority of Thailand in the quasi-public sector in Thailand. The countries in East Asia could not attain high economic growth without the players. Third, we can deduce from the ﬁrst and second results that IZs were quasi-public goods crucial to attracting FDI, and diﬀusing growth throughout a region to attain high economic growth. We showed that a combination of industrial zones with preferential tax, institutional reforms, and physical infrastructure including national highway Route 5 and Haiphong port formed an industrial cluster in northern Vietnam as is shown in Figure 5.3. We emphasized the role of economic agents or players in forming clusters. Private companies constructed industrial zones. Japan’s oﬃcial development assistance supported construction of highway Route 5, redevelopment of Haiphong port and institutional reforms. The Vietnamese government gave tax incentives. Tables 5.10 and 5.11 show these players contributed to the agglomeration of ﬁrms and the enhancement of aggregate growth in the northern Vietnamese economy. Table 5.12 summarizes the Vietnamese model of forming industrial clusters.

135

Figure 5.3

(ODA) Route 5, Haiphong Port Institutional reform (One-stop service, tax incentives)

Japanese companies, other foreign and local companies

Development of northern Vietnam

Industrial clusters

Canon, Honda, Panasonic

Building

Capacity

Nomura Haiphong IZ

Haiphong Haiphong Port

Related and other companies

Anchor companies

+

Route 5

Northern Vietnam

Industrial clusters in northern Vietnam

Canon effects

Thang Long IP Nomura Haiphong IZ

Thang Long IP

Hanoi

136

Table 5.10

Agglomeration in Asia

Japanese companies in the suburbs of Hanoi (June 1, 2002)

Company 1 Inoue Rubber 2 3 4 5 6 7 8 9 10 11 12

Kyoden Goshi Giken Stanley Electric Co., Ltd Sumitomo Metals Daihatsu TakaNicht Denso Toyota Nissin Kogyo Co., Ltd Nippon Sheet Glass Co., Ltd Nippon Carbide Industries Co., Ltd 13 Honda 14 Matsuo Industries Inc. 15 16 17 18 19 20 21

Yamaha Pentax Ajikawa Steel Construction Abe INAX Ebara Kawamura Harness

22 Canon 23 Kyoei Steel 24 Sakurai Manufacture place Co., Ltd 25 Shimadzu 26 Shiroki 27 Shinken 28 Sumitomo Electric/Sumitomo Wiring Systems, Ltd 29 Sumitomo Bakelite Co., Ltd 30 Daiwa Plastic 31 Tsukuba Diecast 32 TOA 33 TOTO 34 NEC 35 Nippon Leakless Co. 36 Parker Processing Co., Ltd 37 Hino

Products Tires and tubes for automobiles and motorcycles Plastic molding for two-wheeled vehicles Metal molding for two-wheeled vehicles Lights for automobiles and motorcycles Metal component for automobiles Commercial car Automobile seats Parts and components for car engines Automobiles Brakes for automobiles and motorcycles Glass for automobiles and building materials Components for automobiles Two-wheeled vehicles Plastic molding parts & steel processing parts for automobiles Two-wheeled vehicles Cameras and optical instruments CAD design Manual printing and manufacturing Sanitary ware Pumps Harnesses for domestic electric appliances and power cords Ink jet printers Cast iron Parts and components for machine tools X-ray devices Moldings Markers and ball-point pens Wiring harnesses for electric products Flexible printing circuit boards Plastic moldings Aluminum diecast Surveillance cameras Sanitary ware Digital telephone exchange machines Gaskets Special coatings for cellular phones Trucks and buses

Agglomeration of exporting firms in northern Vietnam

Table 5.10

137

(continued)

Company

Products

38 39 40 41 42

Optical transmission devices Clothing products Structural steel Pencils Glassware

Fujitsu HOEI MES Mitsubishi Pencil Co., Ltd Ryukyu Glass

Note: 1–15 are companies related to two- or four-wheeled vehicles. Source: Sumitomo Corporation.

Table 5.11

North

Central South

Growth rates of Vietnamese industrial output (%)

Hanoi Haiphong Hungyen Danang Ho Chi Minh City

1999

2000

2001

2002

8.2 17.6 108.7 18.9 6.6 11.5

14.8 19.5 21.6 17.6 15.4 17.5

11.1 20.0 18.2 20.2 16.1 14.1

24.8 24.9 n.a. 19.1 10.4 13.9

Source: General Statistical Oﬃce (2002).

Table 5.12

Capacity building

The Vietnamese model of fostering industrial clusters Pre-conditions

Tasks

Economic agents

Industrial zones

Site development Tax incentives Roads, ports Investment climate Human resources Restaurants, shops

Japanese trading corporations Central and local governments ODA The government and ODA ODA Foreign companies

Infrastructure Institutions Living conditions

138

Agglomeration in Asia

REFERENCES Asian Development Bank (2002), Asian Development Outlook 2002, Oxford: Oxford University Press. Barro, R.J. and X. Sala-i-Martin (1995), Economic Growth, Cambridge, MA: MIT Press. Fujita, M. and N. Hamaguchi (2001), ‘Intermediate goods and the spatial structure of an economy’, Regional Science and Urban Economics, 31, 79–109. General Statistical Oﬃce (2002), Vietnam Statistical Yearbook 2002, Hanoi: Statistical Publishing House. Harian, Benta (1992), ‘25th year anniversary of MIDA’, Kuala Lumpur, August 22. Japan Bank for International Cooperation (2003), Ex-post Evaluation Report on ODA Loan Project, Tokyo: Japan Bank for International Cooperation. Kawakami, M., K. Otsuka and T. Sonobe (2003), ‘Changing roles of innovation and imitation in industrial development: the case of the machine tool industry in Taiwan’, Economic Development and Cultural Change, Chicago: University of Chicago Press. Kennedy, L. (1999), ‘Cooperating for survival: tannery pollution and joint action in the Palar Valley (India)’, World Development, 29 (9), 1673–91. Krugman, P. (1993), ‘The hub eﬀect’, in W.J. Ethier, E. Helpman and J.P. Neary (eds), Trade Policy and Dynamics in International Trade, Cambridge: Cambridge University Press, pp. 29–37. Kuchiki, A. and K. Yamada (1997), ‘Lessons from Japan: industrial policy approach and the East Asian trial’, in L. Emmerij (ed.), Economic and Social Development into the XXI Century, Baltimore, MD: Johns Hopkins University Press, pp. 359–93. Miura, Y. (2002), ‘Viet Nam Report’, 20 September edition, Japan Research Institute, Tokyo. Mori, T. and K. Nishikimi (2002), ‘Economies of transport density and industrial agglomeration’, Regional Science and Urban Economics, 32, 167–200. Suehiro, A. (1996), ‘Expanding central and local areas in Thailand’, Global Area Studies, Tokyo. Torii, K. (1994), ‘Malaysia’ (in Japanese), Development and Environment, 4, Institute of Developing Economies, Tokyo. Urata, S. (2001), ‘Emergence of an FDI-trade nexus and economic growth in East Asia’, in J.E. Stiglitz and S. Yusuf (eds), Rethinking the East Asian Miracle, Oxford and New York: Oxford University Press, pp. 409–59. Weijland, H. (1999), ‘Microenterprise clusters in rural Indonesia: industrial seedbed and policy target’, World Developing, 27 (9), 1515–30. World Bank (1993), The East Asian Miracle, Oxford: Oxford University Press. World Bank (2002), Poverty Reduction Strategy Papers: Good Practices, Washington, DC: World Bank.

6. Industrial agglomeration and regional growth in Korea: focusing on the software and IT service sector Yasushi Ueki 1.

INTRODUCTION

Japan and its followers, the newly industrialized economies (NIEs) and the Association of South East Asian Nations (ASEAN), are known for their highly manufacturing-based economies. With help from government industrial policies, including building export-processing zones (EPZs) and preferential treatment to investors, East and South East Asia have succeeded in attracting foreign direct investment (FDI) from developed countries or in promoting intra-regional FDI. As a result, the region has become the cornerstone of global production networks especially for electrical domestic appliances, computers and peripherals. However, this international division of labor is changing as Mainland China is emerging as a global manufacturing base. China’s plentiful supply of cheap labor and giant market potential attract FDI from all over the world. As Ueki (2003) anticipated, companies began restructuring their production networks. Japanese companies have relocated or plan to relocate 22 bases from ASEAN5 (Indonesia, Malaysia, the Philippines, Singapore and Thailand) to China since 2001 (Nihon Keizai Shimbun, July 25, 2002). These phenomena force countries in the region to open new avenues to establish vital and high-growth economies. Seeing the expansion of new frontiers such as information technology (IT) and biotechnology (often called ‘knowledge-based industries’), most of the Asian countries began shifting their policy priorities to these knowledge-based industries. Generally speaking, these industries greatly depend on cutting-edge technologies. In addition, the speed of technological innovation and product cycle are so rapid and competition is so intense that it is very diﬃcult for innovators to have a technological edge unless they can continue to be innovative. So it has become a top priority mission for industrial 139

140

Agglomeration in Asia

policy makers to build an environment conducive to innovation and entrepreneurship. One of the radical and ideal models for the development of the IT sector can be found in Silicon Valley in the US. Many countries have learned lessons from Silicon Valley’s experience, which is based on relationships among governments, universities, venture businesses, venture capital and public/private supportive/coordination bodies. Another success story, which is the envy especially of developing countries, is Bangalore in India, whose software cluster is based on a system of international division of labor between the US and India. In this case, Indian universities played an important role, mainly as the source of knowledge workers. Observing that one of the common characteristics of the two systems was that universities played a critical role in the formation and development of the industrial clusters, it became one of the consensuses that high-quality engineering universities are indispensable to building an IT cluster in a region. Many governments began planning and implementing policies for promotion of knowledge transfers from universities and/or research institutions. The aim of this chapter is to analyse the factors that have contributed to the growth in industrial agglomerations, speciﬁcally the software industry in Korea, by using regional data in order to analyse the change in the number of employees by industry and by region. The reason for focusing on Korea was the country’s rapid recovery from ﬁnancial crisis in 1997. It is noteworthy that Korea emerged as a frontrunner in broadband dissemination and changed its economic structure to a more Internet-based economy in the process of recovery from the crisis (Ueki 2004).

2. INDUSTRIAL ORGANIZATION AND THE IT INDUSTRY Distribution of Industries by Region According to the data from regional statistics reports issued by the Korean government, business activities are highly concentrated in the capital sphere, which consists of Seoul and its peripherals. Table 6.1, which arranges data on the total number of all 43 industries, indicates Seoul’s dominant position. About one-quarter of all companies and workers are located in Seoul. By adding neighboring Gyonggi, more than 40 per cent of them were represented by the two areas. The concentration of geographical distribution by companies and workers was enhanced between 1998 and 2000. This can be observed from the Herschman–Herﬁndahl Index (HHIi i (share of i) 2), which is a measure of

141

Industrial agglomeration and regional growth in Korea

Table 6.1

Number of establishments and workers by administrative unit Establishments

Seoul Busan Daegu Incheon Gwangju Daejeon Ulsan Gyeonggi Gangwon Chungbuk Chungnam Jeonbuk Jeonnam Gyeongbuk Gyeongnam Jeju Total

Workers

1998

(%)

2000

(%)

610 363 231 247 146 936 117 398 74 520 75 924 50 230 390 677 98 075 82 681 103 906 108 298 117 610 154 551 171 601 32 708

23.8 9.0 5.7 4.6 2.9 3.0 2.0 15.2 3.8 3.2 4.0 4.2 4.6 6.0 6.7 1.3

647 609 234 255 151 937 127 625 79 301 78 768 52 941 434 566 100 648 85 583 105 937 108 483 118 395 158 621 177 906 35 498

24.0 8.7 5.6 4.7 2.9 2.9 2.0 16.1 3.7 3.2 3.9 4.0 4.4 5.9 6.6 1.3

1998 2 923 458 913 888 559 318 543 739 306 742 308 325 274 581 1 861 998 341 179 341 884 409 572 399 853 414 816 632 240 737 180 118 069

(%)

2000

26.4 3 143 416 8.2 950 743 5.0 591 768 4.9 605 334 2.8 344 859 2.8 325 088 2.5 310 920 16.8 2 208 717 3.1 357 422 3.1 379 731 3.7 449 549 3.6 418 237 3.7 436 593 5.7 673 902 6.6 813 409 1.1 128 822

(%) 25.9 7.8 4.9 5.0 2.8 2.7 2.6 18.2 2.9 3.1 3.7 3.4 3.6 5.6 6.7 1.1

2 566 725 100.0 2 698 073 100.0* 11 086 842 100.0 12 138 510 100.0

Note: * Rounded. Source: National Statistical Oﬃce, Korea.

the degree of concentration. Higher HHI indicates more intensive concentration. HHI calculated on the number of establishments increased slightly from 0.112 in 1998 to 0.114 in 2000. HHI on workers also increased from 0.125 to 0.127. IT, R&D and the Education Sector The IT industry consists of hardware, software and IT service. According to the industrial classiﬁcation listed in Appendix Table 6A.1, it seems acceptable to deﬁne the hardware sector as corresponding to ID 19: computer/oﬃce machinery, and software and IT service (hereafter software) to ID 37: computer and related activities. As observed in Table 6.2, only 0.4 per cent of the Korean workers are engaged in the hardware sector. During the two years between 1998 and 2000, the number of both establishments and workers for the hardware sector grew so that the sector captured a larger slice of the share of Korean industry. But it was with software that the most dynamic changes occurred

142

Agglomeration in Asia

Table 6.2

Number of establishments and workers by industry

ID*

Establishment 1998

(%)

2000

Workers (%)

1998

(%)

2000

(%)

3 2111 0.1 2066 0.1 23 962 0.2 21 406 0.2 4–5 5839 0.2 6434 0.2 173 320 1.6 180 827 1.5 6–8 17 170 0.7 20 929 0.8 391 019 3.5 437 573 3.6 9–10 3974 0.2 4596 0.2 76 900 0.7 82 907 0.7 11 3962 0.2 4841 0.2 71 474 0.6 87 455 0.7 12–14 7883 0.3 10 065 0.4 271 485 2.4 307 201 2.5 15–17 13 322 0.5 16 151 0.6 339 176 3.1 371 744 3.1 18 10 281 0.4 12 984 0.5 234 768 2.1 279 844 2.3 19 568 0.0 819 0.0 41 635 0.4 50 064 0.4 20 3852 0.2 5008 0.2 112 164 1.0 139 722 1.2 21 2851 0.1 3961 0.1 218 995 2.0 279 642 2.3 22 1709 0.1 2229 0.1 39 974 0.4 46 486 0.4 23 2589 0.1 3200 0.1 184 449 1.7 203 952 1.7 24 938 0.0 1044 0.0 89 520 0.8 94 392 0.8 25–26 4607 0.2 5849 0.2 79 023 0.7 90 781 0.7 27 1318 0.1 1420 0.1 52 713 0.5 56 629 0.5 28 63 186 2.5 66 621 2.5 711 225 6.4 640 755 5.3 29–31 963 162 37.5 916 685 34.0 2 433 235 21.9 2 493 217 20.5 32 578 175 22.5 607 718 22.5 1 335 601 12.0 1 555 985 12.8 33 218 395 8.5 273 283 10.1 811 297 7.3 896 131 7.4 34–36 127 352 5.0 130 447 4.8 969 562 8.7 943 466 7.8 37 4428 0.2 8145 0.3 57 531 0.5 124 984 1.0 38 1095 0.0 1604 0.1 45 631 0.4 55 188 0.5 39 52 551 2.0 59 992 2.2 419 820 3.8 438 835 3.6 40 94 935 3.7 102 802 3.8 779 622 7.0 921 158 7.6 41 53 979 2.1 65 944 2.4 395 855 3.6 487 902 4.0 42 99 121 3.9 120 517 4.5 248 769 2.2 318 409 2.6 43 1752 0.1 2360 0.1 22 090 0.2 28 340 0.2 44–45 225 620 8.8 240 359 8.9 456 027 4.1 503 515 4.1 0 2 566 725 100.0** 2 698 073 100.0** 11 086 842 100.0** 12 138 510 100.0 Notes: * For details of ID numbers, see Table 6A.1. ** Rounded. Source: National Statistical Oﬃce, Korea.

in this period. The number of establishments increased from 4428 to 8145 and the number of workers more than doubled from 57 531 to 124 984. As a result, the number of workers employed in the software sector has reached more than double the number for the hardware sector. The research and development (R&D) and education sectors, which are often seen as sectors with close relationships to developments in the IT

143

Industrial agglomeration and regional growth in Korea

Table 6.3 Number of workers in IT, R&D and education (percent share of total by administrative unit) Hard (No. 19) Soft (No. 37)

Seoul Busan Daegu Incheon Gwangju Daejeon Ulsan Gyeonggi Gangwon Chungbuk Chungnam Jeonbuk Jeonnam Gyeongbuk Gyeongnam Jeju

R&D (No. 38) Education (No. 40)

1998 2000

1998

2000

1998 2000

1998

2000

8.0 0.3 0.5 5.0 0.1 0.3 0.1 56.2 1.5 4.1 2.0 0.9 0.0 18.9 2.1 0.0

80.7 3.3 2.2 1.2 1.2 1.9 0.2 5.3 0.3 0.3 0.1 0.4 0.6 1.1 1.0 0.1

84.1 2.8 1.3 1.0 1.4 1.8 0.3 4.5 0.3 0.3 0.1 0.4 0.4 0.6 0.6 0.1

17.1 1.1 0.4 3.1 0.7 26.3 0.1 41.0 0.7 1.1 1.4 1.1 0.6 2.5 2.3 0.5

20.8 7.9 4.9 4.5 3.3 3.4 2.1 16.4 4.0 3.6 4.6 5.0 5.0 6.5 6.7 1.3

21.2 7.8 4.9 4.5 3.4 3.3 2.2 18.3 3.6 3.5 4.3 4.6 4.5 6.0 6.5 1.2

10.9 0.3 0.2 4.7 0.0 0.4 0.0 57.1 0.0 2.0 1.4 1.1 0.0 19.2 2.4 0.0

26.1 1.4 1.0 4.3 0.6 22.4 0.2 35.2 0.8 0.6 1.3 0.8 0.5 2.4 2.0 0.5

Source: National Statistical Oﬃce, Korea.

sector, also increased in number and in their share of establishments and workers. The uneven geographical distribution of locations of establishment and workers observed from the data of 43 industries can be seen again with regard to the four knowledge-based industries in Table 6.3. The most intensive concentration is seen in software. Seoul captured more than 80 per cent of the software workers in 1998 and this was increased to 84 per cent in 2000. The hardware sector forms industrial agglomerations in Gyeonggi with more than 55 per cent of the total workers and Gyeongbuk with about 20 per cent. Except for the R&D sector, every administrative unit with 10 per cent or more workers increased its share. In the R&D sector, Seoul deprived Gyeonggi and Daejeon of their shares. On the whole, concentration by location of these sectors to Seoul has progressed during the period. The geographical concentration of software companies within Seoul can also be observed from the data of the Korea Software Industry Association (KOSA). Among 303 KOSA members, 53.2 per cent of the software companies were located in local research complexes in Kangnam district in the

144

Agglomeration in Asia

city of Seoul in 1997 when 87.4 per cent of the software companies were located in Seoul. In fact, over 1500 IT companies concentrated around so-called ‘Teheran Valley’, which is a Korean Silicon Valley located in the southern area of Seoul along Teheran Street. In addition, Teheran Valley is a representative venture town. About 900 venture companies of 7000 venture companies nationwide and of 2900 venture companies in Seoul located around the valley (MIC 2002b).

3. GROWTH FACTORS OF KNOWLEDGE-BASED INDUSTRIES Why has the software sector grown so rapidly? Why are there agglomerations of, in particular, the IT, R&D and education sectors? These questions are the main concerns of this chapter. Eﬃcient creation, acquisition, transmission and the use of information and knowledge may bear a close relationship with these phenomena. In this section, we shall analyse the factors that promote the growth of the software sector and industrial agglomeration. Knowledge-based Economy The IT industry is often categorized as a knowledge-based industry, although there is no precise deﬁnition. Dahlman and Andersson’s (2000 p. 32) deﬁnition is helpful in understanding its image, that is, [A knowledge-based economy (KBE) is] one where knowledge (codiﬁed and tacit) is created, acquired, transmitted and used more eﬀectively by enterprises, organ-izations, individuals and communities for greater economic and social development. The KBE calls for: an economic and institutional regime that provides incentives for the eﬃcient use of existing knowledge, for the creation of new knowledge, for the dismantling of obsolete activities and for the start-up of more eﬃcient new ones; an educated and entrepreneurial population that can both create and use new knowledge; a dynamic information infrastructure that can facilitate eﬀective communication, dissemination and processing of information; an eﬃcient innovation system comprising ﬁrm, science and research centers, universities, think tanks, consultants and other organizations that can interact and tap into the growing stock of global knowledge, assimilate and adapt it to local needs, and use it to create new knowledge.

In summary, the following are the key promotion factors of IT industries: industrial policies; education and entrepreneurs; an information infrastructure; and an innovation system. Market structures, which have close relations with local needs, will also aﬀect the formation of industrial

Industrial agglomeration and regional growth in Korea

145

clusters. Among these factors, the improvement of environments for the innovation and new business will be the main policy issue for the promotion of IT application industries. R&D Policy History of R&D strategy Under the leadership of its government, Korea has undergone a remarkable industrialization process in attempting to catch up with Japan. But the introduction of its R&D strategy has lagged behind Japan and the government’s initiatives have been limited. The government started the process in 1962, undertaking a series of ﬁve-year economic development plans in which speciﬁc industries were given priority. In the 1980s, various tools were introduced to promote private R&D by the government through an incentive program for the private sector to set up R&D laboratories. But R&D was promoted through private sector monopoly rents yielded from domestic market protection, which ﬁnanced R&D conducted by Korean conglomerates known as chaebol. With the exception of the establishment of industrial technology research consortia supported by the government for speciﬁc R&D projects, which concentrated on electronics and machinery industries, the role of the government in R&D promotion has been limited. Although the government’s R&D policy mainly took indirect forms, the government also established national research laboratories such as the Korean Institute of Science and Technology (KIST) in 1966 to support the industry’s technological learning. It also funded university R&D, including the establishment of the Korean Advanced Institute of Science (KAIS) in 1971 (Sakakibara and Dong-Sung 1999). Today, the Korean Advanced Institute of Science and Technology (KAIST), which has its origin in KIST and KAIS, has become one of the centers of excellence in the cutting-edge of technologies, including IT. Daedeok Science Town: R&D center Daedeok Science Town is located in Daejeon Metropolitan City, an industrial city in the center of the country 150 km south of Seoul and 280 km north of Busan. In accordance with the government’s decentralization policy in the 1990s, some governmental organizations such as the Small and Medium Business Administration (SMBA), the Intellectual Property Rights Oﬃce and the Public Procurement Service have relocated to Daejeon. Today, the city is not only one of the most advanced research areas but also the second administrative city of Korea. The national project to build Daejeon was started in 1973 and construction was completed in 1992. As of the end of 2001, there were 116 institutes,

146

Agglomeration in Asia

Table 6.4 Number of institutions and human resources in Daedeok Science Town (December 31, 2001)

Corporate R&D Government corporations Government R&D Highest education Public institutions Supporting institutions Venture company Total

Institutions

(%)

Human resources

(%)

27 18 10 4 9 4 44

23.3 15.5 8.6 3.4 7.8 3.4 37.9

3297 6473 2452 2319 422 37 899

20.7 40.7 15.4 14.6 2.7 0.2 5.7

116

100.0*

15 899

100.0

Note: * Rounded. Source: Daedeok Science Town.

of which 12 were related to the information industry, and 15 899 researchers, of which 4455 were PhD holders, in the 27.8 km2 site. Demographic data in Table 6.4 show that although the private sector presence is largest, human resources are dominated by the governmental sector. Historically speaking, there were only three chaebol R&D institutions in the 1980s, with other private institutions being established in the 1990s (Yoon 2001). Among educational institutions, KAIST, which relocated to the park in 1989, is one of the most renowned. Some 300 PhDs, 600 masters and 500 bachelor-degree students graduate annually. Taking advantage of ‘Daedeok Valley’s’ location and solid scientiﬁc infrastructure as leverage, the Daejeon government instigated policies to promote venture business, aiming to become a leading city in cutting-edge technology in the twenty-ﬁrst century as well as a hub for the logistic distribution industry. The policy includes expansion of facilities for venture companies (for example, the operation of a start-up business incubating center), expansion of investment capital for venture business (for example, the expansion of the Daedeok Venture Investment Society, activation of the Daedeok Angel Mart, and the introduction of venture capital to the city), supporting solid marketing, and construction of a system to foster specialized personnel. Research institutes and universities are closely involved in this policy, especially in the ﬁeld of human resource training and developing companies. To foster management capabilities, KAIST established the Korean Graduate School of Management (KGSM) in 1996. Some institutions

Industrial agglomeration and regional growth in Korea

147

established incubating centers involving the participation of about 350 start-ups. In spite of this favorable environment for venture business, Daejeon’s share of the country’s venture business remains below 10 per cent. By referring to the data before 1998, Yoon (2001) explained Daedeok Science Town’s situation as follows: on the one hand greater priority was given to the goal of national technological policy than the needs of the private sector especially small and medium-sized enterprises (SMEs); on the other, not enough was done to help the local manufacturing base to turn R&D results into businesses. He concluded that the science town is not an agglomeration of cutting-edge industries but a center of technological knowledge with the potential to become an advanced industrial region. In practice, during the dot-com boom at the end of the twentieth century, most dot-com companies located in Seoul. This implies that Yoon’s analysis, the results of which seem to apply largely to the manufacturing sector, cannot completely explain the background for the development of start-ups. Business Promotion Policy Incubation business Like the IT business promotion policies of many other countries, Korea is augmenting centers to incubate businesses to provide a better environment for start-ups. The foundations for the incubation of businesses were laid in the 1990s but it was after the 1997 ﬁnancial crisis that it got into full swing. Two governmental organizations are implementing these policies: the SMBA, which supports centers for business incubation, and the Ministry of Information and Communication (MIC). MIC’s incubation eﬀorts focus on the IT sector, while the SMBA’s is more broadly targeted, supporting software and on-campus incubation. The ministry is establishing incubation businesses in large cities throughout the nation. This policy is promoted through cooperation between MIC and the Korean Industry Promotion Agency (KIPA). In addition to provisions to support policy making by MIC concerning start-ups, human resource management, and the digital content business, KIPA founded i-Park in the US (Silicon Valley) in 1998, China (Beijing) in 2000 and Japan in 2001, to develop foreign IT markets (Sakata et al. 2001). Thanks to this support, the number of incubations has increased from 30 in 1998 to 279 in October 2001, 232 of which are operated by universities, 17 by research institutes and 11 by local governments.

148

Agglomeration in Asia

Venture capital The venture capital industry was developed from policy initiatives in the early 1980s to carry out the government policy of providing funds for, or mainly loans to, SMEs. The launch of KOSDAQ (Korean Securities Dealers Automated Quotation) in July 1996 opened up opportunities for investments. As of the end of 2001, there were 145 venture capital ﬁrms established between 1999 and 2000. Although the industry has experienced a steep decline recently, the IT industry has nevertheless enjoyed the beneﬁts and received two-thirds of total investments in recent years (MIC 2002a). According to SMBA, 28.7 billion won of venture capital funds were invested in 1973 companies in 2000. Some 59.7 per cent of the companies were less than one year old, and 34.9 per cent of funds were invested in them. Two- and three-year-old companies (21.8 per cent) received 29.8 per cent of the funds, most of which were invested in the IT sector. As of June 2000, 32.8 per cent of the funds were invested in electric and electronic businesses and 32.5 per cent in engineering and information businesses. As a result, the number of venture companies listed on KOSDAQ increased from 52 at the end of 1996 to 353 at the end of 2001, 30 per cent of which were IT ﬁrms. These facts imply that there was a change in ﬁnancial sector behavior and in people’s future expectations. In addition, as is often seen in Korea’s industrial policy, the government took strong initiatives to encourage the venture industry by setting up new investment funds themselves and reducing income and capital gains tax (ibid.).

4. REGRESSION ANALYSIS ON GROWTH IN REGIONS Mechanism of Agglomeration Formation The improvements in business and innovative activities may be only prerequisites for the growth of industries. They may not fully explain the dynamic aspects such as industrial growth and agglomeration. In order to analyse these phenomena, the past econometric analysis considered the following aspects as factors aﬀecting regional growth: local competition; diversity; and regional specialization (Glaeser et al. 1992). Bitter local competition caused by the agglomeration is more likely to force companies to seek technological innovations and introduce new technologies. Correspondingly, more diversiﬁed agglomeration is more likely to allow the knowledge spillover between diﬀerent industries. Regional specialization has a trade-oﬀ

Industrial agglomeration and regional growth in Korea

149

aspect. That is to say, regional agglomerations that are more specialized in a speciﬁc industry have more advantages in knowledge spillover within the specialized industry, and disadvantages between diﬀerent industries. Fujita’s study (2001) of new spatial economics also identiﬁed the indigenous mechanism of agglomeration. He identiﬁed three source of agglomeration: diversity or heterogeneity; increasing return; and transportation costs. Diversity is composed of three types: diversity of (i) consumption goods; (ii) intermediary goods for production; and (iii) human resources. Interaction among these three constitutes agglomeration dynamics. Diversity of consumption goods increases real income because it can provide consumers with more opportunities to satisfy their needs. This attracts more people to cities. The resulting market growth enables companies to specialize in speciﬁc ﬁelds of goods and services. In the case of diversity of intermediary goods, complementary relations throughout a supply chain make agglomeration of an industry more productive. The resultant expansion of the market of intermediary goods also increases demand for specialized services. The increasing return, which can be derived from diversity, drives the formation of agglomeration. Diversity of human resource means that each individual has diﬀerent knowledge and information. Knowledge and information are public goods, which mean that their consumption does not decrease their stock. This characteristic brings about eﬀects of increasing return of agglomeration to create knowledge. Knowledge Creation, Agglomeration and IT Interaction between people can be done through face-to-face communications or telecommunications. This implies that diversity of human resources and transportation costs have a close relation in the formation of agglomeration. Transportation costs, which can be interpreted broadly by Fujita (2001), include costs for transfer of people, information, goods and services, and whatever can move spatially. One noteworthy result of spatial economics is the non-linear eﬀect of transportation costs on agglomeration. Without transportation costs, agglomeration would never occur. If face-to-face interactions were indispensable, agglomeration of diversiﬁed people would occur to promote knowledge creation. Ongoing innovations in IT are decreasing transportation costs. This will result in a change in current agglomeration. One of the interesting issues in the knowledge-based economy may be whether IT will substitute for or complement face-to-face communications. Imagawa (2002) indicated that local calls in Japan were made more frequently but the duration of such calls was shorter than long-distance calls despite the fact that

150

Agglomeration in Asia

long-distance calls were more expensive than local calls. This seems to imply that local calls complement and long-distance calls substitute for face-to-face communication. Fujita (2001) insisted that face-to-face communications and IT are complementary, that face-to-face interactions are indispensable as part of the process of technological innovation, and that leading industries such as games and software concentrate in urban areas as a result of the necessity for face-to-face interactions. On the other hand, he also evaluated that the US Air Force, Stanford University and Hewlett-Packard (HP) performed a catalytic role for the so-called ‘big push’ in the initial stages to accelerate high-tech agglomeration in Silicon Valley. The Basic Model of Growth in Regions The regression analysis conducted in this chapter was based on the model developed by Glaeser et al. (1992). The traditional theories have viewed externalities associated with knowledge spillovers as the engine of growth. Glaeser et al. supposed geographical proximity as a factor to facilitate transmissions of ideas and growth in cities. Their paper focused on technological externalities, where innovations and improvements occurring in one ﬁrm increase the productivities of other ﬁrms without full compensation. The model Glaeser et al. developed a simple economic model for expressing the growth in a city, taking knowledge spillover eﬀects into account. The model suppose a ﬁrm in some industry in a given location, which has a production function of output, Atf(lt), where At represents the overall level of technology at time t measured nominally (so changes in A represent changes in technology and changes in price), and lt is the labor input at time t. As the form of the basic production function f(lt), decreasing return is assumed as usual (f 0, f0). Given technologies, prices and wages (wt), the ﬁrm maximizes its proﬁt: At f(lt ) wtlt.

(6.1)

The following can be derived from the ﬁrst order condition (FOC): At f (lt ) wt, log

At1 wt1 f (lt1 ) At log wt log f (lt ) .

(6.2)

Industrial agglomeration and regional growth in Korea

151

The level of technology A, in a city industry is assumed to have both national and local components: At Alocal Anational, which can be transformed as the sum of the growth of national technology in this industry and the growth of local technology as below: log

Alocal,t1 Anational,t1 At1 At log Alocal,t log Anational,t .

(6.3)

The growth of the national technology is assumed to capture the changes in the price of the product as well as shifts in nationwide technology in the industry, and the local technology is assumed to grow at a rate exogenous to the ﬁrm but depending on the various technological externalities present in this industry in the city: log

Alocal,t1 Alocal,t g(specialty, local monopoly, diversity, initial conditions) et1.

(6.4)

If f(lt ) l1 , 0 1 is assumed, the following equation can be t obtained:

log

Anational,t1 f(lt1 ) w log wt1 log A g(specialty, f (lt ) t national,t local monopoly, diversity, initial conditions) et1. (6.5)

Equation (6.5) means that growth in nationwide industry employment is assumed to capture changes in nationwide technology and prices. It is assumed that the wage growth will be constant across city industry. The variables for technological externalities will be explained below. Knowledge-spillover factors and the variables As the externalities to promote the cities’ growth and as factors that have eﬀects on the ‘growth rates’ of industries in diﬀerent cities as shown in equation (6.5), the following three variables were considered and deﬁned to be applied to the statistical analyses implemented in this chapter: diversity of industries in one city; local competition in one city among ﬁrms in one industry; and specialty, or concentration of an industry in a city.

152

Agglomeration in Asia

Diversity If important knowledge for the development of one industry transfers from outside of its industry, variety and diversity of geographically proximate industries rather than geographical specialization promote innovation and growth. This hypothesis is diﬀerent from the local withinindustry externality theory, which supposes that the concentration of an industry in a city helps knowledge spillovers between ﬁrms within the industry, and therefore the growth of that industry and that city. The measure of a (non-)diversity of industries in the city outside the industry in question is deﬁned as the ratio of employees in the ﬁve largest industries other than the industry in question to the total employments. The lower this ratio, the more diverse the city is.

Non-diversity

no. of employees of the top ﬁve industries other than the industry in question . total employment in the city

Non-diversity is ‘the fraction of the number of employees in the top ﬁve industries other than the industry in question accounted for in the initial year’. Competition Local competition fosters the pursuit and rapid adoption of innovation. On the other hand, there is an assumption that local monopoly restricts the ﬂow of ideas to others and so allows externalities to be internalized by the innovator. This internalization promotes innovation and growth. The measure of local competition of an industry in a city is deﬁned as the number of ﬁrms per worker in this industry in this city relative to the number of ﬁrms per worker in this industry in the analysed country (for example: the US, Korea): competition

firms in the city industryworkers in the city industry . firms in the country industryworkers in the country industry

Specialty Firms in an industry will specialize geographically in order to absorb the knowledge spilling over between ﬁrms in that industry. Regionally specialized industries should grow faster because neighboring ﬁrms can learn from each other much better than geographically isolated ﬁrms. This prediction is contrary to the hypothesis that industries located in a highly industrially diversiﬁed area grow faster. The measure of specialization of an industry in a city is deﬁned as the fraction of the city’s employment that this industry represents in that city, relative to the share of the whole industry in national employment:

Industrial agglomeration and regional growth in Korea

153

industry employment in the citytotal employment in the city specialty industry employment in the countrytotal employment in the country.

Results of Previous Studies Glaeser et al. (1992) The data The dataset included 1016 observations of the top six-digit industries in 1956. The data source was the 1956 and 1987 editions of Country Business Patterns (CBP), produced by the Bureau of the Census. The result Local competition and diversity encourage employment growth in industries. The evidence suggests that important knowledge spillovers might occur between rather than within industries. The initial wages in a city-industry growth are uncorrelated with subsequent employment growth. Ahn et al. (2000) The model A ﬁrm in an industry (i), located in a region (j), at time (t) chooses the amount of labor input (L) to maximize its proﬁt: AijtL1 ijt wijtLijt. The main diﬀerence between Glaeser et al. and Ahn et al. is that, in the latter, it is assumed that the growth of regional technology depends on IT improvement and on various externalities as follows:

Aijt1 Li,t1 Aijt Lit

Ijt1 Ijt

f(Sijt ).

In the equation, I is a measure of IT level, and S is a vector of measures of knowledge spillovers and some initial conditions. By introducing this assumption, this model speciﬁes IT improvements as a variable that explains growth of employment. The regional IT index I is deﬁned as follows: Ijt iaiLijt Pj. Here the weight ai is deﬁned by the ratio of the amount of IT service inputs to the total amount of required inputs of industry i, and is constructed based on the input–output analysis table. Pj is a size-normalizing factor. ITA denotes the index normalized by physical area, and ITP by population. The data The sample is extracted from the Population and Housing Census (1995) and the Census on Basic Characteristics of Establishments (1994 and 1998). The dataset is composed of 167 regions corresponding to

154

Agglomeration in Asia

Table 6.5 Results of Ahn et al. (2000) on IT and spillover eﬀects on growth in cities (1)

Competition Specialization Non-diversity ITA ITP

(2)

(3)

sign

signiﬁcance

sign

signiﬁcance

sign

signiﬁcance

–

*

–

*

*

*

*

Note: * indicates the coeﬃcient is signiﬁcantly diﬀerent from zero.

the administrative system in 1998 and the 20 largest industries in each region. The sample size is 3340. The result As arranged in Table 6.5, the IT index normalized by physical area is not statistically signiﬁcant. The eﬀects of knowledge spillovers, except for competition are no longer signiﬁcant when the IT index is included in the model. Ahn et al. indicates two possible explanations for this result: 1.

2.

The initial value for the measures on knowledge spillovers do not explain the recent growth in industries well because of the signiﬁcant changes in IT in recent years. IT improvements would signiﬁcantly reduce the opportunity cost of collecting information and hence could lessen the need for a region’s specialization or diversiﬁcation.

Regression Analysis on Growth in Regions in Korea The basic model The model and method applied for the purpose of verifying the factors driving the growth in regions in Korea are very simple. The model to be estimated in this section is based on that of Glaeser et al. (1992) and is formulated as follows: log

li,j,2000 1log(national level change in industrial structure) i, j li,j,1998 2 (competition) i,j,1998 3 (specialty) i,j,1998 4 (non diversity) i,j,1998

Industrial agglomeration and regional growth in Korea

155

in which li,j,t is the number of employees in an industry (i), in a region (j), in year (t). The deﬁnitions of the knowledge spillover factors are the same as Glaeser et al. ‘National level change’, is added to correct for demand shifts and deﬁned as follows: employment in an industry except the region in 2000 (National level change) employment in an industry except the region in 1998 .

Although the variable of wage growth between 1998 and 2000 is to be included in accordance with equation (6.5), it was excluded because its coeﬃcient does not diﬀer from zero signiﬁcantly at the 5 per cent level as shown by the results of the regression analyses implemented below. Some explanatory variables will be added to this basic model in accordance with the purpose of the analysis and statistical veriﬁcations. The data The data The data used for the analyses conducted were derived from the ‘Report on Mining and Manufacturing Survey’ and other statistics in 1998 and 2000 edited by the National Statistical Oﬃce. The dataset is composed of the 43 industries in 16 administrative units. The total sample number is 659 after removing missing data. The sample number is 618 when tele-density data are applied to the analyses (tele-density data in Ulsan are not available). Description of the original data In addition to employment growth, the indices for competition, specialty and non-diversity can be calculated from the data mentioned above in accordance with Glaeser et al.’s deﬁnitions. Table 6.6 describes the ﬁve fastest- and the ﬁve slowest-growing region industries in terms of employment. Note ﬁrst that the fastest-growing industries are the service sector such as IT service and R&D, while the slowestgrowing industry is the machinery sector, especially the manufacturing of other transportation equipment. Second, smaller region industries grew faster than larger ones. Third, the faster-growing region industries were more competitive than the shrinking ones. Fourth, the service sector grew faster in specialized regions. Results of the regression analysis The descriptive statistics and correlation matrix on the explanatory variables to be included in the basic model described above are shown in Tables 6.7 and 6.8. The number of employees and the wage level in the city industry (region industry) in the initial year were included in Glaeser et al. The correlation matrix indicates that the variables on change in

156

3.236 2.855 2.854 2.833 0.137 0.465 0.491 0.497 0.527

Five slowest-growing region industries Manufacture of Other Transport Equip. Manufacture of Other Transport Equip. Manufacture of Other Transport Equip. Computer & Oﬃce Machinery Coke, Reﬁned Petroleum Products

Seoul Jeonbuk Chungnam Daegu Incheon

Ulsan Jeju Daegu Jeju

Coke, Reﬁned Petroleum Products Computer and Related Activities Computer and Related Activities R&D Manufacture of Basic Metals

Chungbuk

in 1998

in 2000

No. workers

5.833

1113 114 1660 199 1364

127 55 199 6

6

152 53 815 99 719

411 157 568 17

35

Five fastest-growing region industries

00/98

Industry

Fastest- and slowest-growing region industries

City

Table 6.6

1.629 9.209 1.035 2.210 0.840

3.478 2.835 7.329 8.649

27.279

Competition in 1998

0.047 0.035 0.502 0.095 2.207

0.089 0.090 0.086 0.006

0.015

Specialty in 1998

0.537 0.532 0.488 0.491 0.419

0.499 0.641 0.491 0.641

0.471

Non-diversity in 1998

157

Industrial agglomeration and regional growth in Korea

Table 6.7

Descriptive statistics (1)

Log (l2000/l1998) Log (Change in Industrial Structure) Log (W2000/W1998) Log (W1998) Log (l1998) Non-diversity Competition Specialty 1/Specialty

Mean

Std dev

Minimum

Maximum

0.106 0.106 0.095 16.662 8.338 0.497 1.337 1.056 3.855

0.263 0.163 0.091 0.224 1.917 0.058 1.490 1.376 17.106

1.991 0.432 0.127 16.259 1.792 0.327 0.131 0.003 0.045

1.764 0.785 0.276 17.135 12.836 0.641 27.279 22.247 370.928

Note: Number of observations659.

Table 6.8

Correlation matrix (1) Log (l2000/l1998)

Log (l2000/l1998) Log (Change in industrial structure) Log (W2000/W1998) Log (W1998) Log (l1998) Non-diversity Competition Specialty 1/Specialty

Log (l1998) Non-diversity Competition Specialty 1/Specialty

1 0.501

Log (Change in industrial structure)

Log (W2000/W1998)

Log (W1998)

–

– –

– –

–

1

0.126 0.004 0.176 0.007 0.278 0.141 0.189

0.161 0.023 0.160 0.028 0.032 0.069 0.023

1 0.113 0.299 0.067 0.097 0.003 0.026

1 0.010 0.007 0.245 0.097 0.027

Log (l1998)

Non-diversity

Competition

Specialty

1 0.383 0.384 0.286 0.364

–

– –

– – –

Note: Number of observations659.

1 0.096 0.117 0.196

1 0.196 0.359

1 0.131

158

Agglomeration in Asia

industrial structure and competition have relatively high correlations with the explained variable. The result of the regression analysis applied ordinary least squares (OLS) to the basic model is coeﬃcient 1 in the ﬁrst column in Table 6.9. The number of employees and the wage level are added to the basic model in coeﬃcients 2 and 3. As shown in Figure 6.1, the distribution of the residuals derived from coeﬃcient 1 is heterogeneous. The absolute value of the residuals becomes smaller as the specialty index increases. In order to deal with this heterogeneity, the reciprocal of the specialty index is included in the basic and its derivative models. The results of OLS are coeﬃcients 4, 5 and 6 in Table 6.9. By adding the reciprocal of specialty, the t-values of explanatory variables, except for the competition index and wage level, improved. Implications of the results The growth rate in a region industry is higher when the demand for the industry outside the region is increasing. The coeﬃcient of the variable of change in industrial structure in the industry is signiﬁcantly diﬀerent from zero. The initial labor conditions do not inﬂuence the growth rate in regions signiﬁcantly. The negative coeﬃcients of wage level indicate the employment increase in the lower-wage regions. The signs of variables for knowledge-spillover eﬀects are the same as those of Glaeser et al. (1992), though the coeﬃcients of competition and non-diversity in terms of absolute value are much smaller than their estimates. The positive coeﬃcients of local competition and diversity, as well as the negative coeﬃcient of specialty, support the hypothesis that knowledge spillovers would be encouraged between rather than within industries. The eﬀects of tele-density on growth in regions It is possible to hypothesize the eﬀect of telecommunications infrastructure on the growth in regions. It will promote the agglomeration of industries in an area when local calls and data communications complement face-toface communications. It will also encourage decentralization of industrial locations if long-distance calls and data communications replace face-toface communication. The basic model indicated above takes into consideration only the knowledge spillover within an area. For this reason, it is not impossible to analyse the telecommunications eﬀects on the growth in regions in detail, if some variables on penetration of telecommunications are simply added to the model. However, if we take into consideration that tele-density is

159

Notes:

78.680 0.321 659

62.850 0.325 659

0.045* (7.220) 0.240*** ( 1.532) 0.012** ( 1.865)

0.098 (0.929) 0.787* (14.986) 0.000 ( 0.080)

Coeﬃcient 2

63.376 0.322 659

0.040* (6.674) 0.287* ( 1.956)

0.052 ( 1.335) 0.047* (7.815) 0.236*** ( 1.619) 0.011** ( 1.702)

80.892 0.327 659

0.002* (3.116)

0.104 (1.418) 0.792* (15.384)

Coeﬃcient 4

0.962 (1.467)*** 0.786* (15.166)

Coeﬃcient 3

t-statistic in parenthesis. * signiﬁcant at the 5%, ** 10% and *** 15% levels.

F-statistic Adjusted R-squared Number of observations

1/Specialty

Specialty

Non-diversity

Competition

log(W1998)

0.045* (7.737) 0.236*** ( 1.616) 0.012** ( 1.924)

0.092 (1.245) 0.787* (15.190)

Constant

Log (Change in industrial structure) log(l1998)

Coeﬃcient 1

Results of the regression analysis on growth in regions

Variable

Table 6.9

64.628 0.326 659

0.002* (3.084)

0.041* (6.424) 0.276** ( 1.759)

0.088 (0.837) 0.794* (15.197) 0.001 (0.209)

Coeﬃcient 5

65.070 0.327 659

0.002* (2.937)

0.048 ( 1.234) 0.042* (6.760) 0.286* ( 1.950)

0.903 (1.386) 0.790* (15.346)

Coeﬃcient 6

160 8 6 4 2 0 –2 –4 –6 –8 –10 –12

Agglomeration in Asia

5

10

15

20

25

Specialty

Figure 6.1 Standardized residual derived from coeﬃcient 1 in Table 6.9 Table 6.10

Tele-density and Internet penetration by regions (%)

Region

Tele-density (1998)

Internet (2000)

53.5 43.1 46.5 57.9 42 42.7 n.a. 35.3

51.9 43.5 38.5 46.2 42.8 46.7 52.4 50.8

Seoul Busan Daegu Incheon Gwangju Daejeon Ulsan Gyeonggi

Region Gangwon Chungbuk Chungnam Jeonbuk Jeonnam Gyeongbuk Gyeongnam Jeju

Tele-density (1998)

Internet (2000)

44.9 42.3 42.7 41.2 41.1 38.6 42.2 42.7

40.6 37.8 37.5 38.2 36.5 31.4 38.7 44.4

Sources: Regional Statistics Yearbook, Korea National Statistical Oﬃce; MIC, 2002 Internet White Paper, Korea.

lower in the smaller cities and rural areas, it is acceptable to suppose that ﬁrms will choose some lower-cost locations, and keep in touch with their colleagues and collect information via IT communications, when the coeﬃcients of such IT indicators become negative. The data on tele-density, as well as Internet penetration, are listed in Table 6.10. The descriptive statistics and correlation matrix on the explained and explanatory variables are shown in Tables 6.11 and 6.12. The correlation matrix indicates the negative but low correlation coeﬃcient between employment growth and tele-density. The results of OLS are listed in Table 6.13. The eﬀect of tele-density on growth in regions is negative and signiﬁcant only at the 15 per cent level. The signs of other variables and the

Industrial agglomeration and regional growth in Korea

Table 6.11

161

Descriptive statistics (2)

Log (l2000/l1998) Log (Change in industrial structure) Log (W2000/W1998) Log (W1998) Log (l1998) Non-diversity Competition Specialty 1/Specialty Tele-density1998 Log (No. of Employees in R&D Sector)1998 Log (No. of Employees in Education Sector)1998

Mean

Std dev.

Minimum

Maximum

0.102 0.106

0.259 0.163

1.991 0.432

1.764 0.785

0.095 16.662 8.386 0.497 1.364 1.008 3.814 43.783 6.848

0.091 0.224 1.908 0.060 1.528 0.970 17.481 5.416 1.414

0.127 16.259 1.792 0.327 0.196 0.003 0.084 35.300 5.293

0.276 17.135 12.836 0.641 27.279 11.866 370.928 57.900 9.837

10.636

0.627

9.213

11.998

Note: Number of observations618.

coeﬃcient of determination are the same as those of the models without the tele-density variable. It is possible to assume that ﬁrms tend to select lower-cost sites with good eﬀects of knowledge spillover between industries.

5.

CHARACTERISTICS OF THE IT SECTOR

The IT sector, one of the representative knowledge-based industries, is supposed to have diﬀerent characteristics from traditional sectors. This means that some factors diﬀerent from the other sectors can aﬀect growth in the IT sector. On the other hand, though the hardware and the software sectors can be categorized as one sector (that is, the IT sector), the two have diﬀerent features. This consideration may also necessitate the consideration of growth of each sector. This section will analyse these issues based on the model conducted above. Characteristics of the Software Sector Generally speaking, the advantage in promoting the software sector development is that the sector will increase employment opportunities in rural

162

Correlation matrix (2)

Log (l1998) Non-diversity Competition Specialty 1/Specialty Tele-density1998 Log (No. of Employees in R&D Sector)1998 Log (No. of Employees in Education Sector)1998

Log (l2000/l1998) Log (Change in industrial structure) Log (W2000/W1998) Log (W1998) Log (l1998) Non-diversity Competition Specialty 1/Specialty Tele-density1998 Log (No. of Employees in R&D Sector)1998 Log (No. of Employees in Education Sector)1998

Table 6.12

1 0.095 0.145 0.197 0.006 0.447 0.427

Non-diversity

Log (l1998) 1 0.391 0.400 0.360 0.359 0.040 0.271 0.408

1 0.158 0.027 0.153 0.028 0.023 0.093 0.019 0.014 0.019 0.013

1 0.491 0.134 0.008 0.164 0.008 0.279 0.133 0.184 0.045 0.025 0.034

Log (l2000/l1998) Log (Change in industrial structure)

1 0.238 0.361 0.002 0.084 0.065

Competition

1 0.114 0.296 0.067 0.099 0.012 0.020 0.000 0.010 0.018

Log (W2000/W1998)

1 0.168 0.002 0.049 0.033

Specialty

1 0.008 0.008 0.254 0.027 0.016 0.007 0.003 0.000

Log (W1998)

163

Note:

Number of observations618.

1/Specialty Tele-density1998 Log (No. of Employees in R&D Sector)1998 Log (No. of Employees in Education Sector)1998

1 0.004 0.084 0.139

1/Specialty

1 0.007 0.046

Tele-density1998

1 0.612

1

Log (No. of Log (No. of Employees in R&D Employees in Sector)1998 Education Sector)1998

164

Agglomeration in Asia

Table 6.13

Eﬀects of tele-density on growth in regions

Variable

Coeﬃcient 7

Coeﬃcient 8

Coeﬃcient 9

Coeﬃcient 10

Constant

0.186** (1.820) 0.768* (14.441)

1.126** (1.671) 0.766* (14.413) 0.057 ( 1.411) 0.047* (7.813) 0.220*** ( 1.510) 0.007 ( 0.796)

0.204* (2.032) 0.771* (14.643)

0.040* (6.673) 0.274*** ( 1.877)

0.982*** (1.460) 0.769* (14.602) 0.047 ( 1.170) 0.042* (6.737) 0.273** ( 1.872)

0.002* (2.816) 0.002*** ( 1.566)

0.001* (2.658) 0.003*** ( 1.575)

Log(Change in industrial structure) log(W1998) Competition Non-diversity Specialty

0.045* (7.711) 0.219*** ( 1.502) 0.009 ( 0.925)

1/Specialty Tele-density in 1998 F-statistic Adjusted R-squared Number of observations Notes:

0.003*** ( 1.570) 57.271 0.313 618

0.003*** ( 1.582) 48.135 0.314 618

59.345 0.321 618

49.712 0.321 618

t-statistic in parentheses. * signiﬁcant at the 5%, ** 10% and *** 15% levels.

areas. This expectation is based on the assumption that telecommunications will replace face-to-face communications. On the other hand, the formation of the industrial cluster is one of the main strategies for developing a software sector in a region. From the perspective of the analysis in this chapter, these policies should be based on the hypothesis that the coeﬃcient for tele-density will be negative, and for specialty positive. The results of OLS implemented above showed that the specialty index has a negative impact on growth in region industries. This is not always a desirable result to promote the cluster development policy. On the other hand, the OLS results strongly support the positive impact of competition on growth in the region industries, which is consistent with the present conditions in the software sector in Korea. As shown in Table 6.14, the competition index for the software sector is relatively high in comparison with the hardware sector, especially in the smaller regions. The software sector in Seoul, where the largest software cluster in Korea is located, has diﬀerent characteristics from other regions. The competition index in Seoul is much lower than other regions, while the specialty index is much higher.

165

Industrial agglomeration and regional growth in Korea

Table 6.14

Growth of hardware and software industries

Growth City rank

No. of workers 00/98 in 1998 in 2000

Competition Specialty Non-diversity in 1998 in1998 in 1998

Computer & Oﬃce Machinery 37 401 656 266 40 191 652 602 52 186 112

11 21 116 27 9 14 2 19 8 10 28 22 60 179 96 3

Seoul Busan Daegu Incheon Gwangju Daejeon Ulsan Gyeonggi Gangwon Chungbuk Chungnam Jeonbuk Jeonnam Gyeongbuk Gyeongnam Jeju

1.627 3349 5448 1.066 121 129 0.497 199 99 1.144 2075 2374 n.a. 37 n.a. 1.593 118 188 n.a. 29 n.a. 1.222 23 387 28 581 n.a. 616 n.a. 0.600 1703 1021 0.846 853 722 1.482 388 575 n.a. n.a. n.a. 1.224 7867 9626 1.320 893 1179 n.a. n.a. n.a.

2.561 7.875 2.210 1.872 3.962 2.485 2.528 0.796 0.952 0.947 0.687 1.134 n.a. 0.540 1.313 n.a.

Seoul Busan Daegu Incheon Gwangju Daejeon Ulsan Gyeonggi Gangwon Chungbuk Chungnam Jeonbuk Jeonnam Gyeongbuk Gyeongnam Jeju

Computer and Related Activities 2.262 46 440 105 054 0.713 1.806 1925 3477 2.254 1.317 1258 1657 2.417 1.714 696 1193 2.427 2.478 692 1715 2.215 2.039 1076 2194 1.980 3.236 127 411 3.478 1.868 3026 5653 1.580 2.515 165 415 3.543 2.280 164 374 3.803 1.711 83 142 4.853 1.802 253 456 3.903 1.448 366 530 1.881 1.231 642 790 1.781 1.361 563 766 3.323 2.855 55 157 2.835

0.305 0.035 0.095 1.016 0.032 0.102 0.028 3.345 0.481 1.326 0.555 0.258 n.a. 3.313 0.323 n.a.

0.537 0.522 0.491 0.419 0.511 0.506 0.499 0.408 0.583 0.471 0.488 0.532 0.548 0.466 0.439 0.641

3.061 0.406 0.433 0.247 0.435 0.673 0.089 0.313 0.093 0.092 0.039 0.122 0.170 0.196 0.147 0.090

0.537 0.522 0.491 0.419 0.511 0.506 0.499 0.408 0.583 0.471 0.488 0.532 0.548 0.466 0.439 0.641

In order to test the characteristics of the software sector, a dummy variable for the software sector was introduced. The variable equals 1 in the case of the software sector and otherwise 0. The result is indicated in Table 6.15. After separating software’s speciﬁc eﬀects from the rest of the sectors, the eﬀects of negative non-diversity and specialty are retained. On

166

Soft Dummy*1/Specialty

Soft Dummy*Non-diversity

Tele-density in 1998

Specialty

1/Specialty

Non-diversity

Competition

log(W1998)

log(W2000/W1998)

0.002*** ( 1.451) 0.877 (1.266) ( 0.011) ( 1.132)

0.041* (6.766) 0.294* ( 2.001) 0.002* (2.825)

0.203* (2.016) 0.805* (11.804)

0.011 ( 1.193) 0.003*** ( 1.565) 0.985*** (1.460)

0.047** (7.926) 0.258** ( 1.769)

0.202* (1.982) 0.820* (11.954)

Coeﬃcient 11 Coeﬃcient 12

Characteristics of the software sector

Log(Change in industrial structure)

Constant

Variable

Table 6.15

0.002*** (1.460) 0.891 (1.287) 0.011 ( 1.149)

0.044 ( 1.078) 0.043* (6.816) 0.293* ( 1.997)* 0.001 (2.685)

0.929 (1.363) 0.796* (11.582)

Coeﬃcient 13

0.010 ( 1.072) 0.003*** ( 1.575) 0.999*** (1.481)

0.052 ( 1.275) 0.049* (8.000) 0.259** ( 1.776)

1.059*** (1.558) 0.810* (11.717)

Coeﬃcient 14

0.002*** ( 1.447) 0.871 (1.258) 0.010 ( 1.128)

0.041* (6.658) 0.304* ( 2.070) 0.002* (2.854)

0.198* (1.967) 0.802 (11.750) 0.118 (1.215)

Coeﬃcient 15

0.011 ( 1.233) 0.003*** ( 1.562) 0.980*** (1.453)

0.046* (7.813) 0.269** ( 1.839)

0.198** (1.939) 0.817* (11.899) 0.116 (1.195)

Coeﬃcien 16

167

Notes:

0.321 37.462 618

0.010 ( 1.284) 0.320 37.278 618

0.222* (2.694) 0.015** ( 1.920) 33.438 0.321 618

0.010 ( 1.265)

t-statistic in parentheses * signiﬁcant at the 5%, ** 10% and *** 15% levels.

F-statistic Adjusted R-squared Number of observations

Soft Dummy*Tele-density

Soft Dummy*Specialty

33.351 0.321 618

0.221* (2.687) 0.015** ( 1.897)

33.490 0.322 618

0.010 ( 1.310)

33.318 0.320 618

0.223* (2.701) 0.016* ( 1.947)

168

Agglomeration in Asia

the other hand, considering the coeﬃcients with a dummy variable, the signs of the specialty coeﬃcient changed in the software sector, and the negative sign of tele-density became larger than other sectors. Considering its character of a less capital- and material-intensive input structure, the development policy of specialization in the software sector seems to be justiﬁed in periphery economies. Eﬀects of Educational and R&D Institutions on the Growth of the IT Sector It seems to be common sense that educational and R&D institutions will play an important role in the development of the IT sector. Such institutions are assumed to play key roles as an origin of knowledge spillover in the framework of cluster development policy. In order to conﬁrm the relation between educational and R&D institutions and the growth of regions, and the software and hardware sectors, their number in 1998 as initial conditions and dummy variables will be incorporated into the model as explanatory variables. The results of regression analyses are indicated in Tables 6.16 and 6.17. Table 6.16 shows, in the case of the software sector, that the signiﬁcant positive coeﬃcients of R&D and R&D multiplied by the software dummy are not derived. The coeﬃcients of education and education multiplied by the software dummy were derived to be negative, which is contrary to common belief that educational institutions will have positive eﬀects on industrial growth. Table 6.17 shows the results of the analyses applying the same variables as in Table 6.16 to the hardware sector. In contrast to the software sector results, signiﬁcant positive coeﬃcients of R&D multiplied by the hardware dummy were observed.

6.

CONCLUDING REMARKS

The data observed in this chapter indicated agglomerations of some knowledge-intensive sectors such as IT, R&D and education. Some policies implemented by the Korean government have contributed to the growth of the IT and R&D sectors, while the spillover eﬀects of the R&D sector growth on the IT sector have not been veriﬁed. In addition, innovation in IT is expected to be a condition for developing the IT sector. In order to identify the factors to promote growth in regions, and development of the IT sector, some regression analyses have been implemented. Although the series of analyses conducted above are very preliminary, the following factors can be identiﬁed. First, local competition had a

169

Industrial agglomeration and regional growth in Korea

Table 6.16 Eﬀects of educational and R&D institutions on growth in the software sector Variable

Coeﬃcient 17 Coeﬃcient 18 Coeﬃcient 19

Constant Log (Change in industrial structure) Competition Non-diversity Specialty Tele-density in 1998 Soft Dummy*Non-diversity Soft Dummy*Specialty Soft Dummy*Tele-density Log (No. of R&D Employees in 1998) Soft Dummy*Log (No. of R&D Employees in 1998) Log (No. of Education Employees in 1998) Soft Dummy*Log (No. of Education Employees in 1998) F-statistic Adjusted R-squared Number of observations Notes:

0.126 (0.956) 0.825* (12.005) 0.047* (7.962) 0.194 ( 1.189) 0.011 ( 1.172) 0.003** ( 1.650) 1.054*** (1.551) 0.236* (2.760) 0.011 ( 1.197) 0.007 (1.054) 0.036 ( 1.011)

30.013 0.320 618

0.442* (2.019) 0.832* (12.143) 0.046* (7.928) 0.367* ( 2.275) 0.012 ( 1.308) 0.003** ( 1.737) 1.852* (2.428) 0.235* (2.849) 0.000 (0.026)

0.016 ( 1.066) 0.108* ( 2.366)

0.465* (2.118) 0.832* (12.180) 0.047* (8.022) 0.289** ( 1.739) 0.012 ( 1.276) 0.003** ( 1.693) 2.207* (2.687) 0.190* (2.140) 0.002 (0.206) 0.013** (1.676) 0.066 (1.198) 0.031** ( 1.741) 0.173* ( 2.437)

30.791 0.326 618

26.196 0.329 618

t-statistic in parentheses. * signiﬁcant at the 5%, ** 10% and *** 15% levels.

consistently positive impact on growth in regions. Second, high specialization in the software sector helped high growth in the short period, which was unique compared to the system of overall regional growth. Third, the possibility was not denied that location in the software sector may not be limited to high tele-density areas and can disperse throughout the country. These results will support industrial policies focused on the software sector based on the hypothesis that policies can be eﬀective in the periphery, with low tele-density in the whole region.

170

Agglomeration in Asia

Table 6.17 Eﬀects of educational and R&D institutions on growth in the hardware sector Variable

Coeﬃcient 20

Coeﬃcient 21

Constant

0.117 (0.895) 0.774* (14.681) 0.046* (7.940) 0.165 ( 1.026) 0.009 ( 0.990) 0.002*** ( 1.466) 1.582** ( 1.650) 0.087 ( 1.365) 0.011 ( 1.135) 0.005 (0.763) 0.169* (3.490)

0.365** (1.665) 0.773* (14.503) 0.046* (7.773) 0.286** ( 1.777) 0.010 ( 1.053) 0.002 ( 1.393) 1.283 ( 0.719) 0.008 ( 0.089) 0.006 ( 0.532)

Log (Change in industrial structure) Competition Non-diversity Specialty Tele-density in 1998 Hard Dummy*Non-diversity Hard Dummy*Specialty Hard Dummy*Tele-density Log (No. of R&D Employees in 1998) Hard Dummy*Log (No. of R&D Employees in 1998) Log (No. of Education Employees in 1998) Hard Dummy*Log (No. of Education Employees in 1998) F-statistic Adjusted R-squared Number of observations Notes:

0.015 ( 0.946) 0.073 (0.703) 30.969 0.327 618

29.141 0.313 618

Coeﬃcient 22 0.410** (1.887) 0.774* (14.704) 0.046* (7.959) 0.232 ( 1.408) 0.010 ( 1.043) 0.002 ( 1.401) 0.082 (0.045) 0.027 ( 0.313) 0.005 ( 0.369) 0.012*** (1.470) 0.193* (3.529) 0.029** ( 1.637) 0.123 ( 1.050) 26.233 0.329 618

t-statistic in parentheses. * signiﬁcant at the 5%, ** 10% and *** 15% levels.

On the other hand, many venture companies concentrated in Teheran Valley in Seoul. MIC (2002b) cited the following reasons: formation of a networking group to strengthen cooperative ties between venture companies; easy exchange of information by sharing IT infrastructures; numerous advanced intelligent buildings and optical network infrastructures in the Teheran Valley; and the existence of ﬁnancial institutions, associations and government oﬃces that are closely related with venture businesses. All of these points emphasize the merit of the economy of agglomeration.

Industrial agglomeration and regional growth in Korea

171

REFERENCES Ahn, Hyungtaik, Jongjin Kim and Bun Song Lee (2000), ‘Information and telecommunications technology and regional economic growth in Korea’, http://ecobk21. uos.ac.kr/re/sub_re/kjjin/, accessed May 2002. Dahlman, C. and T. Andersson, (2000), Korea and the Knowledge-based Economy: Making the Transition, Washington, DC: World Bank Institute and Paris: OECD. Fujita, M. (2001), ‘Kuukan keizaigaku kara mita kokudo koutuu seisaku’, lecture given at Policy Research Institute for Land, Infrastructure and Transport, Ministry of Land, Infrastructure and Transport (in Japanese), Tokyo, Japan, 25 May. Glaeser, E.L., H.D. Kallal and A. Shleifer (1992), ‘Growth in cities’, Journal of Political Economy, 100 (6), 1126–52. Imagawa, T. (2002), Economic Analysis of Telecommunications, Technology, and Cities in Japan, Tokyo: Taga Shuppan. Ministry of Information and Communication, Republic of Korea (MIC) (2002a), ‘IT Korea 2002’, Seoul. Ministry of Information and Communication, Republic of Korea (MIC) (2002b), ‘IT Korea guide’, Seoul, May. Sakakibara, M. and Dong-Sung Cho (1999), ‘Cooperative R&D in Japan and Korea: a comparison of industrial policy’, Conference Paper 99-1-4, Research Institute of International Trade and Industry, Ministry of International Trade and Industry (MITI), Japan, May. Sakata, I., S. Nobuhara and K. Fujisue (eds) (2001), Conditions for Success in Technology Incubation: Fostering High-tech Start-ups through Technology Transfers from Universities (in Japanese), Tokyo: Tuusho Sangyou Chousakai. Ueki, Y. (2003), ‘Electronic industry in Asia: changing supply chain and the eﬀects’, in E. Giovannetti, M. Kagami and M. Tsuji (eds), The Internet Revolution: A Global Perspective, Cambridge: Cambridge University Press, pp. 82–102. Ueki, Y. (2004), ‘Jumping up to the Internet-based society: lessons from South Korea’, in M. Kagami, M. Tsuji and E. Giovannetti (eds), Information Technology Policy and the Digital Divide: Lessons for Developing Countries, Cheltenhan, UK and Northampton, MA, USA: Edward Elgar, pp. 114–34. Yoon, Myoung-hun (2001), ‘The present situation and prospects for high-tech industrial complexes in South Korea’, (in Japanese), Aziya Kenkyu (Asian Studies), 47 (1), January, 29–48.

172

Agglomeration in Asia

APPENDIX 6A Table 6A.1 ID 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

ID numbers of Korean industries

Industry

ID

Industry

Mining Food Products and Beverages Manufacture of Tobacco Products Textiles, Except Sewn Wearing Apparel Sewn Wearing Apparel & Fur Articles Tanning and Dressing of Leather Wood Products of Wood & Cork Pulp, Paper & Paper Products Publishing, Printing and Reproduction Coke, Reﬁned Petroleum Products Chemicals and Chemical Products Rubber & Plastics Products Non-metallic Mineral Products Manufacture of Basic Metals Fabricated Metal Products Manufacture of Other Machinery Computer & Oﬃce Machinery Electrical Machinery n.e.c. Radio, TV and Communication Equip. Medical, Precision & Optical Instruments Motor Vehicles & Trailers Mfg Manufacture of Other Transport Equip. Furniture, Articles n.e.c. Recycling Electricity, Gas and Water Supply

28 29

Construction Sale of Motor Vehicles and Motorcycles; Retail Sale of Automotive Fuel Wholesale Trade and Commission Trade, Except of Motor Vehicles and Motorcycles Retail Trade, Except Motor Vehicles and Motorcycles Hotels and Restaurants Transport, Storage and Communication Financial and Intermediation Insurance Real Estate Activities Renting of Machinery and Equipment without Operator and of Personal and Household Goods Computer and Related Activities Research and Development Other Business Activities Education Health and Social Work Recreational, Cultural and Sporting Activities Sewage and Refuse Disposal, Sanitation and Similar Activities Membership Organization n.e.c. Other Services Activities All Industries

30

31 32 33 34 35 36

37 38 39 40 41 42 43 44 45 0

7. China’s regional industrial disparity from the viewpoint of industrial agglomeration Koichiro Kimura* 1.

INTRODUCTION

Industrial location is unevenly distributed in the People’s Republic of China. Although China has obviously achieved historical growth as indicated by the annual average GDP growth rate from 1979 to 2001, 9.4 per cent, it is mainly observed in the eastern area (the coastal area), not in the central and western area (the inland area). Therefore, the growth rate does not represent the whole of China’s industry. This means that there is a spatial unevenness of economic activities. Although we need to investigate the relation between income and industrial disparity precisely, it is inferred from income diﬀerential between agriculture and industry and so on, that industrialization raises income in the agricultural inland area. To explain the industrial disparity, industrial agglomeration also has to be considered when investigating industrialization in China; for example, sources of economic growth such as capital accumulation, foreign direct investment, technology diﬀusion and so on.1 Therefore, in this chapter we shall not attempt an investigation of economic growth in the whole of China, but rather shall examine development and underdevelopment from the viewpoint of industrial agglomeration. This work follows theoretical literature on recent geographical economics and empirical literature on China’s industries. The former includes Krugman (1991), Fujita et al. (1999), Fujita and Hamaguchi (2001), Mori and Nishikimi (2002) and Fujita and Thisse (2002); and empirical literature, such as Kim (1995), Ellison and Glaeser (1997), Hanson (1998), Yue (2000) and Sonobe and Otsuka (2006). Spatial economics, the ‘new economic geography’, is a method of analysis for a spatial economic structure with monopolistic competition developed by Dixit and Stiglitz (1977), and transportation costs known as ‘iceberg transport’ which takes melted ice during transport as cost. The latter 173

174

Agglomeration in Asia

empirical literature includes Onoe (1971), Kato (1997, 2000), Qiu (1999), Cai and Du (2000), Seki and Nishizawa (2000), Amiti and Wen (2001), Fujita and Hu (2001), Kuroda (2001), Marukawa (2001), Tuan and Ng (2001), Batisse (2002), Jin and Zhu (2002) and Wei (2002). These studies cover not only spatial disparity and provincial diﬀerences, but also industrial agglomerations. In particular, some authors consider that regional integration can encourage agglomeration through the inter-regional division of labor, for example, Kato (1997, 2000) and Xu (2002). This work refers to Kato (2000) based on Krugman (1991). Kato investigated domestic market integration and regional development through concentration of manufacturing in a few regions modeled by Krugman. However, the investigation put weight on the inter-regional division of labor, rather than on industrial agglomeration directly. The purpose of this chapter is to examine the regional industrial disparity mentioned above from the viewpoint of industrial agglomeration. Industrial agglomeration is a phenomenon whereby many ﬁrms belonging to the same and associated industry concentrate geographically; moreover, they have dealings with one another.2 This chapter attempts to explain Chinese agglomeration using data and personal interviews based on spatial economics. In particular we shall examine the macroeconomic situation between the eastern area and the rest of China.3 The methodology is as follows. First, we shall investigate such concentrations since the beginning of contemporary Chinese history. The work is original in that some macro data and facts are utilized with a basic model of spatial economics to explain industrial agglomeration through linkages in China. In addition, we shall investigate the western development in the context of inevitability of agglomeration in the eastern area as a policy implication for developing economies. The conclusion of this chapter is that linkages resulting in industrial agglomeration observed in the machinery industry, mainly in the eastern area, is one of the factors that explains China’s regional industrial disparity. One policy implication from the conclusion is that there are problems in letting the inland area develop under the core–periphery structure. Therefore the inland development policy should mobilize the characteristics in each inland area with a view to developing pillar industries, a policy that the government is currently practicing. Accordingly, the contribution of this chapter is that some key parameters to explain agglomeration in Krugman (1991) are reconsidered, with some facts utilizing the theory. This will provide some clues for empirical study. The rest of this chapter is organized as follows. In Section 2, we conﬁrm the regional disparity with descriptive statistics, and a brief contemporary history. Section 3 considers the mechanism of industrial agglomeration. In

China’s regional industrial disparity

175

Section 4, as a policy implication, we examine the western development. Section 5 concludes with some remarks.

2. CHANGE IN REGIONAL INDUSTRIAL DISPARITY Index of Regional Industrial Disparity In this section, industrial disparity indices of the mining and manufacturing industries will be examined. We shall deal with the upper administrative divisions, that is to say, provinces, municipalities and autonomous regions.4 There are 31 divisions at present. To explain regional diﬀerentials, we apply Gini’s coeﬃcient as an index of industrial disparity, that is to say, measuring a concentrated (maldistributed) distribution of production value in the whole area. Although there are various indices to explain disparity, such as Gini’s coeﬃcient and Theil’s index, since we do not decompose reasons of inequality, Gini’s coeﬃcient will be utilized here. The coeﬃcient is for scaling income disparity, but, here we let the horizontal axis of the Lorenz curve be the accumulated number of provinces and the vertical axis represent the accumulated gross industrial output value (GIOV).5 We shall call this the ‘industrial disparity index’. Since the concept of ‘industry’ in China is broader than the general statistical classiﬁcation, this GIOV refers to the mining and manufacturing industries. When the industrial disparity index is close to one, there is a big gap in production among provinces, and vice versa. Therefore, if every province has the same ratio of GIOV, namely perfect equality, for example, it is about 3.23 per cent in the present case, then the industrial disparity index is zero. Figure 7.1 shows the changes in China’s regional disparity.6 We see from the ﬁgure that production disparity declined from the time of the nation’s foundation to the latter half of the 1980s, when the degree of disparity starts rising to the present time. This curve is related to given circumstances and policies. A brief history will be introduced here.7 Brief History Chinese history from 1949 to the present is divided into four periods from the viewpoint of industrial agglomeration. The ﬁrst period is the time of the initial conditions during the foundation of China under the leadership of the Chinese Communist Party in 1949. As the value of 1952, 0.5181 is an initial condition (see Figure 7.1), it is the largest number except for the

176

Agglomeration in Asia 0.60 0.55 0.50 0.45

0.35

1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

0.40

Notes: 1. Before 1997, all industrial enterprises. After 1998, all state-owned and non-state-owned industrial enterprises above the designated size refer to all state-owned industrial enterprises plus the non-state-owned industrial enterprises with an annual sales income of over 5 million yuan. 2. The GIOVs are calculated at current prices. Hebei’s ﬁgures for state-owned and collective-owned enterprises before 1988 are calculated at constant prices; since 1989, they are calculated at current prices. 3. The ﬁgure for Guangdong in 1995 has two sets of data in the statistics. The ﬁrst one is selected here. Sources: The ﬁgures for 1957 to 1998 are from Department of Comprehensive Statistics of the National Bureau of Statistics (1999), Comprehensive Statistical Data and Material on 50 years of New China, Beijing: China Statistical Press. The ﬁgures for 1998 to 2001 are from National Bureau of Statistics of China, China Statistical Yearbook, various years.

Figure 7.1 Industrial disparity index of the gross industrial output value, 1952–2001 latter half of the 1990s, and indicates an imbalance of industrial locations. According to Kojima (1997), Chinese industry was in dire straits at that time. Half of the spinning industry, the food industry and general goods was based in Shanghai, and the other half in Tianjin and Qingdao. Although most of the factories were shut down by the USSR, some heavy industry constructed under Japanese occupation remained, such as the Anshan Iron and Steel (Group) Company, located in the northeastern area. The second period is the time of the ﬁrst dispersion before the economic reform. The disparity index in this period was declining by about 0.1. Since China adopted a socialist economy, the government invested in projects of industrialization, giving them priority over heavy industries. In the ﬁrst FiveYear Plan (FYP) from 1953 to 1957, 117 projects of the 156 national priority projects aided by the USSR were located in the central and western areas to redress the production gap. In addition, these projects were ﬁnanced by taxes and the proﬁts of state-owned enterprises (SOEs).8 These industrialization and ﬁnance assistance projects were then relocated to inland provinces.

China’s regional industrial disparity

177

In the mid-1950s, although there was a mood of détente, relations between China and the US were aggravated because of the tensions over Korea and Taiwan. In addition, First Secretary Nikita Khrushchev’s criticism of Joseph Stalin in February 1956 led to a Sino-Soviet controversy over revisionism. Eventually, in 1960 the hostility between China and the USSR came to the surface. China had antagonized both the USSR and the US and this meant it was isolated from both camps, that is, China was isolated internationally. At this point China adopted a slogan for the economic system, ‘self-reliance’. A popular phrase used at that time was the ‘Construction of the Third Front’,9 which, as a result of the national defense strategy in the 1960s to 1970s, was a rearrangement whereby the government began to look to the inland area for the mining and manufacturing industries. According to Marukawa (1993), the ‘Third Front’ includes the whole or most of Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai and Ningxia, or the western parts of Shanxi, Henan, Hubei and Hunan in addition to the other seven provinces. The third period is the time of the second dispersion in the 1980s after the economic reform. Although the Chinese economy was centralized before the economic reform, after 1978, local governments also had decision-making authority because of the decentralization of power and revision of the ﬁnancial system. Therefore, these local governments were determined to have proﬁtable industries in each province. This regional protectionism led to a lordship economy that brought about overlapping investment in some industries. Therefore, as seen in Figure 7.1 in the 1980s, the dispersion continued after the strategy of balanced industrial location. The fourth period is the time of concentration in the 1990s. Setting up special economic zones after the end of the 1970s, particularly in the south coastal and other development areas, drove China’s economic growth, and it is likely that the economic reform in cities after 1984 also achieved results. The Southern Tour Lectures by Deng Xiaoping (who was retired from oﬃcial positions in the government and Chinese Communist Party) at the beginning of 1992 led to high growth, especially in southern coastal cities, following the Tian’anmen incident in 1989. After a reduction in disparity in the eastern area in the 1980s and a rise in the total share of the eastern area as a whole, the industrial disparity index began to increase. Movements of the Production Bases Related to the index and the history, Table 7.1 and Figures 7.1–3 explain the movements of the production bases. Table 7.1 shows the results on the centralization of production by year.

178

1952 1957 1962 1965 1970 1975 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994

Year

First (%)

19.3 16.4 16.4 16.0 13.7 12.7 11.7 11.5 11.3 11.2 10.6 10.0 9.7 10.7 11.1 11.6 12.1 11.8 12.0 11.4 13.1 13.8 13.8

Liaoning Liaoning Liaoning Liaoning Liaoning Liaoning Liaoning Liaoning Jiangsu Jiangsu Jiangsu Jiangsu Shanghai Shanghai Shanghai Shanghai Shandong Shandong Shandong Shandong Guangdong Guangdong Guangdong

13.1 16.2 11.2 11.9 11.7 10.2 9.1 8.8 8.9 9.1 9.0 9.1 9.4 8.9 8.5 7.9 8.2 9.0 9.5 9.4 9.8 10.2 10.2

Second (%)

Jiangsu Shandong Jiangsu Heilongiang Shandong Jiangsu Jiangsu Jiangsu Liaoning Liaoning Liaoning Liaoning Liaoning Liaoning Liaoning Shandong Guangdong Guangdong Guangdong Guangdong Shandong Shandong Shandong

Third (%)

Movements of production bases

Shanghai Shanghai Shanghai Shanghai Shanghai Shanghai Shanghai Shanghai Shanghai Shanghai Shanghai Shanghai Jiangsu Jiangsu Jiangsu Jiangsu Jiangsu Jiangsu Jiangsu Jiangsu Jiangsu Jiangsu Jiangsu

Table 7.1

7.4 6.0 5.8 6.1 6.2 7.1 7.7 8.0 8.8 8.5 8.2 8.1 7.9 7.4 7.1 7.6 7.4 7.7 8.2 9.1 8.7 9.1 9.9

Shandong Jiangsu Guangdong Jiangsu Heilongiang Beijing Shandong Shandong Shandong Shandong Shandong Shandong Shandong Shandong Shandong Liaoning Shanghai Shanghai Shanghai Shanghai Zhejiang Zhejiang Zhejiang

Fourth (%)

5.8 5.7 5.5 6.1 6.0 6.0 6.8 6.5 6.4 6.5 6.6 6.7 6.9 7.0 7.0 6.5 7.3 7.3 7.1 7.0 6.9 7.4 8.1

Guangdong Tianjin Heilongiang Shanxi Jiangsu Shandong Beijing Beijing Beijing Beijing Beijing Beijing Guangdong Zhejiang Zhejiang Guangdong Zhejiang Zhejiang Zhejiang Liaoning Shanghai Shanghai Shanghai

Fifth (%)

5.8 5.6 5.4 5.1 6.0 5.7 5.8 5.9 5.9 5.7 5.6 5.5 5.6 5.7 6.1 6.4 6.4 6.3 6.2 6.5 6.8 6.5 6.0

39.9 38.6 33.4 34.0 31.7 30.0 28.5 28.3 29.0 28.8 29.9 27.2 27.0 26.9 26.7 27.1 27.6 28.6 29.7 29.9 31.6 33.1 33.9

Sum of top three (%) 68.9 66.8 62.5 61.0 61.0 60.5 60.5 60.0 61.2 61.5 60.6 60.0 60.6 60.7 60.6 61.2 62.1 62.2 62.7 64.4 65.8 66.6 66.7

East 23.9 24.6 28.1 29.5 28.4 28.9 28.3 28.9 28.4 28.4 29.0 29.4 28.7 28.6 28.8 28.5 27.6 27.3 26.6 25.2 24.3 23.9 24.3

Central

Each area share (%)

7.2 8.7 9.4 9.5 10.6 10.6 11.2 11.1 10.5 10.0 10.5 10.6 10.7 10.7 10.5 10.2 10.3 10.6 10.7 10.4 9.8 9.4 8.9

West

0.5181 0.4917 0.4588 0.4632 0.4490 0.4333 0.4217 0.4174 0.4235 0.4250 0.4145 0.4112 0.4112 0.4119 0.4143 0.4201 0.4429 0.4433 0.4479 0.4584 0.4745 0.4861 0.5004

Industrial disparity index

179

Source:

See Figure 7.1.

See Figure 7.1.

Note:

11.8 11.9 11.3 11.9 14.5 14.6 14.7

Jiangsu Jiangsu Jiangsu Guangdong Guangdong Guangdong Guangdong

1995 1996 1997 1998 1999 2000 2001

Guangdong Guangdong Guangdong Jiangsu Jiangsu Jiangsu Jiangsu

11.7 10.8 11.2 11.4 12.3 12.2 12.3

Zhejiang Shandong Zhejiang Zhejiang Shandong Shandong Shandong

9.9 9.4 9.4 9.8 9.6 9.7 9.8

Shandong Zhejiang Shandong Shandong Shanghai Zhejiang Zhejiang

9.2 9.1 9.0 9.1 7.5 7.7 8.3

Shanghai Henan Hebei Hubei Zhejiang Shanghai Shanghai

6.5 5.4 5.4 5.8 7.1 7.2 7.3

33.5 32.1 31.9 33.1 36.3 36.5 36.8

66.8 65.2 65.1 67.2 70.1 70.9 71.6

24.6 25.4 25.6 23.3 20.2 19.7 19.1

8.5 9.4 9.3 9.5 9.6 9.4 9.3

0.5119 0.5074 0.5085 0.5219 0.5216 0.5253 0.5311

180

Agglomeration in Asia

Columns 2–6 show how the fortunes of the ﬁve highest-ranking provinces have ﬂuctuated. With regard to the coastal area (except for Liaoning), although the northeastern area had a high share at ﬁrst, the southeastern area increased its share after the mid-1980s. The Yangtze River Delta area has retained a high share as a whole area, but Shanghai itself dropped in rank from the mid-1980s to the mid-1990s. Especially symbolic is that the rankings of Shanghai and Guangdong were reversed in 1989. The Pudong new district in Shanghai started to develop in 1990, but much foreign direct investment had already gone to Guangdong. Jiangsu raised its ranking after the economic reform and ranked ﬁrst from the mid-1980s to the early 1990s. Since the mid-1990s, Zhejiang has been ranked in the top ﬁve provinces in production value. At this time, the Yangtze River Delta was upgraded. Shandong has consistently enjoyed a high share. It has not only a high GIOV, but also a high agricultural value, which accounts for about 10 per cent and ranks ﬁrst in China. Indeed, we can think of Shandong as an independent economic zone. Next, we turn to the inland area, including Liaoning. In the northeastern area, Liaoning and Heilongjiang are also ranked in the table, however Liaoning, which had featured regularly, was dropped from the early 1990s. In the northeastern area, Tianjin was ranked three times before the economic reform. Similarly, Beijing had also been ranked until the ﬁrst half of the 1980s. In the last half of the 1990s, Henan, Hebei and Hubei, belonging to the central area, were ranked ﬁfth. Indeed, provinces in the eastern area generally, have high shares, however, those near the coast or along the Yangtze River were also developed as industrial provinces. We therefore know that many provinces in the western and central areas have only a low share. Column 7 of Table 7.1 shows how the sum of the share of the top three provinces has changed. This change has almost the same shape as the industrial disparity index. First, even the ﬁfth-ranking provinces have shares of about 5–7 per cent, indicating that all unranked provinces have only a low share. Column 8 of Table 7.1 and Figure 7.2 show the eastern, central and western share of GIOV from 1952 to 2001:10 in the eastern area the share of 68.9 per cent in 1952 had decreased to 60.6 per cent by 1986; in the central area the share of 23.9 per cent in 1952 had increased to 28.8 per cent by 1986; and in the western area the share of 7.2 per cent in 1952 had increased to 11.2 per cent by 1978. Therefore, we can guess that the decrease in the eastern share and the increase in the central and western areas was aﬀected by the Construction of the Third Front (especially the western area), the economic reforms in rural areas, and protectionism by local governments. On the other hand, the eastern share has been increasing since the

100 90 80 70 60 50 40 30 20 10 0

181

1952 1957 1962 1965 1970 1975 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

China’s regional industrial disparity

Eastern area

Central area

Western area

Note: See Figure 7.1. The years without ﬁgures are omitted. Source: See Figure 7.1.

Figure 7.2

Ratio of each area (%)

mid-1980s. According to the National Bureau of Statistics of China (2002), the population in 2000 was eastern area: 41.6 per cent; central: 35.3 per cent; and western: 23.1 per cent (year-end); therefore the industrial concentration in the eastern area is stronger than the population distribution. Figure 7.3 shows that there is a relation between the change rates of the industrial disparity index and those of the GIOV. Although change rates of the industrial disparity index were negative before the mid-1980s, from that time both rates start to move in a similar direction. Since the economy at the end of the 1980s was shrinking, we can see from the changes that regional industrial disparity has increased with economic growth in the 1990s. Conclusion China’s regional industrial disparity had been decreasing until the 1980s because of political reasons and regional protectionism. But, the disparity has been increasing since the 1990s. Movements of production bases brought about the change of disparity in terms of the share between the eastern and other areas, and movements between the northeast and the southeast. In addition, agglomerations in some central area provinces near the coast or along the Yangtze River were also found.

182

Agglomeration in Asia GIOV 50

Industrial disparity index 0.06

40

0.05

30

0.04

20

0.03

10

0.02

–10 –20

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

0

0.01 0.00

–30

ñ0.01

–40

ñ0.02

–50

GIOV

Industrial disparity index

ñ0.03

Note: See Figure 7.1. Source: See Figure 7.1.

Figure 7.3 Change rates of GIOV and industrial disparity index, 1980–2001 (%)

3. INDUSTRIAL AGGLOMERATION IN THE EASTERN AREA Deﬁnition of Industrial Agglomeration Here, we shall specify the causes of an industrial agglomeration through linkages, modeled by recent literature on spatial economics such as Fujita et al. (1999). Industrial agglomeration as a phenomenon, that is, production concentrations in some areas, is developed as a result of the following parameters: an increase in the demand for manufactures; a decline in transportation cost; and the importance of division of labor among ﬁrms.11 This is deﬁned by referring to Fujita et al. and Fujita and Hamaguchi (2001), who examined the relations between transactions of intermediate goods and spatial structures. In addition to the papers on spatial economics, the notion of division of labor among ﬁrms is based on Dixit and Stiglitz (1977), explaining the relation between a variety of commodities and scale economies. Here, we shall use the phrase ‘division of labor among ﬁrms’, hence, this notion is related to specialization in the sense of raising the productivity of various manufacturing processes. Krugman (1991) examined an industry sector with imperfect competition due to increasing returns to scale, to ascertain whether consumption goods manufacturers locate uniformly. The result is related to an increase

China’s regional industrial disparity

183

in demand by manufacturers, a decline in transportation cost, and consumers preferring a variety of consumption goods. Here we let the ﬁnal goods of industries be intermediate ones of the machinery industry. Then the relative relations of an increase in demand for the machinery industry, a decline in transportation cost, and the importance of variety generate linkages as well as the deﬁnition of agglomeration.12 It is important to consider the mechanism of industrial agglomeration. The government played a big role in changing industrial locations as shown in the previous section, which showed that the government was instrumental in gathering ﬁrms together into speciﬁc areas. But, it is necessary to conﬁrm whether ﬁrms themselves decided to agglomerate, following governmental intervention, namely what was their economic rationality. Cases introduced in Section 2 were not all gathered by the government. Development zones are examples of this.13 They oﬀer hard and soft infrastructures, such as sites, utilities, the introduction of accountants and lawyers, and so on, to reduce costs for ﬁrms. In addition to the development zones, the most important policy when we consider the development of the eastern area is the introduction of foreign capital, especially after 1992.14 Although we cannot ignore this good treatment strategy for the eastern area, however, we shall focus on the agglomeration process after investment. The Machinery Industry In selecting an industry in order to discuss agglomeration in China, we have chosen the machinery industry, since every industry demands machines to raise productivity. In addition, since the processing and assembly industry, namely, the machinery industry, brings larger and deeper production levels than the basic materials industry (see Murakami 1999 and Kitakyushu City 2002), this leads potentially to industrial agglomeration under transportation costs.15 In fact, as the processing and assembly industry requires a lot of parts to produce a machine, the industry has a tendency to a division of labor among ﬁrms, in other words, it employs a geographical system or a network of suppliers.16 This is consistent with the above-mentioned hypothesis of variety applied to the machinery industry. In particular, this development of deeper production levels corresponds to specialization and a division of labor among ﬁrms.17 To compare the same industrial classiﬁcation revised in 1994, Figures 7.4a and 7.4b, show the industrial disparity indices by sector in 2000 and the machinery industry in 1994, respectively. In Figure 7.4a, since the arithmetic mean of the industrial disparity index of manufacturing is 0.5952, we can see that the regional disparity of the machinery industry is relatively

184

0.4

0.3

Ordinary Machinery Special Purpose Equipment Transport Equipment Electric Equipment and Machinery Electronic and Telecommunications Equipment Instruments, Meters, Cultural and Office Machinery

0.4500

0.5

0.5589 0.5529 0.5277

0.6

0.5564

0.5389 0.5225

0.5714 0.5309

Figure 7.4

Industrial disparity index, by sector, 2000 and 1994

0.7

0.6338

0.6261 0.6995

0.7

0.7005

0.8 0.7529

0.8

0.7572 0.7524

0.7251

0.6957

0.6915 0.6588 0.6358

0.6279

0.6704 0.6535

0.6221

0.5597 0.5478 0.5042 0.6007

0.6

The ﬁgures for 4a are from DITSNBS (2001). The ﬁgures for 4b are from DITSNBS (1996).

0.4

0.5

0.4507

0.4207

b: Industrial disparity index, by sector of the machinery industry, 1994

Notes: 1. See Note 1, Figure 7.1. 2. The GIOVs are calculated at current prices.

Sources:

0.3

Coal Mining and Dressing Extraction of Petroleum and Natural Gas Mining and Dressing of Ferrous Metals Mining and Dressing of Nonferrous Metals Food Processing Food Manufacturing Beverages Tobacco Textiles Papermaking and Paper Products Petroleum Processing and Cooking Raw Chemical Materials and Chemical Products Medical and Pharmaceutical Products Chemical Fiber Nonmetal Mineral Products Smelting and Pressing of Ferrous Metals Smelting and Pressing of Nonferrous Metals Metal Products Ordinary Machinery Special Purpose Equipment Transport Equipment Electric Equipment and Machinery Electronic and Telecommunications Equipment Instruments, Meters, Cultural and Office Machinery Production and Supply of Electric Power, Steam and Hot Water

a: Industrial Disparity Index, by sector, 2000

China’s regional industrial disparity

185

high, except for transport equipment.18 The arithmetic mean of the industrial disparity index of the machinery industry is 0.6768. On the other hand, in Figure 7.4b, the arithmetic mean of the industrial disparity index for the machinery industry is 0.5998. We can see that the machinery industry had been concentrated since 1994. An Explanation for China’s Industrial Agglomeration To consider each parameter, we show the share of the machinery industry’s output value in the mining and manufacturing industries, and the division of labor between ﬁrms using macro data and a personal interview for each proxy, except for transportation cost. The states of the coastal eastern development will be examined using macro data and a personal interview. Share of the machinery industry in the mining and manufacturing industries Let us conﬁrm the changing needs of a machinery industry sector. There are demands from every sector as mentioned above, but we shall focus on those for the machinery industry, which come from the industry itself because of transactions among ﬁrms classiﬁed in that industry.19 Table 7.2 indicates that the share of machinery industry output value in the mining and manufacturing industries has been growing since 1994. This provides a foundation for agglomeration within the machinery industry in some places.20 Some data for intermediate goods transactions We shall examine the state of transactions of intermediate goods among ﬁrms. Since there is agglomeration in the eastern area as shown in Section 2, we shall focus on the eastern area here. But, as our interest here is to explain the relation between agglomeration itself and intermediate goods transactions, such transactions will be examined as follows: this subsection deals with the period from the mid-1980s to the mid-1990s, and the present and the future will be dealt with in the following subsection. Murakami (1999) studied transactions among ﬁrms to investigate the development of the division of labor among ﬁrms. He cited the 1985 industrial census and the analysis by a team of the State Council, which found that the average of external purchases of parts to gross industrial total value in China, 44.96 per cent, was much lower than the average in developed countries, 70–80 per cent (for example, in Japan and the US). In addition, this ﬁgure was an average; representative types of industry, such as metalworking machinery and electric machinery, were lower than the average. Therefore, the low ﬁgures were indicative of the underdevelopment of division of labor among ﬁrms at that time.

186

1994

(%)

1995

(%)

1996

(%)

1997

(%)

1998

Share of the machinery industry in GIOV, 1994–2000 (at current prices) (%)

1999

(%)

2000

(%)

Sources: The ﬁgures for 1994 to 1999 are from DITSNBS (2000), China Industry, Tranport, and Energy Statistical Data and Materials on 50 years (1949–1999), Beijing: China Statistical Press. The ﬁgures for 2000 are from DITSNBS (2000), China Industry Economy Statistical Yearbook 2001, Beijing: China Statistical Press.

Notes: 1. See Figure 7.4. 2. The ﬁgures for values are at current prices.

GIOV 51 353.03 54 946.86 62 740.16 68 352.68 67 737.14 72 707.04 85 673.66 Machinery industry 12 120.80 23.6 12 975.99 23.6 15 093.65 24.1 16 894.54 24.7 17 955.97 26.5 19 892.16 27.4 23 856.58 27.8 Ordinary machinery 2391.75 4.7 2365.69 4.3 2680.92 4.3 2813.35 4.1 2579.80 3.8 2693.90 3.7 3046.95 3.6 Special purpose equipment 1791.90 3.5 1756.54 3.2 1988.14 3.2 2071.02 3.0 1920.27 2.8 1980.71 2.7 2192.63 2.6 Transport equipment 3185.80 6.2 3303.28 6.0 3785.01 6.0 4123.1 6.0 4241.01 6.3 4659.31 6.4 5364.83 6.3 Electric equipment 2327.04 4.5 2594.30 4.7 3059.76 4.9 3366.09 4.9 3628.58 5.4 4021.55 5.5 4834.68 5.6 and machinery Electronic and 1999.86 3.9 2530.48 4.6 3051.09 4.9 3 921.03 5.7 4893.56 7.2 5830.96 8.0 7549.58 8.8 telecommunications equipment Instruments, meters, 424.45 0.8 425.70 0.8 528.73 0.8 599.95 0.9 692.75 1.0 705.73 1.0 867.91 1.0 cultural and oﬃce machinery

Table 7.2

China’s regional industrial disparity

187

After discussing the census results, Murakami analysed the development of transactions among ﬁrms from his enterprise survey in Wuhan, as seen in Table 7.3. The share of Wuhan’s machinery industry in the nationwide total grew from 1.04 per cent in 1994 to 1.24 per cent in 1997, and from 1.23 per cent in 1998 to 1.28 per cent in 2000.21 Therefore, although Wuhan is a city in the central area, it also has agglomeration at the city level. In fact, in Table 7.1, Hubei (including its capital Wuhan) was ranked ﬁfth in 1998. Dongfeng-Citroën Automobile Co. Ltd is also located there, and as a consequence the automobile industry is one of the four major industries of Wuhan. In addition, Wuhan is as much a stronghold of higher education and scientiﬁc research as Beijing, Shanghai and Xi’an. Table 7.3 shows that the ratios of the machinery industry are higher than those of the textile industry because the former is a processing and assembly industry. Moreover, there are increases in the machinery industry ratio after 1985. In addition, the table shows more recently founded SOEs and urban collective-owned enterprises (UCOEs) had placed their orders with outside suppliers. In addition to the survey in Wuhan, Murakami examined the case of Tianjin from 1985 to 1990. Table 7.4 shows that Tianjin’s ratios were higher than Wuhan’s, that is to say, the division of labor among ﬁrms in Tianjin was more developed. But, since the diﬀerences are not excessive, there are limits to the development. Next, we shall utilize China’s inter-regional input–output (IO) table, developed by Okamoto (2002). ‘Inter-regional’ includes the eastern, central and western areas as adopted above. Tables 7.5a and 7.5b summarize the machinery industry without and with electric and electronic equipment, in 1987 and 1997, respectively. In this table and in Tables 7.6 and 7.7, each row indicates the origin of each sector, and each column indicates the destination. Although various facts can be gleaned from the tables, we shall focus on intra-regional transactions of intermediate goods in each area. In 1987, since every intra-regional transaction of the same sector is the largest value except for the intra-regional transaction of electronic equipment in the central area, we can see that production in each area supplied parts for each local demand. On the other hand, in 1997, transactions with the eastern area are larger than the intra-regional transactions in the central and western areas except for the electronics in the central. Table 7.6 shows the ratio of increase between two points in time. As the table indicates, only the ratios of the eastern area, 9.68 and 13.05, exceeded the ratios of total supply, 9.07 and 11.86, and demand, 8.96 and 10.83, of each sector. In other words, intra-area transaction among ﬁrms in the eastern area exceeded output to outside and input from outside.

188

9.4 9.4 11.2 11.4 n.a. n.a.

(21) (21) (14) (16)

1980

6.1 6.2 12.3 10.5 n.a. n.a.

(21) (24) (14) (18)

1985 6.2 7.3 14.7 15.3 n.a. n.a.

(21) (26) (14) (24)

1990

Machinery industry

7.0 9.3 18.0 19.5 5.0 16.1

(21) (27) (14) (29)

1994 2.9 2.8 2.3 2.7 n.a. n.a.

(15) (16) (11) (13)

1980 (15) (16) (11) (15)

3.8 3.6 2.9 2.3 n.a. n.a.

(15) (17) (11) (19)

1990

Textile industry 1985 3.0 2.8 1.9 2.0 n.a. n.a.

External purchase of parts to total external purchase ratio in Wuhan (%)

3.7 3.0 1.6 1.0 2.8 4.7

(15) (19) (11) (17)

1994

Source:

Murakami (1999).

Notes: 1. SOEs: state-owned enterprises; UCOEs: urban collective-owned enterprises; TVEs: township and village enterprises; PEs: private enterprises; JVs: joint-venture enterprises (foreign-aﬃliated enterprises). 2. External purchase of parts to total external purchase ratioExternal purchase of parts / (External purchase of parts External purchase of raw material) 100. 3. The ﬁgures in parentheses are the number of sample ﬁrms. 4. The upper ﬁgures for SOEs and UCOMs indicate samples for all four years, and the lower ﬁgures are all valid samples for each year. 5. n.a.: not available.

TVEs PEs and JVs

UCOEs

SOEs

Table 7.3

189

China’s regional industrial disparity

Table 7.4 External purchase of parts to total external purchase ratio in Tianjin (%)

SOEs UCOEs PEs and JVs

1985

1986

1987

1988

1989

1990

21.1 21.3 30.0

21.3 21.0 30.2

20.4 20.1 29.6

21.1 20.1 28.0

20.7 20.5 28.4

21.2 20.8 26.8

Notes: 1. See Notes 1 and 2 of Table 7.3. 2. The number of samples for SOEs, UCOEs, and PEs and JVs are 89, 43 and 5, respectively. Source: See Table 7.3.

Tables 7.7a and 7.7b are ratios of each area demand to each area supply, that is to say, the value of supply dependence by area. These tables tell us that dependence of supply from the eastern area for the central and the western areas was strengthened from 1987 to 1997, particularly the dependence of the central area. Inﬂuences from the eastern to the western area seem to be weaker than to the central area from the viewpoint of transportation cost in a wide sense. In the same way, Tables 7.7c and 7.7d are ratios of each area supply to each area demand, that is to say, the value of demand dependence by area. These tables also tell us that dependence on supply from the eastern area was strengthened, especially for the central and western areas. A case of intermediate goods transactions This subsection will discuss the case of transaction of parts among ﬁrms in the eastern area. This case is based on a personal interview (12 September 2002, in Guangzhou) at Japanese transport equipment Company A (established in 1998) producing automobiles in Guangzhou, Guangdong. By March 2002, Company A had produced 100 000 cars since beginning operations. The company assembles about 1800 parts to produce an automobile at their assembly factory. About 60 per cent of the items are purchased in China and the rest are imported from foreign countries. Some 70 per cent of the 40 per cent are imported from Japan and the remaining 30 per cent are imported from the US, Malaysia and Thailand. Half of the intraChina purchasing is from Guangdong, about 30 per cent from Shanghai and about 20 per cent from other areas. Therefore, quantitatively, many parts are produced in the eastern area, where there are two large industrial

190

38 986 008

Column total

28 975 880

2 870 148 23 211 517 521 811 1 272 488 80 414 1 019 502

E-Electro

2 253 971

297 622 1 651 930 60 930 178 580 12 949 51 960

E-Electro

6 012 811

3 317 625 285 305 2 124 994 55 104 202 485 27 298

C-General

1 275 958

445 577 86 615 613 833 63 007 58 680 8246

C-General

632 573

106 134 350 693 67 981 67 733 6478 33 554

C-Electro

289 592

27 656 120 810 38 100 87 882 3642 11 502

C-Electro

1 747 015

689 067 116 659 84 364 4307 673 789 178 829

W-General

628 957

116 748 29 222 32 492 4294 377 832 68 369

W-General

Source:

Okamoto (2002).

Notes: 1. E, C and W: eastern, central and western areas, respectively. 2. General: machinery industry without electric and electronic equipment; Electro: electric and electronic equipment.

27 053 211 5 694 137 4 918 440 312 161 757 960 250 099

E-General E-Electro C-General C-Electro W-General W-Electro

E-General

3 873 221

Column total

b: 1997

2 532 019 625 392 518 366 67 607 110 166 19 671

E-General

E-General E-Electro C-General C-Electro W-General W-Electro

a: 1987

Table 7.5 Inter-regional input–output table, 1987 and 1997

2 156 534

114 605 745 507 14 031 27 523 112 064 1 142 804

W-Electro

239 345

9921 55 802 2761 8201 32 106 130 554

W-Electro

78 510 821

34 150 790 30 403 818 7 731 621 1 739 316 1 833 190 2 652 086

Row total

8 561 044

3 4 29 543 2 5 69 771 1 2 66 482 409 571 595 375 290 302

Row total

191

China’s regional industrial disparity

Table 7.6

Ratio of increase between 1987 and 1997 EECCWWRatio to General Electro General Electro General Electro total

E-General E-Electro C-General C-Electro W-General W-Electro Ratio to total

9.68 8.10 8.49 3.62 5.88 11.71

8.64 13.05 7.56 6.13 5.21 18.62

6.45 2.29 2.46 0.13 2.45 2.31

2.84 1.90 0.78 0.23 0.78 1.92

4.90 2.99 1.60 0.00 0.78 1.62

10.55 12.36 4.08 2.36 2.49 7.75

8.96 10.83 5.10 3.25 2.08 8.14

9.07

11.86

3.71

1.18

1.78

8.01

8.17

Note: See Table 7.5. Source: See Table 7.5.

agglomerations of automobile companies. One of these, in the Yangtze River Delta area, has a longer history than the other, Guangdong, because European and American automobile companies were established before Company A in the Yangtze River Delta area. But the agglomeration in Guangdong has also been developed since there was an investment input. Half of the items are purchased in Guangdong, where Company A is located. There are many intermediate goods companies located in the Pearl River Delta area, near Company A. In particular, the suppliers are located in Guangzhou, Dongguan, Foshan and Zhongshan. In addition, companies making large volume parts located near Company A after taking transportation costs into consideration. Consequently, many direct parts companies for Company A are gathered in a speciﬁc area.22 Next, let us examine the situation from a capital point of view. Company A has dealings with about 100 companies, about 75 per cent of which are Japanese companies, about 20 per cent are Chinese, and the others are European and American. In addition, by value, 90 per cent or more are purchased from Japanese companies with about 5–6 per cent purchased from Chinese companies. Here we should provide some supplementary explanation. According to China’s customs statistics, the import value of machinery has been increasing the same as export value (see Table 7.8).23 Therefore, we see from the foreign trade statistics that China needs intermediate goods from foreign countries in order to export more machinery because of the underdevelopment of the supporting industry. In fact as seen in Table 7.9, the value of parts and accessories for four-wheeled vehicles has been increasing in the same way as the export value. On the other hand, export values of parts

192

0.79 0.19 0.64 0.18 0.41 0.09

0.50

Ratio to total

E-General

0.45

0.74 0.24 0.41 0.17 0.19 0.07

E-General E-Electro C-General C-Electro W-General W-Electro

b: Supply, 1997

Ratio to total

E-General E-Electro C-General C-Electro W-General W-Electro

E-General

0.37

0.08 0.76 0.07 0.73 0.04 0.38

E-Electro

0.26

0.09 0.64 0.05 0.44 0.02 0.18

E-Electro

0.08

0.10 0.01 0.27 0.03 0.11 0.01

C-General

0.15

0.13 0.03 0.48 0.15 0.10 0.03

C-General

0.01

0.00 0.01 0.01 0.04 0.00 0.01

C-Electro

0.03

0.01 0.05 0.03 0.21 0.01 0.04

C-Electro

Dependence ratio of supply and demand, 1987 and 1997

a: Supply, 1987

Table 7.7

0.02

0.02 0.00 0.01 0.00 0.37 0.07

W-General

0.07

0.03 0.01 0.03 0.01 0.63 0.24

W-General

0.03

0.00 0.02 0.00 0.02 0.06 0.43

W-Electro

0.03

0.00 0.02 0.00 0.02 0.05 0.45

W-Electro

1.00

1.00 1.00 1.00 1.00 1.00 1.00

Ratio to total

1.00

1.00 1.00 1.00 1.00 1.00 1.00

Ratio to total

193

Source:

See Table 7.5.

See Table 7.5.

1.00

Ratio to total

Note:

0.69 0.15 0.13 0.01 0.02 0.01

E-General E-Electro C-General C-Electro W-General W-Electro

E-General

1.00

Ratio to total

d: Demand, 1997

0.65 0.16 0.13 0.02 0.03 0.01

E-General

E-General E-Electro C-General C-Electro W-General W-Electro

c: Demand, 1987

1.00

0.10 0.80 0.02 0.04 0.00 0.04

E-Electro

1.00

0.13 0.73 0.03 0.08 0.01 0.02

E-Electro

1.00

0.55 0.05 0.35 0.01 0.03 0.00

C-General

1.00

0.35 0.07 0.48 0.05 0.05 0.01

C-General

1.00

0.17 0.55 0.11 0.11 0.01 0.05

C-Electro

1.00

0.10 0.42 0.13 0.30 0.01 0.04

C-Electro

1.00

0.39 0.07 0.05 0.00 0.39 0.10

W-General

1.00

0.19 0.05 0.05 0.01 0.60 0.11

W-General

1.00

0.05 0.35 0.01 0.01 0.05 0.53

W-Electro

1.00

0.04 0.23 0.01 0.03 0.13 0.55

W-Electro

1.00

0.43 0.39 0.10 0.02 0.02 0.03

Ratio to total

1.00

0.40 0.30 0.15 0.05 0.07 0.03

Ratio to total

194

Agglomeration in Asia

Table 7.8

Export Import

Foreign trade value of machinery, 1995–2001 (US$m) 1995

1996

1997

1998

1999

2000

2001

36 691.95 57 023.09

40 612.90 59 120.60

50 129.12 57 010.24

56 721.07 61 636.60

65 713.23 75 018.08

90 706.42 99 766.05

102 944.89 117 040.53

Source: China Department of Customs (World Trade Atlas provided by Global Trade Services, Inc.).

Table 7.9 Foreign trade value of four-wheeled vehicles and motorcycles, 1995–2001 (US$m) 1995

1997

1998

1999

2000

2001

Four-wheeled vehicles Export 223.97 206.39 Import 1536.25 841.80

247.77 726.14

201.47 855.84

160.26 850.98

252.69 1217.01

262.26 1765.89

Parts and accessories Export 373.26 379.64 Import 748.11 1043.93

444.66 901.91

527.07 931.32

778.95 1244.19

1121.31 2103.63

1350.60 2514.80

42.98 1.51

61.84 2.37

64.83 0.32

127.34 2.32

747.07 3.00

747.42 1.27

367.76 233.22

480.99 180.20

488.47 170.26

484.91 213.24

657.55 237.66

752.86 185.18

Motorcycles Export Import

46.81 23.82

Parts and accessories Export 290.28 Import 226.22

1996

Notes: Four-wheeled vehicles: 8701–8705 in HS code; Parts and accessories under fourwheeled vehicles: 8708 (for 8701–8705) in HS code; Motorcycles: 8711 in HS code; Parts and accessories under motorcycles: 8714 (for 8711–8713) in HS code. Source: See Table 7.6.

and accessories for motorcycles and four-wheeled vehicles, which have a well-developed industry, exceed import values. This is the present situation, in 2002, but the intention is to raise the value and the ratio of local content in order to reduce costs in the future. Therefore, the low share held by Chinese companies makes it clear that there are few local players even in the eastern area conducting transactions with Company A. However, it can be seen from the example of Company A that there is the possibility for agglomeration development, including by Chinese ﬁrms.

China’s regional industrial disparity

195

Conclusion To consider the mechanism for the change in disparity we deﬁned an industrial agglomeration as production concentrations in some areas. We focused on the machinery industry on account of its importance regarding productivity for all other industries and its character as the processing and assembly industry. At the same time, we saw that the disparity of machinery industry production among the provinces has increased. To investigate the inevitability of agglomeration we examined the following issues: (i) growth in the share of the machinery industry in the mining and manufacturing industries; (ii) an increase in intermediate goods transaction among ﬁrms in the city; and (iii) an increase in transactions in the eastern area. In addition, the actual state of transactions among ﬁrms was introduced with reference to the speciﬁc case of a Japanese motorcar company. Thus it was shown that linkages, namely, linkage-based spatial concentrations, are a factor in industrial agglomeration. We also explained that the inevitability of agglomeration in the eastern area and near provinces is a core area; that is to say, the machinery industry in the area is growing steadily.

4.

THE WESTERN DEVELOPMENT

The 10th FYP and the Western Development The 10th FYP (2001–05) was ratiﬁed at the fourth plenum of the ninth National People’s Congress in 2001. In an attempt to correct the regional disparity between the coastal and the inland areas the plan has put weight on the western development.24 The push for the western development is thought to have originated in President Jiang Zeming’s speech at Xi’an in 1999. In the 10th FYP, the government put weight on western infrastructure, scientiﬁc research and education, the ecological environment, ethnic minorities, natural resources, agriculture and so on. Some major infrastructural projects, such as the transfer of natural gas and electric power from the west to the east, and railways from Qinghai to Tibet, are planned. The Big Push in a Core–Periphery Structure This western development represents a case of the big-push theory as discussed by Kato (2000). Here, however, we see that the big-push theory proposed by Rosenstein-Rodan (1943) is to produce the intra-industry linkages (both forward and backward) artiﬁcially.25 Therefore, the greater the

196

Agglomeration in Asia

investment in the western area the greater the possibility of moving to a higher equilibrium. But, the relation between the coastal and inland areas is the core– periphery structure in the context of agglomeration.26 That is, the problem facing the western area is in some aspects a structural one. The big-push theory deals with point economy, that is to say, non-space economy. Therefore, unless we consider neo-classical economic growth completely here, it is more diﬃcult to escape from the state of underdevelopment of the periphery. In addition, since Deng Xiaoping proposed allowing some people and areas to get rich ﬁrst, the possibility of achieving this end will be more diﬃcult because of the lock-in eﬀect. Moreover, even if the present periphery becomes a center of production, then the existing core will become the new periphery. Therefore, the structural problem will remain. Given these circumstances, not only for the western area but also for the rest of China, a careful consideration of what industries should be invested in to produce linkages is important to avoid wasteful investments. Nevertheless, the development strategy of the 10th FYP is to promote characteristics for pillar industries in the western area, such as agriculture, food processing, tourism, Chinese medicine and so on. To withhold judgment on the 10th FYP, including the western development, in order to focus on particular features seems to be an appropriate development strategy in our context.27 With regard to Xi’an, Shaanxi, for example, characteristics typical of industries such as electronics and telecommunications, aviation and so on, were noted in the 10th FYP for that area. According to Uno (1999), although the economic structure in Xi’an was meager before the foundation of modern China, many large-scale companies were established, and ﬁrms and institutions related to national defense moved there during the ﬁrst FYP (1953–57), the second FYP (1958–62), and the Construction of the Third Front. Massive ﬁnancial help was given to defray the expense of construction costs in the non-industrial city. As a result, Xi’an has gained a large foothold in the machinery, spinning and weaving, and war industries in the northwestern area. To further raise this status, the industries mentioned above have been reinforced by making the most of beneﬁts from past investments, in other words, utilizing the history before the economic reform. Conclusion In this section, the western development was seen as a peripheral area policy. After introducing the western development, we discussed the diﬃculties of attaining the big push in the context of structural results, and the core–periphery structure. However, the possibility of utilizing the char-

China’s regional industrial disparity

197

acteristics of the western area was discussed, in order to encourage development in spite of the structural problem.

5.

CONCLUDING REMARKS

In this chapter, we examined regional industrial disparity from the viewpoint of industrial agglomeration. We shall summarize the main points again. The concentration with the eastern area as the center became apparent in the 1990s after dispersions during two periods. One of the factors to explain the disparity was the idea of industrial agglomeration. From the share of the machinery industry in the mining and manufacturing industries and intermediate goods transaction, the disparity was inevitable, based on the working of linkages. A policy implication and a counter-plan were evident in the core–periphery structure between the agglomerated area and the other areas. To develop the inland area, the importance of promoting characteristics for the pillar industries was introduced. These results lead us to the conclusion that the agglomeration eﬀect in the eastern area is one of the factors that account for China’s regional industrial disparity, and that western development has to develop their own pillar industries based on their regional comparative advantages. But some issues still remain to be solved. First, transportation cost, which is one of the key issues in deciding whether to agglomerate was not dealt with here. Second, although there are bigger GIOV gaps than those of population among the three areas as discussed in Section 2, we need to normalize the industrial value of each province by, for example, population or industrial structure, in order to analyse the situation more precisely. Third, this chapter tried to explain Chinese agglomeration statistically, however, the empirical conclusions still remain. Finally, the lower levels of administrative division should also be analysed. Here, we discussed only the upper echelons: the provincial and the three area levels.

NOTES *

1.

I would like to thank Dr Yumiko Okamto (Doshisha University) for her comments on a workshop held by IDE, Dr Tetsushi Sonobe (Foundation for Advanced Studies on International Development) and Dr Koji Nishikimi (IDE-JETRO) for their advice, Dr Naoki Murakami (Nihon University) and Mr Makato Muto (Nippon Hyoronsha) for their permission on copyright, Mr John Gallagher for his English proofreading and editing work, Ms Kie Ono for her administrative support and Mr Shigeru Togashi for his editing work. Although the role of foreign direct investment is important in the industrial agglomeration in China, here we concentrated on China’s domestic distribution in accordance with macroeconomic data and some facts.

198 2. 3. 4.

5. 6. 7. 8. 9. 10.

11. 12.

13.

14. 15.

16.

Agglomeration in Asia This is almost the same as the deﬁnition by Marukawa (2001), but he supposes increasing returns to scale at the area, rather than the ﬁrm level. See Marukawa (2001) for a detailed investigation of local industries. There are 22 provinces, four municipalities and ﬁve autonomous regions at present. Hainan was made a separate province from Guangdong in 1988. Chongqing became a separate municipality from Sichuan in 1997. We exclude Hong Kong Special Administrative Region, Macao Special Administrative Region and Taiwan. Although the number of ﬁrms is important to the concept of agglomeration, we shall take up the gross industrial output value here. The criterion of ﬁrms was changed after 1998. Because of this, the share of provinces that have comparatively more large-scale enterprises will be overestimated. See, in particular, Kojima (1997) and Kato (2000). In addition, they are relocated to the least-developed areas as a means of ﬁnance assistance. See, for example, Marukawa (1993, 2002). We shall not assess the policy, but it can be seen as a priority investment to the inland area for reasons of national defense. These three economic zones were adopted in the 7th FYP (1986–90). The eastern area includes 12 province-class divisions: Liaoning, Beijing, Tianjin, Hebei, Shandong, Shanghai, Jiangsu, Zhejiang, Fujian, Guangdong, Hainan and Guangxi. The central area includes nine province-class divisions: Jilin, Heilongjiang, Inner Mongolia, Shanxi, Henan, Anhui, Hubei, Hunan and Jiangxi. The western area includes 10 province-class divisions: Xinjiang, Tibet, Qinghai, Gansu, Ningxia, Shaanxi, Sichuan, Chongqing (at that time part of Sichuan), Yunnan and Guizhou. According to the National Bureau of Statistics of China (2002), in 2001 the GDP in the areas was as follows: eastern was 59.6 per cent; central 26.9 per cent; and western 13.6 per cent (current prices). Sonobe and Otsuka (2006) analysed Taiwan’s industrial location. It is not easy to conclude whether transportation costs are high or low. According to a personal interview (September 10, 2002, in Tianjin) at a Japanese food-processing Company B (established in 1997) producing bean paste from azuki beans in Tianjin, it was claimed that transportation costs had decreased. At ﬁrst, the ﬁrm located there to purchase high-quality beans, but after 1995 the supply decreased because of a bad harvest; since then beans have been purchased from the northeastern area, speciﬁcally from Heilongjiang. But, there is no incentive to move closer to raw materials with the aim of reducing the transportation cost (the raw materials share of total cost is over 60 per cent.). In addition, with regard to weight transported (Okamoto 2002), a dataset concluded that the weight of inter-region (east, central and west) transportation increased from 1987 to 1997. In addition, there are data on the development of transportation infrastructure and increases in weight. As mentioned above, there are some facts to support the development of transportation in China. However, it is diﬃcult to conclude that transportation cost has increased or decreased without an investigation of the bottleneck discussed by Uchida (1988). There are reasons to set up development zones for every level of government. For example (a personal interview, September 2–6, 2002, in Ningbo and Cixi at county level), Cixi in Ningbo city, Zhejiang, set up some zones to realize scale economies. Ningbo in particular, where there were a lot of small factories, has developed a cluster economy. Other reasons include environmental conservation and investment selection for speciﬁc industries. On the relation between foreign capital and development in the eastern area, see Ishihara (1998). Kitakyushu City (2002) found, from the viewpoint of industrial agglomeration, that the problem of industrial structure weighted on the basic material industry in the city, compared with Higashi-Osaka City which was weighted on the processing and assembly industry; the problem is recognized in the Kita-Kyushu industrial area generally. In this context of a network of suppliers, see, for example, Marukawa (1994) and Ohara (2001).

China’s regional industrial disparity 17. 18.

19.

20. 21.

22.

23.

24. 25. 26. 27.

199

It is necessary to examine economic activities based on a number of ﬁrms to make a valid comparison between these characteristics of the machinery industry and the basic materials industry. The arithmetical mean of the industrial disparity index including the mining and quarrying industry was 0.6079 in 2000. Since locations for mining and high production provinces correspond, the mining and quarrying industry is a typical type located by available resources. In particular, the high index of the extraction of petroleum and natural gas industry suggests that the large-scale oilﬁelds of Daqin, Shengli, and Liaohe have high shares. This treatment may not be natural, because, if we applied Krugman’s (1991) deﬁnition of industrial agglomeration, all types of manufacturing have the importance of variety. In addition, because all manufacturing is like this, the machinery industry would tend to agglomerate more in the same area as other manufacturing agglomerations. But, we focus on the machinery industry to distinguish it from the others. According to Fujita et al. (1999), if the condition is that there are a lot of (more than two) places set up, there is the possibility of agglomeration in some places. Since data from the same period as in Table 7.3 are not available, data after 1994 are used here to show increasing ratios of external purchases. The period from 1994 to 2000 was divided into two periods, because the criterion of ﬁrms was changed after 1998 as noted in Figure 7.1. The percentages are calculated from Table 7.2 and the Wuhan Statistical Yearbook, compiled by the Wuhan Statistical Bureau. There is a special case that illustrates the situation of gathering around a core assembler such as the Beijing Economic–Technological Development Area, according to a personal interview (September 9, 2002 in Beijing). The development area has some jointventure own zones, where there is a concentration of intermediate goods companies. Export (import) share of machinery in total export (import) was 38.6 (48.1) in 2001 (the machinery industry includes 84 to 93 in the HS code). Furthermore, ordinary machinery (84) and electrical machinery (85) are ranked ﬁrst and second, respectively. Therefore, the share of machinery in China’s foreign trade is high. Although the western area includes 10 province-class divisions as mentioned above, there is scope in the development policy for 12 province-class divisions, including Guangxi in the eastern area and Inner Mongolia in the central area. Murphy et al. (1989) formulized the general idea. Fujita and Thisse (2002) examined the development of a periphery area under some speciﬁc conditions. In addition, large infrastructure projects are essential in the 10th FYP. Note that infrastructure is important to the development of industries, however, there is the possibility that the core–periphery structure could develop further if the transportation cost between the core and periphery areas was reduced.

REFERENCES Amiti, Mary and Mei Wen (2001), ‘Spatial distribution of manufacturing in China’, in Peter Lloyd and Zhang Xiao-guang (eds), Models of the Chinese Economy, Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 271–89. Batisse, Cécile (2002), ‘Dynamic externalities and local growth: a panel data analysis applied to Chinese provinces’, China Economic Review, 13 (2–3), 231–51. Cai, Fang and Du Yang (2000), ‘Convergence and divergence of regional economic growth in China’ (in Chinese), Economic Research Journal, No.10, 30–37. DITSNBS (Department of Industrial and Transport Statistics of National Bureau of Statistics) (1996), China Industrial Economic Statistical Yearbook 1995, Beijing: China Statistical Press.

200

Agglomeration in Asia

DITSNBS (2001), China Industrial Economic Statistical Yearbook 2001, Beijing: China Statistical Press. Dixit, Avinash K. and Joseph E. Stiglitz (1977), ‘Monopolistic competition and optimum product diversity’, American Economic Review, 67 (3), 297–308. Ellison, Glenn and Edward L. Glaeser (1997), ‘Geographic concentration in U.S. manufacturing industries: a dartboard approach’, Journal of Political Economy, 105 (5), 889–927. Fujita, Masahisa and Nobuaki Hamaguchi (2001), ‘Intermediate goods and the spatial structure of an economy’, Regional Science and Urban Economics, 31 (1), 79–109. Fujita, Masahisa and Dapeng Hu (2001), ‘Regional disparity in China 1985–1994: the eﬀects of globalization and economic liberalization’, Annals of Regional Science, 35 (1), 3–37. Fujita, Masahisa, Paul Krugman and Anthony J. Venables (1999), The Spatial Economy, Cambridge, MA: MIT Press. Fujita, Masahisa and Jacques-François Thisse (2002), Economics of Agglomeration, Cambridge, UK and New York: Cambridge University Press. Hanson, Gordon H. (1998), ‘Market potential, increasing returns, and geographic concentration’, NEBR Working Paper No. 6429. Ishihara, Kyoichi (ed.) (1998), The Chinese Economy and Foreign Capital (in Japanese), Tokyo: Institute of Developing Economies. Jin, Xiangrong and Zhu Xiwei (2002), ‘The origin and evolution of specialized industrial districts: a historical and theoretical approach’ (in Chinese), Economic Research Journal No. 8, 74–82. Kato, Hiroyuki (1997), Economic Development and Marketization in China (in Japanese), Nagoya: University of Nagoya Press. Kato, Hiroyuki (2000), ‘Integration of domestic market and regional development in China’ (in Japanese), in Katsuji Nakagane (ed.), Structural Change in Contemporary China, Vol. 2, Tokyo: University of Tokyo Press, pp. 107–30. Kim, Sukkoo (1995), ‘Expansion of markets and the geographic distribution of economic activities: the trends in U.S. regional manufacturing structure, 1860–1987’, Quarterly Journal of Economics, 110 (4), 881–908. Kitakyushu City (2002), Industry Report of Kitakyushu City 2001 (in Japanese), Kitakyushu: Kitakyushu Urban Association. Kojima, Reeitsu (1997), Contemporary China’s Economy (in Japanese), Tokyo: Iwanami Shoten. Krugman, Paul (1991), ‘Increasing returns and economic geography’, Journal of Political Economy, 99 (3), 483–99. Kuroda, Atsuo (2001), ‘China’ (in Japanese), in Toyojiro Maruya and Koichi Ishikawa (eds), Impacts of Made in China, Tokyo: JETRO, pp. 17–34. Marukawa, Tomoo (1993), ‘The “Third Front construction” in China (I) (II)’ (in Japanese), Ajia Keizai, 34 (2), 61–80; 34 (3), 76–88. Marukawa, Tomoo (1994), ‘The formation of inter-ﬁrm relations in China: the case of the automobile industry’ (in Japanese), Ajia Keizai, 94 (10), 2–32. Marukawa, Tomoo (2001), ‘Industrial agglomeration in China’ (in Japanese), in Mitsuhiro Seki (ed.), Industrial Agglomeration in Asia, Chiba: Institute of Developing Economies (JETRO), pp. 29–61. Marukawa, Tomoo (2002), ‘The “third front” of China revisited’ (in Japanese), Ajia Keizai, 43 (12), 67–80.

China’s regional industrial disparity

201

Mori, Tomoya and Koji Nishikimi (2002), ‘Economics of transport density and industrial agglomeration’, Regional Science and Urban Economics, 32 (2), 167–200. Murakami, Naoki (1999), ‘Industrial organization and relations among ﬁrms’ (in Japanese), in Ryoshin Minami and Fumio Makino (eds), Ordeals of being a Big Power, Tokyo: Nippon Hyoronsha, pp. 123–39. Murphy, Kevin M., Andrei Shleifer and Robert W. Vishny (1989), ‘Industrialization and the Big Push’, Journal of Political Economy, 97 (5), 1003–26. National Bureau of Statistics of China (2002), China Statistical Yearbook 2002, Beijing: China Statistics Press. Ohara, Moriki (2001), ‘The supplier system of the Chinese motorcycle industry: a comparative study with the Japanese system in view of the mechanisms of risk management and capability upgrading’ (in Japanese), Ajia Keizai, 42 (4), 2–38. Okamoto, Nobuhiro (ed.) (2002), China’s Inter-Regional Industrial Structure, Vol. 1 (in Japanese), Chiba: Institute of Developing Economies (JETRO). Onoe, Etsuzo (1971), A Study on Industrial Location in China (in Japanese), Tokyo: Institute of Developing Economies. Qiu, Baoxing (1999), A Study of Agglomerations of Small-Sized Enterprises (in Chinese), Shanghai: Fudan University Press. Rosenstein-Rodan, Paul. N. (1943), ‘Problems of industrialization of Eastern and South-Eastern Europe’, Economic Journal, 53 (June–September), 202–11. Seki, Mitsuhiro and Masaki Nishizawa (2000), China’s Challenging Inland Industry (in Japanese), Tokyo: Shinhyoron. Sonobe, Tetsushi and Keijiro Otsuka (2006), ‘The division of labor and the formation of industrial clusters in Taiwan’, Review of Development Economics, 10 (1), 71–86. Tuan, Chyau and Linda F.Y. Ng (2001), ‘Regional division of labor from agglomeration economies’ perspective: some evidence’, Journal of Asian Economics, 12 (1), 65–85. Uchida, Tomoyuki (1988), ‘Reforms of traﬃc and transport control institutions’ (in Japanese), in Reeitsu Kojima (ed.), China’s Economic Reforms, Tokyo: Keisoshobo, pp. 111–41. Uno, Kazuo (1999), ‘Xi’an’ (in Japanese), Contemporary Chinese Dictionary, Tokyo: Iwanami Shoten. Wei, Houkai (2002), ‘Concentration status quo of manufacturing and international comparison in China’ (in Chinese), China Industrial Economy, 166 (1), 41–9. Wuhan Statistical Bureau (various years), Wuhan Statistical Yearbook, Beijing: China Statistics Press. Xu, Xinpeng (2002), ‘Have the Chinese provinces become integrated under reform?’, China Economic Review, 13 (2–3), 116–33. Yue, Ximing (2000), ‘An empirical analysis of industrial location decision in Japanese prefectures’ (in Japanese), JCER (The Japan Center for Economic Research) Economic Journal, 41 (September), 92–109.

PART II

Agglomeration in Italy

8. Italian comparative advantages, persistence and change in overall specialization Luca De Benedictis* 1.

INTRODUCTION

The structure of Italian exports, its characteristics, its change over time, its causes and implications in terms of income and growth, employment and fragility with respect to shocks has been an issue on which at least three generations of Italian economists have debated and even strongly disagreed.1 If we have to specify one single motivation at the origin of this articulated debate it would be the perplexity associated with the evidence of Italy being a large industrialized country and of having an export structure more similar to that of a newly industrialized country, such as Taiwan or Thailand, than of its fellow members of the G7. The anomaly in the trade structure is what makes Italy a case in the international trading system. This chapter explores the structure of Italian exports making use of the Revealed Comparative Advantage Index associated with the name of Bela Balassa, focusing on the export structure itself, on its change over time and on its degree of persistence. The analysis is developed with the use of nonparametric statistical techniques that allow us to estimate the empirical distribution of the Balassa Index and to track its dynamic change during three decades, from the 1970s to the present. The persistence in the pattern of comparative advantage is then related to four interacting theoretical explanations based on the role of dynamic scale economies, Marshallian externalities, quality and vertical diﬀerentiation, and the absence of market incentives. From an empirical viewpoint the persistence in the pattern of comparative advantage is subsequently conditioned to the presence of industrial districts, in order to show whether it is the local structure of production that originates such a relevant degree of persistence. The empirical strategy followed in the chapter is largely dependent on data availability. The ideal dataset consisting of a panel of highly spatially and sectorally disaggregated data is unfortunately unavailable; we shall 205

206

Agglomeration in Italy

therefore move from one dataset to the other when a speciﬁc question requires a speciﬁc level of data disaggregation. More properly, we shall use the new OECD Stan database (OECD 2002) when an international comparison is required but the sectoral disaggregation should be around 20–30 sectors in order to make the visual representation meaningful; we shall use the World Bank TradeCAN database (ECLAC–World Bank 1999) when we require a higher level of sectoral disaggregation; and ﬁnally we shall use the Italian ISTAT dataset (ISTAT 2002) when we require in the analysis of Italian exports a higher level of spatial disaggregation. More detailed explanation on the data used will be given below; for the moment let us start with a necessary preliminary step regarding measurement issues.

2.

MEASURING COMPARATIVE ADVANTAGES

Among the metrics used in the analysis of bilateral and multilateral trade ﬂows,2 the ﬁrst and still most widely used measure built on one single information variable3 is the Balassa (1965) Index of Revealed Comparative Advantage (RCA). The traditional way of deﬁning the Balassa Index – which we shall indicate synthetically as b – is: Xij Xwj bij X X i w

(8.1)

where – for every time period t considered – i denotes a speciﬁc country, w indicates the world economy (that is, the entire set of countries considered in the analysis), and j is a speciﬁc sector. b is, therefore, a sectoral relative export measure in terms of share of world exports. Since the numerator (cXij /Xwj) ranges from 0 (the country is not exporting products belonging to that particular sector) to 1 (the country is an international monopolist in such a category of products), and the denominator (1b Xi Xw ) – which is the relative economic dimension of the country, in export terms – also ranges from 0 to 1, then b ranges between 0 and b. We can therefore write equation (8.1) – neglecting subscripts – as: b c.b

(8.2)

where c is the sectoral market share (competitiveness) of country i in sector j, and b is the time-variant upper bound of b (for all sectors j ). From equation (8.2) we can say that when b [0,1] (which is equivalent to saying that c 1b) the country has a comparative disadvantage in sector j; while it has a comparative advantage in sector j if b[1, b] (when c 1b). The

Italian comparative advantages

207

demarcation value is given by the condition c 1b, corresponding to the case where the country displays a sectoral market share equal to its total share of world exports. For every country, the distribution of b is characterized by a ﬁxed lower bound (0), a time-variant upper bound (b), and an invariant demarcation value (1). De Benedictis and Tamberi (2004) give a detailed description of the characteristics of the distribution of b, as far as the necessity of the present contest is concerned, it is suﬃcient to remark the asymmetry in the distribution, and the inverse relation between the degree of asymmetry and the country’s share of total world exports, 1/b (that is, small countries tend to be characterized by high bs). We shall now use b to discriminate between countries that reveal a comparative advantage in a particular sector and countries that do not, and we shall then order countries according to the speciﬁc value of the b.

3.

SCATTER PLOTS

In Figure 8.1 we have collected six scatter plots for Italy, Spain, Japan, Germany, the US and the UK in order to compare trade structures and trade patterns internationally. We shall take the ﬁrst plot on the upper left (Italy) as an example, since the discussion extends to the other ﬁve plots for analogy. On the horizontal axis we have the sectoral b values for 1970, and on the vertical axis we have the corresponding b values for 1998.4 The two lines drawn in correspondence of the demarcation value b1 separate sectors with revealed comparative disadvantage from sectors with revealed comparative advantage, and deﬁne four quadrants. The two quadrants along the main diagonal contain sectors that modiﬁed their relative position, from comparative disadvantage to comparative advantage (upper quadrant to the left) or vice versa (lower quadrant to the right). The two quadrants along the secondary diagonal contain sectors that did not modify their position in terms of comparative advantage (upper quadrant to the right) or comparative disadvantage (lower quadrant to the left), even if relative changes are evident inside each one of the four quadrants. The 45 dotted line visualizes a condition of pure persistence.5 The data used are collected by the OECD (2002), at two-digit levels of the SITC (rev.2) classiﬁcation; the sectors included are 27 manufacturing sectors6 that we have roughly divided into traditional sectors, advanced sectors and other, respectively indicated in Figure 8.1 by the symbols, (), () and (). Both the choice of the level of sectoral disaggregation and the speciﬁc clustering are for expositional and visual reasons.7

0.0

0.0

0

0.5

0.5

2

1.5

2.0

1.5

2.0

2.5

6

3.0

traditional sectors

USA 1970

1.0

Japan 1970

1.0

4 Italy 1970

advanced sectors

0.2

0

4

0.8

1.0

Spain 1970

UK 1970

1.5

1.2

6

2.0

Germany 1970

1.0

0.6

others

0.5

0.4

2

2.5

1.4

8

3.0

Comparative advantage, 1970–1998: Italy, Spain, Japan, Germany, the US and the UK

OECD (2002).

Figure 8.1

Source:

4

2

0

Italy 1998

Japan 1998

USA 1998

1.0

0.0

2.5

1.5

0.5

Spain 1998 Germany 1998 UK 1998

3.5 2.0 0.5 1.2 0.8 0.4 1.5 0.5

208

Italian comparative advantages

209

The information content of Figure 8.1 is largely self-explanatory. Starting from Italy it is evident that it had, and still has, very strong RCAs in traditional sectors such as Textiles, Wearing Apparel, Leather, Footwear, Furniture, and Pottery and China (note that the scale of the axis is very large with respect to the other countries considered in Figure 8.1), it has improved its RCA in many traditional sectors (which lie above the 45 dotted line) but not in the top ones (Footwear, and Wearing Apparel), and it has also improved its RCA in some other sectors such as Machinery and Other Manufacturing. What is also remarkable is the persistence in the structure of comparative advantages: almost all sectors are in the two quadrants along the secondary diagonal. The only other country that shows such a high persistence (a low dispersion around the 45 line) is the US. We shall come back to this point later. The opposite case is represented by Spain. From the 1970s to the 1990s Spain changed markedly its pattern of comparative advantage. Many sectors with comparative advantage – traditional sectors but not only – lie below the 45º line (with the noteworthy exception of Non-Metallic Products), while some sectors – both traditional and advanced – moved from a condition of comparative disadvantage to one of comparative advantage. Spain, which for many reasons can be considered similar to Italy, followed a very diﬀerent path during the period. On the other hand, Japan and Germany are somewhat similar cases. Both have in common an export structure characterized by a limited number of sectors with comparative advantages – both advanced sectors (NonElectrical Machinery, Transport Equipment, and Professional Goods) and other sectors (Rubber Products); in both cases a conspicuous number of sectors moved toward a condition of revealed comparative disadvantage; both are characterized by a high dispersion around the 45º line. The US instead follows a fairly stable path. The large majority of sectors lie close to the 45º line; as for Japan and Germany the value of the bs is quite small with only one exception being Tobacco, with a b around 2; ﬁnally, the sectors characterized by RCA are mainly advanced sectors. As for the UK, it follows a fairly stable path, with the peculiarity of a relevant number of traditional and other sectors – such as Leather and Products, Footwear, Tobacco, Rubber Products, Non-Ferrous Metals, Metal Products, and Transport Equipment – losing their original comparative advantage. After this bird’s-eye inspection of comparative advantages in six industrialized countries we can go back to Italy and re-examine in the light of our ﬁndings the point around which the debate on the Italian pattern of comparative advantages was developed: the persistence of the Italian structure of comparative advantage in the traditional sectors.

210

Agglomeration in Italy

First, we have seen that this view is only partially true. In Figure 8.1 it was evident that the top RCA traditional sectors such as Footwear and Wearing Apparel decreased their b values from 1970 to 1998, while other sectors such as machinery have increased their b values. It should be manifest that the Italian structure of comparative advantages is by now characterized by two diﬀerent – although interconnected – components: the traditional sectors and the machines (mainly used in the production of traditional products and now exported). This point will be discussed further when we discuss the shape of the estimated probability density function of Italian bs. Second, so as to properly verify the persistence of the Italian structure of comparative advantages, we have to be sure that the persistence is not apparent, since the change in the structure is hidden by its intra-sectoral nature; rephrasing it, the changes in the traditional sectors along, let’s say, the quality ladder, cannot result in the data as the level of aggregation is not very reﬁned. To exclude this eventuality we have to switch to a diﬀerent dataset, which should make it possible to deﬁne the b values (it should contain information on all countries, or at least the aggregate of world exports) at the higher possible disaggregation. We have chosen the ECLAC–World Bank (2000) dataset, which includes data on 97 countries, covering, according to the collectors ‘well over 90% of world trade’, and which at 4 digits of the SITC (revision 2) classiﬁcation includes 540 manufacturing sectors (and more than one thousand overall sectors).8 From such a dataset we can obtain very detailed information on sectoral RCAs, but we have to pay the cost of losing insights on the ﬁrst decade of our previous analysis, since the ECLAC dataset covers a time span of 14 years, from 1985 to 1998. It is not proﬁtable to go through the description of every speciﬁc sectoral b, since what matters in the present context is the possibility of correctly verifying the persistence of the Italian structure of comparative advantages. One possibility would be to test the AR(1) characteristic of the various sectoral time series (de Nardis 1997), but the probability of having diﬀerent series with diﬀerent autoregressive structures is much more than strictly positive, and that would not help to answer our question concerning the persistence of Italian overall structure of sectoral comparative advantages. We would then rely on a simple visual description of Euclidean distance between the sectoral bs of 1985 and 1998. In Figure 8.2 we replicated the scatter plot for Italy contained in Figure 8.1 using the ECLAC–World Bank dataset. The years considered are 1986 and 1998. Since the data9 showed – not surprisingly – a strong right skewness we shall examine it on a log scale, so as to emphasize both comparative advantages (log(b)0) and disadvantages (log(b) 0).

211

–4

–2

log(ita98)

0

2

Italian comparative advantages

–6

–4

–2

0

2

log(ita85)

Note: The dotted line is the pure persistence 45 line; the continuous line is the OLS regression. Source: ECLAC–World Bank (2000).

Figure 8.2

Persistence of Italian comparative advantages, 1985–1998

Figure 8.2 gives an indication that partially corrects the one contained in Figure 8.1. The large majority of sectors are still located in the two quadrants along the secondary diagonal containing sectors that did not modify their relative condition in terms of comparative advantage or disadvantage, but what shows up – and that was unobservable with highly aggregated data – is that many sectors that were comparatively disadvantaged became, in less than 15 years, comparatively advantaged, moving from the lower quadrant along the secondary diagonal to the upper quadrant along the main diagonal. Two-thirds of sectors with RCA in 1986 lie above the 45º line, indicating an increase in b, while the remaining one-third decreased their b value. One further indication is that sectors with revealed comparative disadvantage are much more dispersed around the 45º line than sectors with comparative advantages.

212

Agglomeration in Italy

Finally, the resulting degree of persistence of the Italian structure of comparative advantages is still high even when using highly disaggregated data. To show this, just observe in Figure 8.2 the distance between the dotted 45º line and the continuous line corresponding to the linear OLS regression. In case of pure persistence the two lines would overlap, while they would be perpendicular to each other in the case of complete structural change. In Figure 8.2 the two lines are very close to each other, implying a high degree of persistence in mean terms.10 Since what Figure 8.2 seems to indicate is both persistence and change in the distribution of RCA we shall now move to the estimation of such density.

4.

KERNEL DENSITY

In this section we shall give some emphasis to the ordinal content of the Balassa Index of RCA and no emphasis at all to the sectoral characteristics of the structure of the Italian RCA. This is not because they are not important but because we want to concentrate on the shape and the dynamics of the overall structure of the Italian RCA, leaving aside the intra-distributional changes that we have previously observed. Using nonparametric statistics (Lehmann 1975), we shall estimate the probability density function (PDF) of the Italian sectoral bs. As for the scatter plot in Figure 8.2, given the skewness of the bs we shall carry out the analysis on a log scale. We shall use a kernel density estimate,11 adopting a Gaussian kernel function and a normal optimal smoothing parameter, as suggested by Silverman (1986) and Bowman and Azzalini (1997). Since the aim is to compare PDFs along time, we shall keep the bandwidth of the PDF (the smoothing parameter) constant: we shall estimate it for 1985 and apply it to the following years. The results are set out in Figure 8.3. The vertical line in Figure 8.3 indicates the ﬁxed demarcation value, log (1); if we take such value as a focus point, we can observe that the distribution is bell-shaped but still skewed, even on a log scale, and that the density is higher around the demarcation value (in the interval 0.3 b3), conﬁrming the information of Figure 8.2. Starting from the estimated PDF for 1986, the shape of the distribution is relatively centered and is characterized by very high b values in the right tail and by a remarkable bimodality. The principal mode indicates that in 1986 more than 50 per cent of the sectors had a b value below the demarcation value, while the second mode to the right of the distribution indicates that a conspicuous number of sectors were characterized by high b values. The estimated PDF for 1998 shows a noteworthy evolution. The major change is that the overall distribution has moved to the right, with a small

213

Italian comparative advantages Italy

0.3 0.2 0.1 0.0

Estimated PDF

0.4

1985 1988

log log log log log log (0.0025) (0.0067) (0.0183) (0.0498) (0.1353) (0.3679)

log(1)

log log log (2.7183) (7.3891)(20.0855)

Manufactures

Source: ECLAC-World Bank (2000).

Figure 8.3

Kernel density of Italian comparative advantage, 1985–1998

increase in the number of sectors with very low b values and with a comprehensive increase in the bs. A further major change is that the bimodality has disappeared, with sectors following a pattern of convergence toward the mode. Finally, the mode itself has moved to the right and is now to the right of the demarcation value: more than 50 per cent of the sectors show a revealed comparative advantage. In spite of what we said regarding the persistence of the Italian structure of comparative advantage, the estimated PDFs show a certain degree of mobility. What should we conclude? The contradiction is only apparent since we did not test the proximity of the regression line to the 45º line drawn in Figure 8.2. In other words, can we say that the evident persistence in the Italian structure of comparative advantages implies the immobility of the overall structure? The answer is: no. In order to evaluate the statistical signiﬁcance of the mobility in the estimated distribution, from 1986 and 1998, a two-tailed Wilcoxon’s signed rank test was performed. This test was chosen instead of a more traditional t-test because it did not require the assumption of normally distributed data.12 The null hypothesis was the absence of diﬀerence between the 1985 and the 1998 series. At a level of conﬁdence of p 0.05, the test strongly rejected the null hypothesis. The characteristics of the structure and of the dynamics of Italian comparative advantages evidenced so far can be summarized as follows. The Italian anomaly is related to the sharp diﬀerence in terms of export structure with respect to other industrialized countries: Italy shows a strong

214

Agglomeration in Italy

RCA in traditional sectors and in recent years in machines. The intensity of such RCA is very high compared to other OECD countries, so that the distribution of the b values is markedly skewed. Some changes are occurring, however: top bs are decreasing and Machines and other sectors comparatively disadvantaged in 1985 are now becoming more relevant; the overall distribution is moving to the right, the mode is unique (in 1998) and is above 1; persistence is high but this does not imply immobility in the structure of comparative advantages.

5.

SOME THEORETICAL EXPLANATIONS

From a theoretical viewpoint there are at least three main potential explanations of the actual structure of Italian comparative advantages: (a) the factor proportion theory of comparative advantages, associated with the predictions of the Heckscher–Ohlin theorem; (b) the theory of dynamic scale economies, Marshallian externalities and agglomeration, related both to the new economic geography and to the literature on industrial districts; and (c) a theory of vertical diﬀerentiation, quality ladders and the absence of market incentives. Even if the three approaches are not mutually incompatible it is better to keep them separate in order to highlight the key elements of each speciﬁc approach. Starting from the factor proportion theory, the Heckscher–Ohlin theorem predicts that Italy would specialize in traditional sectors – producing and exporting labor-intensive goods – if the country is a labor-intensive country. But, is Italy a labor-intensive country? A positive answer would be empirically supported only for the period immediately after the Second World War. From the mid-1950s, Italy started a process of capital accumulation that brought the country almost to the same average level of capital–labor ratio of other large industrialized countries. By now, according to the Heckscher–Ohlin theorem, Italy should export capital-intensive goods to the rest of the world. Is the factor proportion theory completely misleading? It is diﬃcult to give a clear-cut answer. We prefer a less neat but more articulated response. We have seen that the Italian structure of comparative advantages is not homogeneous: Italy is exporting both traditional goods and machines, so it is exporting both labor- and capital-intensive goods. How can this be reconciled with the factor proportion theory? The answer is that Italy is exporting traditional goods to capital-intensive countries and machines to labor-intensive countries. Thus a geographically diﬀerentiated export structure is the explanation for the Italian structure of comparative advantages in terms of factor proportions. But what the factor proportion theory cannot explain is the high degree of persistence in the very structure.

Italian comparative advantages

215

If persistence is the issue, then dynamic economies of scale is the element to look at. De Benedictis and Padoan (1999) and Epifani (1999) oﬀer diﬀerent but converging interpretations of the persistence of the Italian structure of comparative advantages; both emphasize the role played by dynamic economies of scale. In a Ricardian model (De Benedictis and Padoan) or in a Heckscher–Ohlin model (Epifani) with no impediment to trade, a country specializes according to comparative advantages. With dynamic economies of scale the country will become more and more eﬃcient in the production of the goods it was initially exporting, so that learning by doing will freeze the country at its initial structure of comparative advantages. If dynamic economies of scale are quite eﬀective they can nullify the eﬀect of a change in factor proportions and the export structure would remain the same. In this case Italy is still exporting traditional goods not because it is a labor-intensive country but because it was a labor-intensive country. But why does Italy have such strong dynamic economies of scale? The answer is the diﬀusion of industrial districts in the late 1950s, associated with Marshallian externalities and agglomeration forces. A further question is why are the industrial districts producing traditional goods still able to compete with labor-intensive countries? Is there a diﬀerent element that together with dynamic economies of scale explains the persistence of the Italian structure of comparative advantages? This time the key words are ‘quality’ and ‘incentives’ (de Nardis and Traù 1999; De Benedictis and Tamberi 2000). Italy is not really (not yet, at least) competing with the newly industrialized countries since the quality of Italian traditional products is higher than that of China or Singapore. Labor cost corrected for quality still gives Italy an advantage in traditional sectors. On the other hand, a very uniform distribution of proﬁts across manufacturing sectors does not generate an incentive to shift from traditional production to advanced sectors. In the next section we shall explore the possibility of empirically supporting the second potential explanation of the actual structure of Italian comparative advantages, and, in particular we shall explicitly consider the link between the persistence in the structure of Italian comparative advantages and the presence of industrial districts at a provincial level of disaggregation.

6.

INDUSTRIAL DISTRICTS AND PERSISTENCE

The analysis of the role of the industrial districts in shaping the structure of the Italian economy has a long and fruitful tradition13 that has recently been enriched by the collection and elaboration of data by the Italian National Statistical Institute (ISTAT 1997). We shall make use of these data,14 jointly

216

Agglomeration in Italy

examining the export structure of the 104 Italian provinces and the presence of industrial districts in each province. Also in this case the necessity to expand one dimension of the dataset (the space dimension) is possible only at the cost of shrinking a diﬀerent dimension (the time dimension). The time span of the data used covers only 10 years, from 1991 to 2001, while the sectoral aggregation allows us to consider the 35 manufacturing sectors15 of the SITC (rev.2) classiﬁcation (ISTAT 2002). The data on industrial districts (ISTAT 1997) identify 616 local labor systems, which in our dataset – excluding the food districts and counting all the districts in each province as one singular piece of information – reduces to 259 districts specialized in the production and export of Paper; Chemicals; Machinery; Miscellaneous Manufactured Articles (Toys, Musical Instruments and Jewellery); Leather and Footwear; Furniture and Non-Metallic Mineral Manufactures; Textiles, Apparel and Clothing. Noteworthy, in some sectors where the scale of production is a relevant factor, the presence of industrial districts is null or very limited (Chemicals, Iron and Steel, Road Vehicles); the majority of the districts are concentrated in the traditional sectors (Leather, Textiles, Furniture, Apparel and Clothing, Footwear, Miscellaneous Manufactured Articles) and in sectors producing machines (Power-generating Machinery, Specialized Machinery, Metal-working Machinery) and is localized in the northeast and central Italian provinces; the number of districts in the South of Italy is very limited. The information on the presence of sectoral industrial districts in the 104 Italian provinces can be summarized in a dummy matrix that – in order to connect the districts to the Italian export structure and to its persistence – can be related to the RCA index described in equation (8.1) in the following way. Let us deﬁne the sectoral export share of each province relative to the world sectoral export share as: Xijp Xwj ijp X X , ip w

(8.3)

where the p subscript stands for ‘province’, and Xijp Xwj ijp X X , ip w

(8.4)

which identiﬁes the sectoral export share of each province relative to the national sectoral export share. We can now redeﬁne the RCA index described in equation (8.1) as: · c · b,

(8.5)

Italian comparative advantages

217

showing that the sectoral RCA of each province ( ) is positively related to the national RCA index (b) and to the sectoral export share of each province and negatively related to the national sectoral export share (). From the identity equation (8.5) we can also derive the following expression: b

Xij

· Xj ,

(8.6)

which indicates that the national sectoral RCA index is the weighted sum of the sectoral RCA of each province. In Figure 8.4 we have used the log transformation of identity equation (8.6) in the case of the footwear sector. Having the 1991 values expressed in logs on the horizontal axis and the 2001 values also expressed in logs on the vertical axis, the spots identify the sectoral RCA of the 104 Italian provinces: the grey spots correspond to the provinces characterized by the presence of industrial districts, the white spots to the provinces characterized by the absence of industrial districts, and the black spot identiﬁes the b value for Italy. Two things are worth noting: the remarkable presence of industrial districts in provinces with comparative advantages, and the higher degree of persistence (lower dispersion around the 45º line) in the RCA of the same provinces. The same result extends to other sectors characterized by small ﬁrms, high territorial concentration and low–medium technological content, such as textiles or specialized machinery. But on aggregate the result is less clear-cut and the correlation between the y values and the dummy matrix is positive and signiﬁcative but not very strong (0.23), indicating that only for certain sectors is the district eﬀect enhancing comparative advantages and persistence, but that the attribution of the strong persistence in the structure of Italian export to the district eﬀect still needs some further research.

7.

CONCLUSIONS

In this chapter we have explored the structure of Italian exports, making use of the Balassa Index of Revealed Comparative Advantage, focusing on the export structure itself, on its change over time and on its degree of persistence. The analysis has been developed through nonparametric statistical techniques that allow us to estimate the empirical distribution of the Balassa Index and to track its dynamic change during three decades, from the 1970s to the present. The persistence in the pattern of comparative advantage is then related to three interacting theoretical explanations

218

–3

–2

–1

0

l og( b⬘) 1991

–4

–3

–2

-1

0

1

2

3

1

2

3

Footwear sector in the Italian provinces, 1991–2001

ISTAT (1997, 2002).

Figure 8.4

Sources:

Note: Grey spots identify provinces characterized by the presence of industrial districts (according to the ISTAT classiﬁcation); the black spot identiﬁes the b value for Italy.

log(b⬘) 2001

Italian comparative advantages

219

based on: factor proportions; the role of dynamic scale economies and of Marshallian externalities; and the role of quality and vertical diﬀerentiation and the absence of market incentives. From an empirical viewpoint the persistence in the pattern of comparative advantage has been subsequently conditioned to the presence of industrial districts, in order to show how the local structure of production is inﬂuencing such a relevant degree of persistence.

NOTES *

1.

2. 3.

4.

5. 6.

I am very grateful to MIUR (PRIN PIE 2003) and JETRO for ﬁnancial support and to Giancarlo Spagnolo for his encouragement. Many of the arguments of this paper come from the inspiring discussions I had over the years with Massimo Tamberi: however, all errors and omissions are my own responsibility. We do not want to enter into the details of a debate that is still alive (if somehow dormant) and which has not yet been properly schematized, organized and fully explored. The easier way of building up a taxonomy from the individual position of each singular contribution is to arbitrarily separate the ones that emphasized the pros from those that underlined the cons of the Italian pattern of comparative advantage. On the cons side, among many, Onida (1978), Modiano (1982), Epifani (1999) and Guerrieri and Rossi (2000), taking into account the characteristics of the structure of Italian exports, stressed the Italian dependence on imported technology, the low rate of R&D investment, the small size of ﬁrms which is at odds with scale economies, the limited eﬃciency of the credit market, and the role of family in the ownership and management of ﬁrms. Conti (1978), Iapadre (1996) and de Nardis (1997) related the rigidity of the Italian export structure to the possible risks in terms of a sluggish growth rate, a low employment capacity, and the fragility with respect to exogenous shocks, such as the European Monetary Union or a drop in world demand. On the pros side, Becattini (1999) and Fabiani et al. (1998) gave strong and robust evidence of the eﬃciency of small ﬁrms belonging to industrial districts, the local organization of production that prevails in many industrial sectors constituting the core of the Italian export structure. See Vollrath (1991) and De Benedictis and Tamberi (2004) for a general discussion on the various measures used in the literature. The variable generally used – as we shall in this chapter – is exports. Recently the availability of international comparable datasets has extended the choice of researchers to other possibilities which are used – somewhat acritically – as possible alternatives: imports, employment, value added, production and patents. The choice of the two years 1970 and 1998 – which correspond to the initial and the ﬁnal year of the time series – does not inﬂuence the results. We veriﬁed the robustness of the analysis to the choice of a particular year. Deriving b we have used the total (and sectoral) export of OECD countries as w variable(s); which – according to OECD (2002) – corresponds to more than 70 per cent of total world exports. We call it ‘pure persistence’ since it implies not only that Italy has not changed its export structure, but also that all other countries considered did not change their structure. The sectors considered in the 2-digit OECD (2002) dataset are: Food, Beverages, Tobacco, Textiles, Wearing Apparel, Leather and Products, Footwear, Wood Products, Furniture and Fixtures, Paper and Products, Printing and Publishing, Industrial Chemicals, Other Chemicals, Petroleum Reﬁneries and Products, Rubber Products, Plastic Products (nec), Pottery and China, Glass, Non-Metallic Products (nec), Iron and Steel, Non-Ferrous Metals, Metal Products, Non-Electrical Machinery, Electrical Machinery, Transport Equipment, Professional Goods, Other Manufacturing.

220

Agglomeration in Italy

7. The numerical values of the bs for Italy, Spain, Japan, Germany, the US and the UK derived from the OECD (2002) database for 1970 and 1998 are included in Appendix 6A. What the bs show is that, for example, in 1970, the share of Italian exports in the textiles sector was proportional to the world share to an order of magnitude of 1.42; and that in 1970 the share of the Italian textiles sector with respect to the Italian professional goods sector (the relative contribution to total export) was (1.42)(0.59) 0.84 times greater than for the sum of the countries considered in the sample set; and, ﬁnally, that in 1970 the share of the Italian textiles sector was (1.42)(0.91) 1.29 times greater than the same share for Germany. Therefore, using the bs in a cardinal way allows the possibility of demarcation and of sectoral and country ranking. 8. The deﬁnition of manufactures includes sectors with SITC codes from 5 to 9. Food sectors (codes 0, 1, 2 and 4) were not included because of the diﬃculties in separating raw material from manufactures. 9. Ten sectors were excluded from the analysis since there was no data available for 1998. The total number of manufacturing sectors in the dataset is therefore 530. 10. Since we used a linear OLS regression the estimated gives predictions in expected terms, which means that on average the degree of persistence is high. In order to take into account the possible inﬂuence of outliers we have also performed a robust regression, as suggested by Yohai et al. (1991): the diﬀerence was not noticeable. 11. The easier way of describing kernel densities is to relate them to histograms. As in Bowman and Azzalini (1997) the histogram may be written as: f(b)

n

I(b bj ; h), j1

where bj is the data (with j1, . . ., n); bj is the center of the interval in which bj falls, and I(b bj ; h) is the indicator function of the interval [–h, h]. The kernel estimator has the form: f (b)

n

I(b bj ; h). j1

12. 13. 14.

15.

Since the forms of the b distribution are so diﬀerent, we do not try to estimate parametrically the possible speciﬁc distribution. See Becattini (1999) for a recent summing up and Fabiani et al. (1998) for a quantiﬁcation of the district eﬀect in ﬁrms’ eﬃciency. We use the ISTAT (1997) classiﬁcation since it is the most widely used. Many other classiﬁcations are available, however (IPI 2002). No classiﬁcation is uncriticizable and each classiﬁcation tends to over- or underestimate the phenomenon. As an example, the Riviera del Brenta footwear district analysed in this volume, by Rabellotti (Chapter 9), is not included in the ISTAT (1997) classiﬁcation. The sectors considered are: Organic Chemicals; Inorganic Chemicals; Dyeing and Tanning; Medicinal and Pharmaceutical Products; Essential Oils for Perfume; Fertilizers; Plastics in Primary Forms; Plastics in Non-primary Forms; Other Chemical Materials; Leather; Rubber Manufactures; Wood (excluding Furniture); Paper; Textiles; Non-metallic Mineral Manufactures; Iron and Steel; Non-ferrous Metals; Manufactures of Metal; Power-generating Machinery; Specialized Machinery; Metalworking Machinery; Other Industrial Machinery and Parts; Oﬃce Machines; Telecommunication and Sound Recording Apparatus; Electrical Machinery; Road Vehicles; Other Transport Equipment; Prefabricated Buildings and Sanitary; Furniture; Travel Goods and Handbags; Apparel and Clothing; Footwear; Professional and Scientiﬁc Instruments; Photo-optical Goods and Watches; Miscellaneous Manufactured Articles.

Italian comparative advantages

221

REFERENCES Balassa, Bela (1965), ‘Trade liberalisation and revealed comparative advantage’, Manchester School of Economics and Social Studies, 33, 99–123. Becattini, Giacomo (1999), Distretti industriali e made in Italy. Le basi socioculturali del nostro sviluppo, Turin: Boringhieri. Bowman, Adrian W. and Adelchi Azzalini (1997), Applied Smoothing Techniques for Data Analysis, Oxford: Clarendon Press. Conti, Giuliano (1978), ‘La posizione dell’Italia nella divisione internazionale del lavoro’, in Alessandrini Piero (ed.), Specializzazione e competitività internazionale dell’Italia, Bologna: Il Mulino. De Benedictis, Luca and Pier Carlo Padoan (1999), ‘Dynamic scale economies, specialization, and the cost of the single currency’, International Journal of Development Planning Literature, 14 (4), 549–60. De Benedictis, Luca and Massimo Tamberi (2000), ‘La specializzazione internazionale dell’Italia: anomalie, dinamica e persistenza’, in Rapporto sull’industria italiana, Maggio: Centro Studi Conﬁndustria. De Benedictis, Luca and Massimo Tamberi (2004), ‘Overall specialization empirics: techniques and applications’, Open Economies Review, 15 (4), 323–46. de Nardis, Sergio (1997), ‘Persistenza e cambiamento delle specializzazioni manifatturiere: l’industria italiana nel confronto con i principali paesi’, Rivista di Politica Economica, 137 (1), 89–105. de Nardis, Sergio and Fabrizio Traù (1999), ‘Specializzazione settoriale e qualità dei prodotti: misure della pressione competitiva sull’industria italiana’, Rivista italiana degli economisti, 4 (2), 177–212. ECLAC–World Bank (1999), TradeCAN database, Washington, DC: World Bank. Epifani, Paolo (1999), ‘Sulle determinanti del modello di specializzazione internazionale dell’Italia’, Politica Economica, 15 (2), 195–224. Fabiani, Silvia, Guido Pellegrini, E. Romagnolo and Luca F. Signorini (1998), ‘L’eﬃcienza delle imprese nei distretti industriali’, Sviluppo Locale, 5 (9), 42–73. Guerrieri, Paolo and Salvatore Rossi (2000), ‘Vantaggi competitivi reali nell’Europa con un solo mercato e una sola moneta’, in Pier Carlo Padoan (ed.), L’euro e i mercati reali, chapter 1, Bologna: Il Mulino. Iapadre, Lelio (1996), ‘La collocazione internazionale dell’economia italiana: indicatori statistici e tendenze recenti’, Economia Italiana, 3, 437–83. IPI (Istituto per la promozione industriale) (2002), L’esperienza italiana dei distretti industriali, Rome: Istituto per la promozione industriale, Ministero delle attivitá produttive. ISTAT (1997), I sistemi locali del lavoro 1991, edited by F. Sforzi, Rome: Istituto Poligraﬁco e Zecca dello Stato. ISTAT (2002), Dati di commercio estero delle province italiane, October, www.coeweb.istat.it/. Lehmann, Erich L. (1975), Nonparametrics: Statistical Methods Based on Rank, San Francisco, CA: Holden Day. Modiano, Pietro (1982), ‘Competitivitá e collocazione internazionale delle esportazioni italiane: il problema dei prodotti tradizionali’, Economia e Politica Industriale, 33. Organization for Economic Cooperation and Development (OECD) (2002), STAN Database -2002 edition, Paris.

222

Agglomeration in Italy

Onida, Fabrizio (1978), Industria italiana e commercio internazionale, Bologna: Il Mulino. Silverman, Bernard W. (1986), Density Estimation for Statistics and Data Analysis, London: Chapman Hall. Vollrath, Thomas L. (1991), ‘A theoretical evaluation of alternative trade intensity measures of revealed comparative advantage’, Weltwirtschaftliches Archiv, 127, 265–80. Yohai, V., W.A Stahel and R.H. Zamar (1991), ‘A procedure for robust estimation and inference in linear regression’, in W.A. Stahel and S.W. Weisberg (eds), Directions in Robust Statistics and Diagnostics, Part II, New York: SpringerVerlag, pp. 365–74.

223

0 0 1 1 1 0 0 0

1.30 2.10 1.36 1.25 2.23 0.49 0.30 1.16

0.49 1.20 0.02 1.42 3.38 2.16 7.41 0.54 1.81 0.27 1.13 0.74 0.78 2.23

0 0 0 1 1 1 1 0 1 0 0 2 2 2

Food Beverages Tobacco Textiles Wearing Apparel Leather and Products Footwear Wood Products Furniture and Fixtures Paper and Products Printing and Publishing Industrial Chemicals Other Chemicals Petroleum Reﬁneries and Products Rubber Products Plastic Products, nec Pottery, China etc. Glass and Products Non-Metallic Products, nec Iron and Steel Non-Ferrous Metals Metal Products

Italy 1970

0.97 1.67 1.58 1.13 3.28 1.05 0.53 1.70

0.76 1.25 0.15 2.29 3.14 4.88 5.53 0.46 3.44 0.57 0.94 0.65 0.73 0.70

Italy 1998

OECD data: b–1970, 1998

SEC

Table 8A.1

APPENDIX 8A

1.29 1.90 2.95 0.61 0.71 2.01 0.34 1.08

0.41 0.03 0.00 1.79 1.23 0.69 0.37 0.59 0.34 0.22 0.51 0.89 0.34 0.12

Japan 1970

1.20 0.28 1.51 0.85 0.55 1.24 0.49 0.71

0.07 0.04 0.14 0.49 0.06 0.12 0.02 0.03 0.08 0.17 0.16 0.80 0.64 0.29

Japan 1998

0.59 0.62 0.12 0.80 0.61 0.50 0.73 0.78

1.03 0.08 2.28 0.35 0.29 0.44 0.06 0.70 0.30 0.92 1.20 1.20 1.15 0.76

US 1970

0.69 0.87 0.44 0.82 0.36 0.33 0.74 0.85

0.93 0.40 2.67 0.73 0.68 0.47 0.23 0.85 0.54 0.95 1.26 1.12 0.94 0.78

US 1998

1.33 0.75 1.70 0.84 0.98 0.64 1.23 1.22

0.39 2.94 2.11 1.13 0.82 1.44 0.66 0.13 0.65 0.33 1.54 0.89 1.52 1.06

UK 1970

0.96 0.90 1.69 0.61 0.70 0.89 0.91 0.82

0.68 2.02 0.80 1.01 1.03 0.76 0.61 0.14 0.52 0.54 1.97 1.06 1.53 1.06

UK 1998

2.38 0.67 0.98 0.74 1.34 0.38 0.76 1.23

3.04 3.86 0.65 0.84 1.32 2.90 8.38 1.59 1.85 0.29 3.95 0.72 0.60 2.35

Spain 1970

1.89 1.03 1.31 1.15 3.46 1.31 0.95 1.09

1.41 1.68 0.24 1.04 0.76 2.17 3.54 0.59 1.24 0.70 1.30 0.76 0.70 1.41

Spain 1998

0.87 1.03 1.41 1.07 1.14 1.04 0.58 1.33

0.37 0.25 0.33 0.91 0.67 0.94 0.40 0.33 1.50 0.32 0.93 1.27 1.17 0.86

Germany 1970

1.10 1.11 0.95 1.00 0.80 1.08 0.91 1.22

0.73 0.45 0.71 0.88 0.85 0.59 0.44 0.42 0.83 0.88 1.07 1.23 1.08 0.48

Germany 1998

224

Source:

Our elaborations on OECD (2002).

0 is ‘other’, 1 is ‘traditional’ and 2 is ‘advanced’.

Note:

1.16 0.63 0.58 0.57 1.79

1.32 0.83 0.68 0.59 1.39

Non-Electrical Machinery Electrical Machinery Transport Equipment Professional Goods Other Manufacturing

0.61 1.89 1.07 1.38 1.47

Japan 1970

2 2 2 2 0

Italy 1998

Italy 1970

(continued)

SEC

Table 8A.1

1.34 1.75 1.24 1.65 0.72

Japan 1998 1.44 1.01 1.26 1.48 0.72

US 1970 1.21 1.21 0.98 1.37 0.68

US 1998 1.17 0.91 1.02 1.13 2.13

UK 1970 1.10 1.04 0.86 1.11 1.89

UK 1998 0.57 0.50 0.65 0.23 0.62

Spain 1970

0.49 0.60 1.61 0.35 0.56

Spain 1998

1.36 1.09 1.03 1.32 0.80

Germany 1970

1.09 0.91 1.14 1.03 0.54

Germany 1998

9. Globalization, industrial districts and value chains Roberta Rabellotti* 1.

INTRODUCTION

This study is concerned with the impact of global transformations, such as international relocation of production, increasing concentration in distribution, mergers of ﬁrms on local competitiveness in Italian industrial districts. The aim is to integrate the typical industrial district approach,1 traditionally focused on analysing the local sources of competitive advantages, with the global value chain literature which stresses that activities such as design, production and distribution are often located in diﬀerent regions or countries (Kaplinsky 1998 and Gereﬃ 1999). In the typical industrial district, most of the activities along the value chain have traditionally been locally undertaken and competitiveness of producers has mainly come from intra-cluster vertical and horizontal relationships. Recent changes in global markets and particularly the increasing concentration of global trade and retailing in the hands of transnational companies suggest that more attention needs to be paid to external linkages. This study tries to ﬁll this gap in the literature on industrial districts using some of the insights of the global value chain approach. By doing so it takes into account activities taking place outside the districts in particular to understand the signiﬁcance of the relationships with key external actors.2 The following questions are addressed. Is globalization pushing Italian industrial districts toward new value chains? What types of governance characterize the relationships between local and outside actors? Do the chains’ leaders come from inside or outside the districts? Does the integration of industrial districts in global value chains enhance or weaken local upgrading strategies? The chapter is organized as follows. After the introduction in the second section, industrial districts and their focus on internal linkages are compared with some of the main issues that have arisen in the literature on value chains. The importance of analysing the model of governance, in other words how the various activities in the chains are coordinated, is stressed 225

226

Agglomeration in Italy

and it is pointed out that coordination may take place through arm’s-length market relations or non-market relationships. The diﬀerent models of governance may inﬂuence the opportunity of ﬁrms to upgrade within the chain. Various forms of upgrading are considered: improvements in process, improvements in product and changing functional positions. In order to understand what may be implied in terms of upgrading for a company located in an industrial district to be part of a value chain, the author reports some of the ﬁndings of a study undertaken in Riviera del Brenta, a highly specialized and strongly export-oriented shoe district, located near Venice. The analysis is based on primary data collected in Brenta, with interviews conducted at shoe manufacturing enterprises, and qualitative information collected through in-depth interviews at trade organizations and fashion ﬁrms, and with retailers, international buyers and other key informants. The chapter analyses the recent entry of some global fashion leaders into the footwear sector. An increasing number of fashion companies, originating in other sectors, have penetrated the footwear industry looking for skilled manufacturers in order to outsource the production of shoes sold with their brand names. This is why in Brenta many producers have begun to work as subcontractors, abandoning key activities such as design and sales, which are the core competencies of fashion companies. The latter have become the new lead ﬁrms of the chain. Furthermore, the chapter analyses how global linkages have aﬀected local linkages, stressing that the latter have weakened in recent years: in backward linkages the weakening is due to the increasing outsourcing to Romania and other nearby low-wage countries and in forward linkages it is due to the forced abandonment of activities such as design and marketing, notably in the top brand value chain. The ﬁnal section draws together the main conclusions and reﬂects on how Brenta’s future evolutionary path is inﬂuenced by its recent redeﬁned participation in global chains.

2.

INDUSTRIAL DISTRICTS AND VALUE CHAINS

The literature on industrial districts in advanced and less-developed countries has shown that clustering helps local enterprises overcome growth constraints and compete in distant markets.3 The general argument is that competitiveness of producers mainly comes from intra-cluster vertical and horizontal relationships generating collective eﬃciency, namely increasing returns from incidental economies of agglomeration and active cooperation (Schmitz 1995 and 1999; Rabellotti 1997).

Globalization, industrial districts and value chains

227

Nevertheless, recent changes in production systems, distribution channels and ﬁnancial markets, accelerated by the globalization of product markets and the spread of information technologies, suggest that more attention needs to be paid to external linkages.4 Furthermore, industrial boundaries are blurring and the shape of industries is no longer conforming to the standard industrial classiﬁcation (Mytelka 2000). This is the case for instance of the footwear sector, analysed in this chapter, which is increasingly integrated in the fashion industry, dominated by a few multi-product oligopolies, exploiting economies of scale and scope in activities such as distribution, marketing and branding across traditionally separated industrial sectors such as shoes, clothing, glassware, perfumes and leather accessories. Accordingly, to understand the eﬀect of these changes on a district it becomes necessary to adopt an analysis which pays attention to linkages with actors external to the district. The research on global value chains seeks to understand the nature of these relationships and their implications for upgrading. An important aspect stressed in the literature on value chains is that the various activities in the chain are subject to some degree of governance or coordination (Gereﬃ 1999). At any point in the chain, activities are deﬁned by three key parameters: what is to be produced (design of products); how it is to be produced (deﬁnition of production process: technology, quality standards) and how much has to be produced (Kaplinsky and Readman 2001). Coordination may take place through arm’s-length market relations or non-market relationships. In the latter case, following Humphrey and Schmitz (2000), we distinguish among three types of governance: (i) the network implying cooperation between ﬁrms of more or less equal power which share their competencies within the chain; (ii) a quasi-hierarchy involving relationships between legally independent ﬁrms in which one is subordinated to the other and where the leader in the chain deﬁnes the rules that the rest of the actors have to comply with; and (iii) hierarchy when the local ﬁrm is owned by an external ﬁrm. Among the diﬀerent forms of governance, the literature on value chains, which is mainly concerned with developing countries, stresses the importance of the quasi-hierarchy type, distinguishing between those cases when coordination is undertaken by buyers (‘buyer-driven chain’) and those in which producers play the key role (‘producer-driven chains’) (Gereﬃ 1994). Moreover, several authors conclude that the increasing concentration of retailing in developed countries makes buyer-driven chains a growing phenomenon (Gereﬃ 1999; Dolan and Humphrey 2000). In this chapter we aim to contribute to this debate by broadening the analysis to include developed countries, showing that quasi-hierarchy may also be a relevant form of governance in linkages existing between producers from

228

Agglomeration in Italy

one of the main shoe clusters in Italy and the lead fashion ﬁrms governing the chain. Furthermore, we shall address the following question: how does the insertion into global value chains aﬀect local upgrading strategies? The concept of upgrading is used here in the sense proposed by Humphrey and Schmitz (2002). Process upgrading means transforming inputs into outputs more eﬃciently by reorganizing the production system or introducing superior technology; product upgrading can be deﬁned as moving into more sophisticated product lines; and functional upgrading is acquiring new, superior functions in the chain, such as design or marketing. Besides, from our empirical analysis in Brenta a diﬀerent possible form of functional upgrading coming to the fore is the externalization of low value-added functions combined with a focus on more advanced activities or higher value-added segments of the market. This chapter presents some of the ﬁndings of a survey undertaken in Brenta, investigating problems and opportunities arising when a district gets involved in a global value chain. Focusing on external linkages, the model of governance of the main value chain in Brenta is analysed and the upgrading eﬀect on local ﬁrms is explored. In addition, we analyse how global linkages have aﬀected local relationships.

3.

THE BRENTA SHOE DISTRICT

In Brenta the origins of the footwear industry date back to the beginning of the last century. During the footwear industry boom after the Second World War, the sector progressively absorbed most of the rural workforce available in the area. In the 1960s, the local enterprises expanded and increased their exports, specializing in the upper segments of the market. In 2001, 88 per cent of the shoes produced in the area were medium–high and high-priced women’s shoes with an average ex-factory price of €58. Since the second half of the 1980s, the area has suﬀered from increasing competition in the international market and sales have stagnated, ﬂuctuating between 7.9 and 8.8 million pairs (mainly due to exchange rate ﬂuctuations). In value terms, however, sales continued to increase in most years (ANCI 2001). It may be useful to add that Brenta’s performance is in line with that of the rest of the Italian footwear industry, which suﬀered from stagnating European demand and from increasing international competition (ibid.). In terms of the structure of the district,5 there were 993 ﬁrms in 2001. These comprised 436 shoe producers, 412 ﬁrms manufacturing inputs such as heels, soles and lasts, 69 design ﬁrms and 76 trading companies. The

Globalization, industrial districts and value chains

229

number of ﬁrms has decreased continuously from the second half of the 1980s as has the number of employees. Shoe factories in Brenta are characterized by their small size: 78 per cent of the ﬁrms employing 40 per cent of the workers have fewer than 20 employees and 18 per cent, corresponding to 34 per cent of total employment in the shoe industry, have between 20 and 49 employees (ISTAT 1996). The average size is 15 employees. The total value of shoe production is about €948 million, close to 12 per cent of the total turnover of the Italian footwear industry. Over the years, Brenta’s orientation toward the external market increased from about 70 per cent of sales exported at the beginning of the 1980s to 89 per cent in 2001. This represents more than 10 per cent of total Italian exports in the sector (ACRIB 2002). In what follows, the ﬁndings of a questionnaire survey are presented: 40 shoe manufacturing enterprises were interviewed, randomly selected from a list provided by ACRIB (Associazione Calzaturiﬁci della Riviera del Brenta), the local entrepreneurial association of footwear producers. Qualitative information was also collected through in-depth interviews at trade organizations and fashion ﬁrms, and with retailers, foreign buyers and other key informants. The size distribution of our sample (40 ﬁrms) reﬂects the fact that the majority of ﬁrms in Brenta are small. Seven ﬁrms (17.5 per cent of the total sample) have fewer than 20 employees, 20 ﬁrms (50 per cent) have between 20 and 49 workers, a further nine employ between 50 and 99 people and only four companies (10 per cent) have more than 99 workers. The survey explores how and when ﬁrms sell their products, identifying the main global value chains, discussing the issue of governance and ﬁrms’ upgrading.

4.

THE MARKETS

In Riviera del Brenta the main market has traditionally been Europe, particularly Germany, and to a more limited extent France, Great Britain and the rest of the EU. Among sample ﬁrms, 35 per cent sell between 50 and 90 per cent and 17.5 per cent more than 90 per cent of their production to Europe. Sales to Italy are less than 10 per cent for the majority of the sample (53 per cent of ﬁrms) and less than 50 per cent for another 23 per cent of ﬁrms. In the European market, Brenta’s companies are selling to a variety of customers. In the UK they typically sell to large buyers or department stores, in France and Italy they supply mainly to independent retailers and in Germany to buying groups.6

230

Agglomeration in Italy

Regarding the rest of the world, Brenta’s market penetration is more diﬃcult due to the small size of local enterprises, geographical distances and the large investments involved. Only nine ﬁrms (23 per cent of the sample) export between 10 and 50 per cent of their production to the USA and 12 (30 per cent) to other countries, mainly in the Far East, Russia and the Middle East. It should be added that exports to the USA have increased considerably in the past years, boosted by the weakness of the euro. Brenta’s presence in the Far East, the Middle East and Russia is very volatile, and sales mainly occur through export agents. A potentially important market is Japan, however, protectionism still represents an obstacle to further development.7 Furthermore, the demand from other Asian countries, such as Hong Kong, is slowly recovering after the ﬁnancial crisis in 1997. Russia also remains a very unpredictable market. Overall, none of the sample ﬁrms maintains very stable relationships with customers in these distant markets; many ﬁrms see them as opportunities to diversify their sales but none of them systematically invests in marketing. Thus, it appears that sample ﬁrms are able to maintain many exit options, selling in many diﬀerent geographical markets and to a large variety of customers, such as independent retailers, department stores and large buyers. The only exceptions are eight ﬁrms selling more than 50 per cent of their production to large buyers in the European market (seven in Europe and one in Italy). The ability to combine many customers is also conﬁrmed by the limited amount of production sold by each ﬁrm to its two main customers (to whom on average, sample ﬁrms sell only 16 per cent of their production).8 The above analysis of the main markets of Brenta has nevertheless neglected a category of increasingly important customers for the district – the high-fashion companies. Many ﬁrms in Brenta have recently begun to work as subcontractors to some leading global fashion ﬁrms. The next section analyses this value chain characterized by high fashion and global reach.

5.

THE TOP BRAND CHAIN

The world market of luxury goods is estimated at about €46 billion in 2001, 60 per cent of which corresponds to clothing and 8 per cent to footwear. In this rich and rapidly increasing market, the Italian market share is 30 per cent, followed by France with 25 per cent. In recent years, the luxury fashion system went through important changes, turning it into an oligopoly dominated by a few multi-product giants. The growth strategy of many companies has been characterized by a similar pattern. First, successful ﬁrms established their brand names in

Globalization, industrial districts and value chains

231

speciﬁc product lines (for example, three among the most important and largest companies in the industry, LVMH, Gucci and Prada, began producing and selling leather goods). Second, they capitalized on their brand names and diversiﬁed to other segments (in the cases named above, they entered into clothing, footwear, glassware, perfumes and wines). Finally, they have also begun to grow through the acquisition of other well-known existing brands. During the last ﬁve years, mergers and acquisitions in the fashion industry have increased from 31 in 1997 to 155 in 2001. In the shoe sector, there was a total number of 15 operations (Pambianco 2002). The economic logic behind these growth strategies is a search for scale and scope economies in activities other than manufacturing, such as branding, marketing, advertising, and opening of mono-brand shops in the most exclusive and expensive streets in the world. The increasing concentration of fashion enterprises in the intangible phases of the value chain may be explained by the growing concentration of rents in these activities. According to Kaplinsky (2000), in the past decade the barriers to entry in manufacturing have begun to fall and consequently the rent going to production activities has shrunk in favor of rents accruing to activities outside the area of production. This explains the growing concentration of investments in areas like branding, advertising, marketing and sales where capital cost barriers to entry are high. Creating and maintaining a global brand is very expensive and only those with access to international ﬁnancial markets are able to aﬀord it. Therefore, leading ﬁrms are expanding through mergers and acquisitions, capitalizing on their core competencies such as design, advertising, marketing and band naming, which are no longer sector speciﬁc. This implies that we can consider the footwear or clothing industries as subsystems of the global fashion system. The top brand value chain can be regarded as a subtype of the ‘buyerdriven’ chain because the lead ﬁrms are the owners of top global brands, controlling activities connected with intangible characteristics of the products such as design, brand name, marketing and distribution. Gereﬃ’s work is more focused on ‘buyer-driven’ chains ‘in which large retailers, branded marketers and branded manufacturers play the pivotal roles in setting up decentralized production networks in a variety of exporting countries, typically located in the Third World’ (Gereﬃ 1999, pp. 41–2). Although he discusses the importance of the creation of brands, he is more concerned with global brands in mass markets, such as Liz Claiborne, Nike and Reebok. In the luxury market, barriers to entry are supposed to be higher, and returns from branding and marketing very high. Therefore, following Kaplinsky (1998, 2000), economic rents in this chain are assumed to be higher than in other types of chains.

232

Agglomeration in Italy

The next section shows that, in the top brand chain, developed-country producers (and not only developing countries as in many Gereﬃ studies) are aﬀected by increasing buyer concentration. It follows the analysis of the rapid expansion of the top brand chain in Brenta, investigating the uneven relationship between local footwear producers and top global brand leaders.

6.

BRENTA IN THE TOP BRAND CHAIN

Recently, the high and medium–high segments of the footwear industry have increasingly attracted the interest and the ﬁnancial capital of wellknown top brands and luxury multi-product oligopolies from outside the shoe world.9 Some world top luxury companies, looking for highly skilled manufacturing capabilities to begin footwear production, identiﬁed the Riviera del Brenta as a preferred area in which to ﬁnd subcontractors. The beginning of this trend corresponded with a diﬃcult time in Brenta because local ﬁrms were facing the end of the positive impact on exports of the 1992 devaluation of the lira. Among our sample ﬁrms, 17 enterprises (corresponding to 42.5 per cent of the sample) work as subcontractors to high-fashion companies, producing shoes sold with globally known top brands. In ﬁve cases (12.5 per cent), they work exclusively as subcontractors, while four of them (10 per cent) make between 50 and 89 per cent of their total production for highfashion companies and the remaining eight (20 per cent of the sample) make less than 50 per cent (Table 9.1).10 According to our interviews, it appears that in many cases the amount of production made as subcontractors is still rather changeable, depending on the season. Furthermore, some ﬁrms do not have direct contact with high-fashion companies, instead they are subcontractors to other local enterprises who are the direct subcontractors. In most of the cases investigated, fashion companies provide the design and Brenta manufacturers take care of all production phases, including product development11 and purchase of raw materials and components. After that, shoes are sold by fashion companies with their brand names. In our sample, design is totally controlled by fashion companies in 65 per cent of the ﬁrms working as subcontractors, while 35 per cent of them contribute to design. This often means that fashion companies give producers some ideas and sketches to be transformed into a shoe.12 Sales are undertaken by fashion companies in 82 per cent of the cases (Table 9.2), while product development and purchase of inputs are carried out in most of the cases by the subcontracting ﬁrms. Nevertheless, according to many ﬁrms interviewed, fashion companies are increasingly becoming involved in these

233

Globalization, industrial districts and value chains

Table 9.1

Production for high-fashion companies among sample ﬁrms

% of total production

No. of enterprises

% of total sample

0 1–49 50–89 90–

23 8 4 5

57.5 20.5 10.0 12.5

Total

40

100.0*

Note: * Rounded. Source: Author’s survey.

Table 9.2 Internal functions in sample ﬁrms working as subcontractors to high-fashion companies No. of sample ﬁrms* Design Product development Purchase of components Sale

Not undertaken

Partially undertaken

Totally internal

11 (65%) 2 (12%) 1 (6%) 14 (82%)

4 (23%) 2 (12%) 3 (18%) 2 (12%)

2 (12%) 13 (76%) 13 (76%) 1 (6%)

Note: * The total number of sample ﬁrms working as subcontractors is 17. Ratio of the total number of subcontracting enterprises in the sample in parentheses. Source: Author’s survey.

activities by directly selecting suppliers, sometimes even through acquisitions of ﬁrms, and extending their control on quality and delivery conditions backwards along the chain. From what has been said so far, it appears that Brenta has been undergoing a process of functional downgrading. Traditionally the design and acquisition of inputs were controlled locally, and carried out inside the ﬁrms or inside the district. More recently, with the advent of the luxury fashion companies, local enterprises are moving out of design and sale. There are also signs of luxury fashion companies extending their control backwards along the chain. Integration in the luxury fashion value chain is thus causing a process of functional downgrading at district level in those activities that are the typical core cross-sector competencies of luxury fashion companies, namely design, branding and sales.

234

Agglomeration in Italy

Nevertheless, although Brenta is showing a trajectory of functional downgrading resulting from the integration in top brand chains, it has to be underlined that many of the leading companies in those chains are Italian. Therefore, by moving from a narrow district perspective to take into consideration the evolution of the Italian fashion system as a whole, our conclusions may be very diﬀerent. The Italian luxury goods industry is deﬁnitely undergoing a process of functional upgrading and concentration in rent-rich activities by exploiting across sectors its core competencies in design, branding and marketing. This is very diﬀerent from what is often occurring in developing countries. Small Brenta producers are abandoning some key activities which have moved to the headquarters of the chains’ leaders in Milan. In the case of producers taking part in global chains in the developing world, these activities are never carried out within the country, instead they are ﬁxed in New York, London or other cities in the developed world. Apart from design, the functional downgrading that is occurring in Brenta concerns activities in which local enterprises, probably including the majority of small Italian footwear ﬁrms, have traditionally been rather weak, that is, branding, marketing and sales strategy. In our sample, 60 per cent of ﬁrms do not perform any marketing at all and the existing brand names are sometimes recognized at national level (mainly in Germany) but never globally. A reason for the very limited local investments in these activities is the average company size. Firms are too small to aﬀord very expensive strategies in marketing or advertising and to impose a brand name in the global market. Local entrepreneurs are aware of their weaknesses and are beginning to accept that in the global market high production skills are no longer enough to sell their products; brand names and aggressive marketing strategies have become unavoidable competitive factors. Therefore, many of them agree that becoming subcontractors to luxury fashion companies is a way to face the challenge of globalization despite the cost of functional downgrading. Furthermore and quite unexpectedly, our empirical analysis shows that this choice is not an impoverishing strategy. In fact, Table 9.3 shows a statistically signiﬁcant positive relationship between performance13 and the share of production sold to high-fashion companies. There is, of course, a problem of causality in interpreting these ﬁndings because it is impossible to know a priori whether positive performance is caused by participation in the luxury value chain or quite the reverse, high-fashion companies select their subcontractors among better-performing enterprises. In order to address the causality problem we tested the relationship between performance and production of subcontractors to top brand

235

1.000 0.461** 0.028 0.142 0.256*

Product innovation

1.000 0.213 0.068 0.092

Process innovation

1.000 0.126 0.161

No. of employees

1.000 0.447**

Ex-factory price

1.000

% of production as subcontractor

Source:

Author’s survey.

Notes: Some of the variables in this study (process and product innovation) are ordinal variables which can be ordered qualitatively in a limited number of ordered classes. Therefore, to analyse correlation we calculate the Kendall coeﬃcient which is the more appropriate for ordinal variables. However, alternative correlation measures, such as Spearman and Pearson coeﬃcients, were also calculated and no diﬀerences in terms of statistically associations were found. * Correlation is signiﬁcant at the 0.05 level; ** correlation is signiﬁcant at the 0.01 level.

1.000 0.104 0.042 0.051 0.319** 0.229*

Performance

Kendall correlation coeﬃcients

Performance Production innovation Process innovation No. of employees Ex-factory price % of production as subcontractor

Table 9.3

236

Agglomeration in Italy

companies with OLS (ordinary least squares) regression analysis. Among the independent variables included (indicators of process and product innovation and number of employees) the amount of production as subcontractors is the only statistically signiﬁcant variable explaining performance. Nevertheless, the cause–eﬀect relationship remains unclear given that if the amount of production as subcontractors becomes the dependent variable, then the beta coeﬃcient of performance is the only one that is statistically signiﬁcant among regressors. On the other hand, although our sample survey does not provide a suﬃciently robust empirical test of the hypothesis that performance is positively inﬂuenced by the participation in the luxury value chain, we may corroborate our argument by looking at the economic theory of rent.14 In fact, the extent of subcontracting shoes is derived from the demand of luxury shoes and given that ﬁnal consumers are prepared to pay a high price then top brand companies are also willing to pay a relatively high price to their high-quality subcontractors, sharing with them a (small) proportion of their rent. In other words, we argue that top brand companies are exploiting ﬁnal consumers’ willingness to pay very high prices for luxury goods, earning a rent or a super-normal proﬁt above production costs. This rent, it seems, is to some extent shared within the chain in order to guarantee high and consistent quality and respect of delivery conditions. The high rents earned in the top brand value chain also explain why Brenta enterprises are increasingly using their internal production capacity to make shoes as subcontractors to high-fashion companies and outsourcing abroad the rest of their production. Decentralization to Romania and other neighboring countries is a necessity for reducing costs, given that price competition is severe even in the high-quality market. In this respect, outsourcing may be interpreted as a strategy of functional upgrading of Brenta ﬁrms: moving low value-added activities abroad and at home focusing on production for the rent-rich luxury market. Once more with respect to upgrading, Table 9.3 shows that there is a positive and statistically signiﬁcant relationship between the amount of production made for high-fashion companies and the degree of product upgrading. The role played by top brand companies in product innovation is conﬁrmed by the fact that 60 per cent of sample ﬁrms stress the importance of their assistance in this ﬁeld. Sample ﬁrms identiﬁed a number of advantages coming from their activity as subcontractors. The most important are: the size of orders (all ﬁrms except one); their regularity (all ﬁrms except two); and, for 70 per cent of them the prestige of working for a world-renowned top brand. A further advantage, very much emphasized by the interviewed ﬁrms, is the reduction of costs because they no longer have to produce a sample set. Among the

Globalization, industrial districts and value chains

237

disadvantages, imposition of delivery conditions and timing, lack of direct market access and the loss of independence were stressed by 75 per cent of sample ﬁrms. These disadvantages are strongly related to the issue of governance within the chain. Coordination of the value chain is clearly in the hands of top brand companies who keep their control on rent-rich activities such as design, branding and marketing. They are also increasingly becoming more directly involved in shoe and component production. Nevertheless, only two sample ﬁrms have clearly deﬁned their relationship with high-fashion companies as quasi-hierarchical, while the remaining ones reported that there is some degree of cooperation. A clear assessment of how hierarchical these relationships are is a diﬃcult task. There are mixed signs: in many cases ﬁrms are not fully dependent on the top brand value chain, producing less than 50 per cent of total production as subcontractors. In some cases they even contribute to design but, on the other hand, most of them suﬀer from functional downgrading and are losing their direct link with the market. Furthermore, 75 per cent complain about dependence on high-fashion companies. We may conclude that the most common type of governance within this chain is somewhere in between network and quasi-hierarchy. The clear leaders in the chain are the top brand companies. They are deﬁnitely setting the parameters that the rest of the actors have to comply with but in many cases they are also cooperating with their highly qualiﬁed partners to obtain top-quality products and besides, very importantly, they are willing to share with them part of their rent in order to acquire their production skills. The local entrepreneurial association is pushing Brenta ﬁrms to increase cooperation with high-fashion companies by making themselves unavoidable partners and eventually acquiring licenses.15 The question is, for how long will this strategy be successful, because some high-fashion companies are already setting up their own production facilities and others are beginning to decentralize production of more standardized shoes to Romania and other low-wage countries. They will probably need Brenta ﬁrms as subcontractors of top-quality shoes for a long time but competition is tough within the district and Italy more generally. Following this strategy, Brenta ﬁrms may seek to build up a cooperative relationship with the chain’s leaders, capitalizing on their excellence in production but they may progressively lose their design capability and the direct market access. To conclude, it is useful to summarize our main empirical ﬁndings in connection with our reference point on value chains (Gereﬃ 1999). The top brand value chain conﬁrms Gereﬃ’s trend toward ‘buyer-driven’ chains. However, there are also some new ﬁndings. Producers from developed countries are involved in the chain and their participation, far from having

238

Agglomeration in Italy

an upgrading eﬀect (much stressed in Gereﬃ’s work), has a functional downgrading eﬀect instead. Nevertheless, although ﬁrms have abandoned some key functions, their performance is still very positive because they have the prospect of sharing with the chain’s leaders (most of which are Italian) the high rents of the luxury industry. The next section analyses how global linkages have aﬀected local linkages within Brenta.

7.

LOCAL GOVERNANCE IN BRENTA

While Brenta never lived up to the romantic view of the Third Italy, much advertised in the past literature on industrial districts, it has nevertheless been traditionally characterized by strong local linkages at both a vertical and a horizontal level. As documented in a previous work carried out in Brenta in 1992 (Rabellotti 1997), local shoe producers had intensive and cooperative linkages with their local suppliers and subcontractors and frequent informal contacts with other shoe ﬁrms in the district. Compared with Marche, another major footwear district analysed at that time, entrepreneurs in Brenta showed a more cooperative attitude, also attributing more importance to these informal relationships. In addition to cooperative vertical and horizontal links among ﬁrms, Brenta is also characterized by the existence of several collective institutions, that is, ACRIB, the local entrepreneurial association established more than 30 years ago; an export consortium (Consorzio Maestri Calzaturieri del Brenta) created in 1967, and a technological and training institute (Consorzio Centro Veneto Calzaturiero) set up in 1986.16 Among many initiatives organized by these bodies, a signiﬁcant example is the collective organization at international fairs, managed by the Consorzio Maestri Calzaturieri. For example, at the September 2000 international shoe fair in Dusseldorf, 81 exhibiting producers from Brenta participated in a space collectively organized with a common look and identiﬁed by the logo of the consortium. However, a signiﬁcant change can be noted by comparing local governance in the early 1990s with that at the turn of the century. Links within the cluster have become less important and links with outside actors have become more important. In 2000, among the sample ﬁrms, only 25 per cent think that relationships with other local shoe enterprises are important, while the remaining 70 per cent attach no importance to them at all. Furthermore, during recent years, cooperation among shoe enterprises has decreased within the district according to 30 per cent of sample ﬁrms. A better, but far from good, evaluation is given to the relationship with the local entrepreneurial association ACRIB, which is considered very

Globalization, industrial districts and value chains

239

important by 15 per cent of the enterprises, important by 37.5 per cent and unimportant by 40 per cent. Among local links, only relationships with suppliers17 assume a real importance. They are very important for 65 per cent of sample ﬁrms and important for 15 per cent (Table 9.4). Regarding the importance attributed to the reputation of the district, most of the sample ﬁrms believe that their main buyers and high-fashion companies do not attach much importance to this aspect. This is conﬁrmed by the interviews with buyers, who did not select ‘Made in Brenta’ or ‘Made in Italy’ among the main factors of competitiveness. In fact, the cluster has never made a real eﬀort to promote its collective image, apart from joint participation at the fairs. Furthermore, the ‘Made in Italy’ factor is progressively losing its impact on the market because customers are looking more for individual top brands and because in many cases shoes are no longer made in Italy but in nearby low-wage countries. In contrast, links with actors external to the cluster are considered increasingly crucial. In particular, relationships with main customers are regarded as very important by 65 per cent of the sample ﬁrms and important by 15 per cent and those with high-fashion companies are very important for 70 per cent of the sample ﬁrms working as subcontractors and important for 13 per cent (Table 9.4). In summary, external linkages have recently assumed a greater importance while relationships within the cluster have become less important, except for links with suppliers which remain crucial. Among the entrepreneurs interviewed there is a diﬀused feeling of disillusionment regarding local cooperation and the possibility of repositioning themselves in the world market through joint initiatives.18 Nevertheless, there have been a few joint projects recently, among them the opening of a showroom for 13 ﬁrms in New York, sponsored by ACRIB with 50 per cent of its cost funded by public subsidies, has been particularly successful. Boosted by the euro devaluation of recent years, this initiative has been a real success for at least seven partners with more than 200 000 pairs sold by the showroom in its ﬁrst year of activity, compared to an initial objective of 50 000. Following this successful experience, a new project has been established to set up a trading company in joint venture with a local partner in Shanghai, to sell to the top segment of the Chinese market. This project should also receive ﬁnancial support from the Veneto regional government and from the Italian Ministry of External Trade. Another project sponsored by ACRIB is the setting up of an electronic information network connecting shoe enterprises, suppliers of components, retailers and some local branches of banks. The e-commerce initiative, which is in a pilot phase, is aimed at supplying business-to-business services.

240

65.0 7.5 15.0 65.0 70.0

12 7 15 6 2

30.0 17.5 37.5 15.0 13.0

2 28 16 4 3

5.0 70.0 40.0 10.0 17.0

Source:

Author’s survey.

Note: * 17 sample ﬁrms are working as subcontractors for high-fashion companies. The percentages are calculated on the 17 ﬁrms which answered the questionnaire.

26 3 6 26 12

0 2 3 4 0

Very important % of Important % of Unimportant % of Missing (no. of ﬁrms) sample ﬁrms (no. of ﬁrms) sample ﬁrms (no. of ﬁrms) sample ﬁrms no. of ﬁrms

Importance of linkages according to sample ﬁrms

Suppliers Shoe enterprises ACRIB Main customers High-fashion companies*

Table 9.4

Globalization, industrial districts and value chains

241

Furthermore, there is a regional government project19 involving ACRIB and other local institutions that trains immigrants from outside Europe, providing them with the essential services (that is, housing, residence permits, visas for their families) needed to facilitate their integration into the local community. This initiative addresses a crucial issue in Brenta; given that the district is located in a region characterized by full employment and the footwear industry is not considered very appealing in terms of pay and working conditions, there is a structural shortage of skilled labor. Availability of highly qualiﬁed workers was traditionally one of the main advantages of ﬁrms located in industrial districts like Brenta. Consequently, the diﬃculty in ﬁnding skilled workers is having a negative impact on the district by increasing labor costs, decreasing ﬂexibility and becoming an obstacle to further expansion. This is why local initiatives addressing this constraint are particularly important for the future of Brenta. To conclude, there has been a change in the relative importance of local and global linkages. Vertically, there is the weakening of local linkages, that is, backwards with subcontractors due to increasing decentralization of production to Romania and other nearby countries and forwards due to increasing presence in the district of high-fashion companies. Horizontally, there are a few collective initiatives promoted by ACRIB but compared with the past, Brenta enterprises are investing more in their external linkages with buyers or high-fashion companies than in local relationships.

8.

CONCLUSIONS

This chapter studied the impact of global transformations on local competitiveness in industrial districts. This was done by integrating the typical industrial district approach with the global value chain approach. Primary data on one case study, the Brenta shoe district, were used to address the following questions. Is globalization pushing industrial districts toward new value chains? What types of governance characterize the relationships between local and outside actors? Do the chains’ leaders come from inside or outside the districts? Does the integration of industrial clusters in global value chains enhance or weaken local upgrading strategies? This section presents the main conclusions. First, in Brenta most of the shoe enterprises feed into a variety of chains. If we exclude from our sample ﬁrms producing more than 50 per cent of their production as subcontractors of high-fashion companies (22.5 per cent of the sample) and enterprises which sell more than 50 per cent of their production to large buyers (17.5 per cent) we are left with 60 per cent of

242

Agglomeration in Italy

ﬁrms combining several diﬀerent chains. Therefore, our ﬁrst conclusion is that Brenta shoe enterprises often face several exit options, and dependence on one value chain is limited to a reduced number of cases. Second, the top brand value chain is of recent origin. Its increasing importance follows a global boom of luxury, branded goods in clothing, leather accessories, shoes, perfumes and so on. This is a cross-sector global value chain, characterized by very high entry barriers, given the costs of creating and maintaining a global brand. It can be deﬁned as a subtype of Gereﬃ’s buyer-driven value chain, with some peculiarities. Third, to be part of this chain, Brenta’s shoe producers accept a functional ‘downgrading’ by abandoning design and sales, which are the key competencies of the leaders of the chain, and focusing on production. Their relationships with top brand companies can be deﬁned as somewhere in between network and quasi-hierarchy, but it is clear that the leaders of the chain are not located in Brenta. This story draws attention to at least two new insights, that is, not only upgrading but also downgrading can occur within global value chains and this may happen even to leading producers of developed countries. Fourth, these ﬁrms, which have given up their design function and have become subcontractors, perform better than the other local producers in terms of sales and proﬁts as well as process and product upgrading. It appears that the luxury brand companies share some of their high rents with their skillful subcontractors. Fifth, furthermore, although a process of functional downgrading is occurring at district level, by taking a more systemic perspective our conclusions are diﬀerent. The Italian high-fashion luxury industry is generally undergoing a process of functional upgrading and concentrating investments in rent-rich activities linked with intangible characteristics of the products. The global leaders in the chain exploit cross-industry economies of scale and scope in branding, marketing and advertising. Most of the global leaders of this value chain are Italian companies. Therefore, if the Brenta district is experiencing functional downgrading the Italian fashion system, in contrast, is experiencing functional upgrading. Sixth, the integration of Brenta into the top brand value chain is causing conﬂicts within the district: some producers use their internal production capacity to satisfy the demand of shoes coming from high-fashion companies and outsource (often to Romania) the production of their own sample set with a decrease in quality. Nonetheless, these changes occurring in the district can also be interpreted as a form of functional upgrading, with Brenta enterprises moving the less proﬁtable production abroad in order to use their internal capacity to satisfy the demand from the more proﬁtable top brand chain.

Globalization, industrial districts and value chains

243

Seventh, there are also conﬂicts arising between both linkages within the district and outside it. Comparing our ﬁndings with a previous survey in the early 1990s, it is clear that Brenta enterprises are now attributing less importance to relationships with other local ﬁrms than before. Backward and forward local linkages are weakening, while external linkages with buyers and top brand companies have recently assumed a major role. Eighth, the conclusion that can be drawn is that intangible activities are increasingly becoming the major assets in the top brand industry. In the past, Italian industrial districts, such as Brenta, have built their excellence on a mix of skills in design, fashion and production but the small size of ﬁrms has limited their capability to face the massive investments required to control intangibles in the global market. These intangible activities have become the core competencies of a few large cross-industry companies dominating the top brand value chain, which is assuming a leading role in Brenta. Its expansion allows local footwear ﬁrms to continue exploiting their traditional comparative advantage of highly skilled producers, maintaining a good performance. Nevertheless, the top brand oligopolies, with their huge proﬁts and large ﬁnancial capital availability, are the global leaders of this segment of market. Finally it is not clear what this will mean for the future of Brenta. Local ﬁrms are winning a place in the rent-rich global top brand market but by focusing only on their production skills they oﬀer capabilities which can increasingly be found in other clusters in the world, and this may slowly erode their competitiveness and independence. To date, a mix of factors prevent luxury companies from searching for alternative subcontractors in countries such as Romania or Brazil. They include the size of rents, which downplays the cost factor, and their lack of experience in the shoe sector. These factors induce them to search for very skillful subcontractors that are already able to produce for the top-quality market, and, of course, the higher transaction costs involved in the relationships with more distant and less-qualiﬁed subcontractors. Nevertheless, these conditions may change and the dynamic comparative advantage of Brenta’s producers may one day vanish.

NOTES *

The chapter presents some of the ﬁndings of research undertaken by the author within a project of the Institute of Development Studies at the University of Sussex and the Institute for Development and Peace at the University of Duisburg. I am very grateful to Hubert Schmitz, John Humphrey, Raphie Kaplinsky, Jorge Katz and participants at workshops held at the Institute of Development Studies and at the University of Molise for providing valuable comments on earlier drafts of the study. The shortcomings are my responsibility alone.

244 1. 2. 3. 4.

5. 6.

7. 8. 9.

10.

11. 12. 13.

14. 15.

16. 17.

Agglomeration in Italy By typical industrial district we refer to the Marshallian concept and particularly to its Italian variant. For a collection of classical papers see Pyke et al. (1990). The need for using the market channel approach to study industrial clusters was ﬁrst advocated by Knorringa (1999) in his work on Agra, a footwear district in India. For a summary of the argument and evidence see Schmitz (1995). Markusen (1996), broadening the deﬁnition of an industrial district, discusses four types of districts. In the ‘satellite platform’ type, consisting of a congregation of branch facilities of externally based multi-plant ﬁrms, she acknowledges the importance of external linkages. For a detailed analysis of the main characteristics of the Brenta district, also see Rabellotti (1997). In another paper (Rabellotti 2001), we present an analysis of the main characteristics of the German chain. A key feature of this is that independent retailers are organized in large, powerful buying groups. These groups are network organizations supplying credit and information to their members, helping them to reduce transaction costs and risks. In Japan the tariﬀ on imported shoes ranges from 23.6 per cent to 41.3 per cent for leather footwear. In the EU it ranges from 4.6 per cent to a maximum of 8 per cent. Sixty-six per cent of the sample ﬁrm’s main customers originate in Europe, 11 per cent in the USA, 6 per cent in Italy and the remaining 17 per cent constitutes the rest of the world. In Italy among the top 10 companies in the luxury industry there is only one footwear ﬁrm – Tod’s, ranking at 9th position with a sales value of US$202 million. It originated in Marche, the largest Italian footwear cluster, and was recently listed on the Milan Stock Exchange. According to the local entrepreneurial association, in 2000 the amount of production by Brenta’s enterprises as subcontractors to high-fashion companies reached 50 per cent of total production in the area (personal communication with the director of ACRIB). If these estimates are correct, subcontractors are underrepresented in our sample. There is a distinction between creative and technical design; in this chapter the former is simply called ‘design’ while technical design, including size developing, is called ‘product development’. In the rest of this section, the proportion of sample ﬁrms is intended as a ratio of the 17 enterprises which work as subcontractors to high-fashion companies. The index of performance is generated by attaching equal weight to each performance variable. The ﬁve possible values – from strong increase to strong decrease – were coded on a range of +2 to – 2, respectively, with no change coded 0. The index for each ﬁrm was then constructed by adding up the actual values and dividing them by the number of variables. To test the robustness of this indicator, an alternative performance indicator was also obtained with a principal component analysis, estimated as a linear variable. The correlation coeﬃcient between the two indicators is 0.9 combination of the original (signiﬁcant at 0.01 level). A summary of rent theory in the history of economic thought, with a particular focus on land, is presented in Camagni (1992). According to some sector experts, acquiring licenses could be an alternative strategy to get into the luxury value chains. This strategy was adopted in the 1970s by one of the leading Brenta ﬁrms and has been rather successful in the past. This ﬁrm owns seven licenses of globally known brand names such as Yves Saint Laurent and Calvin Klein and it takes full care of their design and distribution. Nevertheless, there is a recent trend among fashion leaders to reacquire licenses. For example, one of the licenses of this named ﬁrm was not renewed after many years. Furthermore, in 2001, the French group LVMH bought 45 per cent of the shares of this same enterprise A detailed description of their main activities can be found in Rabellotti (1997). Most of the suppliers of components are located in Brenta. In contrast, raw material suppliers are typically located in other clusters that specialize in leather production. Finally with regard to machinery, production is concentrated in Vigevano, an old footwear cluster transformed into a mechanical district, but in Brenta there are usually local dealers with whom shoe ﬁrms interact (Rabellotti 1997).

Globalization, industrial districts and value chains 18. 19.

245

A similar result was found by Schmitz (1999) in the largest Brazilian shoe cluster, the Sinos Valley, where producers with the closest ties to global buyers were least interested in collective local initiatives. This project is part of a local agreement (patto territoriale), an instrument of industrial and infrastructural policy at local level recently introduced in Italy and much implemented in the southern regions.

REFERENCES ACRIB (various years), Rilevamenti statistici, Strá: Associazione Calzaturiﬁci della Riviera del Brenta. ANCI (various years), L’industria calzaturiera italiana, Milan: Associazione Nazionale Calzaturiﬁci Italiani. Camagni, R. (1992), Economia Urbana, Rome: La Nuova Italia Scientiﬁca. Dolan, C. and J. Humphrey (2000), ‘Governance and trade in fresh vegetables: the impact of UK supermarkets on the African horticulture industry’, Journal of Development Studies, 37 (2), 147–73. Gereﬃ, G. (1999), ‘International trade and industrial upgrading in the apparel commodity chain’, Journal of International Economics, 48, 37–70. Humphrey, J. and H. Schmitz (2000), ‘Governance and upgrading: linking industrial cluster and global value chain research’, IDS Working Paper 120, Institute of Development Studies, Brighton: University of Sussex. Humphrey, J. and H. Schmitz (2002), ‘Developing country ﬁrms in the world economy: governance and upgrading in global value chains’, INEF (Institut für Entwicklung und Frieden) report, no. 61, Institute for Development and Peace, Duisburg: Univerity of Duisburg. ISTAT (1996), Industrial Census, preliminary results available at www.istat.it/. Kaplinsky, R. (1998), ‘Globalisation, industrialisation and sustainable growth: the pursuit of the Nth rent’, IDS Discussion Paper 365, Institute of Development Studies, Brighton: University of Sussex. Kaplinsky, R. (2000), ‘Spreading the gains from globalisation: what can be learned from value chain analysis?’, IDS Working Paper 110, Institute of Development Studies, Brighton: University of Sussex. Kaplinsky, R. and J. Readman (2001), ‘How can SME producers serve global markets and sustain income growth?’, University of Sussex and University of Brighton, mimeo. Knorringa, P. (1999), ‘Agra: an old cluster facing the new competition’, World Development, 27 (9), 1587–604. Markusen, A. (1996), ‘Sticky places in slippery space: a typology of industrial districts’, Economic Geography, 72, 293–313. Mytelka, L. (2000), ‘Local systems of innovation in a globalized world economy’, Industry and Innovation, 7 (1), 15–32. Pambianco (2002), ‘Osservatorio delle principali operazioni di merger & acquisition avvenute nei settori del Made in Italy a livello mondiale’, mimeo, www. pambianco.com/. Pyke, F., G. Becattini and W. Sengenberger (eds) (1990), Industrial Districts and Interﬁrm Cooperation in Italy, Geneva: International Institute for Labour Studies, ILO.

246

Agglomeration in Italy

Rabellotti, R. (1997), External Economies and Cooperation in Industrial Districts. A Comparison of Italy and Mexico, London: Macmillan. Rabellotti, R. (2001), ‘The eﬀect of globalisation on industrial districts in Italy: the case of Brenta’, IDS Working Paper, No. 144, Institute of Development Studies, Brighton: University of Sussex. Schmitz, H. (1995), ‘Collective eﬃciency: growth path for small-scale industry’, Journal of Development Studies, 31 (9), 1627–50. Schmitz, H. (1999), ‘Increasing returns and collective eﬃciency’, Cambridge Journal of Economics, 23 (4), 465–83.

10. The competitive advantage of a region: industrial districts in Emilia-Romagna Enrico Santarelli 1.

INTRODUCTION

This chapter investigates some crucial aspects of the recent development of industrial districts in the Emilia-Romagna region of Italy – one of the most developed and dynamic areas in the country, with a GDP per capita and an annual growth rate of GDP above the national average – where this type of spatial agglomeration of industrial ﬁrms has ﬂourished since the period immediately after the Second World War. Industrial districts are so intimately bound up with modern economic growth in this region that the typical organization of industrial activities characterized by the widespread presence of such spatial clusters and of small and medium-sized enterprises (SMEs) is usually labeled, after Brusco (1982), the ‘Emilian model’. The chapter is organized as follows. First, Section 2 shows that an identiﬁcation problem still aﬀects the analysis of industrial districts, since six diﬀerent sources provide six diﬀerent deﬁnitions and maps of industrial districts. Comparison of the technological and organizational features of two of the most important industrial districts in the Emilia-Romagna region, the biomedical and the ceramic tile districts, oﬀers some useful hints for identiﬁcation of the sources of competitive advantage of these typical location clusters (Section 3). In general, the interaction between dynamic industries at the local level – namely, an industry producing for the ﬁnal market and an industry producing specialized machinery for the other industry – gives rise to a virtuous circle in which the growth performance of one industry reinforces the growth performance of the other, on the basis of technological links, market opportunities and input sharing. In-depth analysis of the biomedical cluster allows conclusions to be drawn about knowledge transfer and the development of product innovations in a district characterized by the presence of multinational corporations (MNCs) and an increasing amount of inward foreign 247

248

Agglomeration in Italy

direct investment (FDI). In-depth analysis of the ceramic tile cluster conﬁrms the importance of large ﬁrms in the recent development of local agglomerations, whereas comparison of the economic performance of a sample of ﬁrms in the ceramic tile district and one in the whole industry in Italy shows that over the 1998–2000 period the former performed better than the latter in terms of average annual growth rate of the value of total sales, whereas non-district ﬁrms displayed a most favorable dynamic in terms of net income, return on equity and cash ﬂow. On this basis, Section 4 undertakes an econometric analysis aimed at comparing the technological strength (in terms of patents registered with the European Patent Oﬃce) of innovative ﬁrms located within and outside industrial districts, in order to determine whether the prediction that innovative activity favors those ﬁrms or industries with direct access to knowledge-producing inputs also applies to the case of industrial districts. The analysis deals with the population of ﬁrms whose headquarters are in the Emilia-Romagna region, which registered at least one patent with the European Patent Oﬃce during the 1986–95 period. Results from panel model estimates show that being located within an industrial district resulted in a technological advantage during the overall 1986–95 period. However, when this period is broken down into two subperiods (1986–90 and 1991–95) it was found that such advantage was strong in the ﬁrst one, whereas it was lost in the ﬁrst half of the 1990s. The concluding section makes some comments on the future of spatially concentrated industrial districts vis-à-vis the diﬀusion of information and communication technologies (ICTs).

2. ALTERNATIVE LANDSCAPES: A COMPARISON OF SIX DIFFERENT MAPS OF INDUSTRIAL DISTRICTS The idea of the industrial district can be traced back to Alfred Marshall’s Principles of Economics (1890), when he compares the advantages of vertically integrated, large ﬁrms with those of spatially concentrated, small ﬁrms beneﬁting from division of labor and network externalities. However, the modern deﬁnition of an industrial district, and the revival of this unit of investigation in economic analysis is due to the Italian economists Giacomo Becattini (1979, 1990) and Sebastiano Brusco (1982, 1986), who respectively describe the industrial district as (i) a local system characterized by the active integration between a community of people and a community of industrial ﬁrms;1 and (ii) a ﬂexible specialization system – typical of the Emilia-Romagna, Marche, Tuscany and Veneto regions of

The competitive advantage of a region: Emilia-Romagna

249

Italy – characterized by the widespread presence of ﬁrms with fewer than 200 employees which, by subcontracting many stages of production to other (equally small) ﬁrms, are able to mobilize a labor force 10 times larger than the labor force on their pay roll. These local systems are characterized by a strong incentive to invest in advanced production machinery, which is usefully employed thanks to a strong polarization of skills. Accordingly, Brusco (1986, p. 90) identiﬁes a further and highly signiﬁcant feature of industrial districts, namely ‘the presence, in an area that produces a certain commodity, of ﬁrms that produce the machinery necessary for the production of the commodity’ (italics added). There is a main implication concerning the economic function of SMEs in industrial districts, namely that through the support of the network embedded in the industrial district, the small ﬁrm is able to oﬀset any sizeinherent cost disadvantages (see Audretsch et al. 1999). As pointed out by Grabher (1993), the social structure underlying local clusters contributes to the viability of small ﬁrms that would otherwise be vulnerable if they were operating in an isolated context. This view, which stems from Piore and Sabel’s (1984) work, depicts industrial districts as the result of a pathdependent process including a past of handicraft production, widely shared social values and trading links to foreign markets. In eﬀect, in such an environment, the various ﬁrms – either producing a given commodity or the machinery necessary for its production – are linked together by a complex network of external economies and diseconomies, practices of cost sharing, and sequences of historical events which shape both inter-ﬁrm and interpersonal relationships. This gives rise to the emergence, within the same industry, of asymmetric productive structures among regions based on the adoption of quality-improving technologies (Giovannetti 2000). This ‘romantic’ portrait, mostly centered around the key role played by SMEs in traditional industries substantially immune from competition by mass-production industries, has partly been changed following the works by Gianni Lorenzoni and his co-authors (Lorenzoni and Ornati 1988; Lazerson and Lorenzoni 1999), who focused in-depth on the recent evolution of industrial districts. Lorenzoni contends that ‘focal ﬁrms’ – deﬁned as those ﬁrms that occupy strategically central positions in the industrial district thanks to the great number of relationships that they have with both customers and suppliers – look decisive in expanding the district’s horizons by enabling incorporation of new technologies, organizational skills and markets (Lazerson and Lorenzoni 1999, p. 362). These quite large leading ﬁrms are strategic centers that enable the emergence of a form of hierarchy more akin to the generation and transfer of new knowledge (Lorenzoni and Baden-Fuller 1995; Boari and Lipparini 1999).

250

Agglomeration in Italy

Translation of the above socio-economic and organizational deﬁnitions into a statistical procedure for identiﬁcation of industrial districts is not straightforward (Becchetti et al. 2002). In fact, at least six diﬀerent lists and classiﬁcations of industrial districts are currently available, each of them provided by a diﬀerent source: ISTAT (1997), Censis (2001), CNEL/CerisCNR (1997), Club dei Distretti (1999), Fondazione G. Brodolini (1996) and Il Sole-24 Ore (Moussanet and Paolazzi 1992). The methodology developed at Statistics Italy (ISTAT 1997) is based on previous work by Sforzi (1985), who extended Becattini’s original intuition. According to this approach, which emphasizes the importance of SMEs in industrial district development, identiﬁcation of industrial districts proceeds through the following stages: 1. 2.

3. 4.

identiﬁcation of local systems specializing in manufacturing; identiﬁcation of local systems specializing in manufacturing that are dominated by SMEs (with fewer than 250 employees, according to the EU classiﬁcation); identiﬁcation of the main industry in local systems specializing in manufacturing that are dominated by SMEs; and identiﬁcation, as industrial districts, of those local systems specializing in manufacturing that are dominated by SMEs within which the main industry is also dominated by SMEs.

Application of this procedure leads to the identiﬁcation of 24 districts in the Emilia-Romagna region (see Table 10.1). These include the biomedical district of Mirandola and the ceramic tile district of Sassuolo, which will be analysed in the next section, both covering a wide range of specialization: food, paper and printing, chemicals, rubber and plastics, mechanical engineering, toys, leather and leather products, household appliances, textiles and clothing. The CNEL/Ceris-CNR (1997) Report identiﬁes industrial districts according to the ﬁve criteria suggested by the Ministry of Industry: (i) a Table 10.1 Number of districts in Emilia-Romagna and Italy according to alternative maps ISTAT Emilia-Romagna Italy Emilia-Romagna as a % of Italy

24 199 12.06

Il Sole-24 Fondazione CNEL/ Club dei Censis Ore G. Brodolini Ceris-CNR Distretti 7 65 10.77

9 100 9.00

11 84 13.10

11 87 12.64

7 51 13.73

The competitive advantage of a region: Emilia-Romagna

251

share of manufacturing employment at least 30 per cent above the national average; (ii) a number of local units with a resident population ratio above the national average; (iii) clear specialization in a given manufacturing production, with a share of employment in the relevant ﬁliere at least 30 per cent above the national average; (iv) a share of employment in the relevant ﬁliere amounting to at least 30 per cent of local employment in manufacturing; and (v) at least 50 per cent of employment in the relevant ﬁliere accounted for by small ﬁrms. The resulting map of industrial districts in Emilia-Romagna identiﬁes 11 local systems displaying such features, the second largest number in Italy, after Lombardia, which ranks ﬁrst with 22 industrial districts. The methodology employed at Fondazione G. Brodolini (1996) combines statistical ﬁgures provided by oﬃcial sources2 with qualitative information originating from ﬁeld research. Interviews with industry insiders enabled us to focus on the evolution of inter-ﬁrm relationships within industrial districts and the existence of a district-speciﬁc system of social rules. From application of this procedure only nine industrial districts are identiﬁed in Emilia-Romagna, corresponding to 9 per cent of those found in the whole country. The other sources summarized in Table 10.1 are less robust from a methodological viewpoint, being more the result of direct surveys and case studies than the outcome of the application of a rigorous methodology. Nevertheless, they yield a picture of industrial districts in Emilia-Romagna that does not contrast markedly with those furnished by ISTAT and CNEL/Ceris-CNR. Unlike most Italian regions, the regional government of EmiliaRomagna has decided not to identify the ‘oﬃcial’ industrial districts present in its territory, despite being obliged to do so in fulﬁllment of Law 317 of 1991 and D.M. 21 April 1993 by the Ministry of Industry. The rationale for this decision is (i) an intention to leave local (either municipal or provincial) authorities free to decide how to organize their territory on the basis of industrial aggregates likely to change quickly according to the dynamics of both external trade and new technology, and (ii) a refusal to adopt the somewhat contradictory identiﬁcation criteria suggested by national laws. In fact, the regional government has pursued a strategy of favoring a market-led process of local agglomerations, in the belief that the clustering processes characterizing the emergence of industrial districts are spontaneous and not policy driven. Irrespective of their total number and productive specialization, industrial districts in Emilia-Romagna are local clusters in which it is possible to identify a ﬁliere organization in which the producers and users of specialized industrial machinery are closely integrated. One of the strengths of

252

Agglomeration in Italy

this networked system is the technological progressiveness of the mechanical engineering industry, which has de facto represented one of the engines of economic growth in the region since the second half of the twentieth century. Distributed along the ‘Via Emilia’, this industry has been the main external source of innovation for small ﬁrms characterized by weak in-house research and development (R&D) and, in general, by a very low level of autonomous innovative capabilities. In eﬀect, if one develops a matrix of inter-industry technology ﬂows linking R&D expenditures to patented innovations originating in the mechanical engineering industry and used by small ﬁrms in consumer goods industries, the latter exhibit a strong relationship between the technological capabilities employed in the production processes and economic performance (see Rosenberg 1976; Santarelli and Sterlacchini 1994). These inter-industry spillovers are important for both research-intensive and mature or low-tech industries. Although at odds with part of the literature on the geography of spillovers (see, among others, Henderson et al. 1995; Kim 1995; Audretsch and Feldman 1996) – emphasizing that spillovers are conﬁned to industries where new economic knowledge plays a greater role, that is, high-tech industries – this ﬁnding has been conﬁrmed for the industrial districts of Emilia-Romagna by Forni and Paba (2002). Today, the mechanical engineering industry accounts for approximately 8.1 per cent of total employment3 and 8.9 per cent of value added4 in the region. Its integration with the local industries which use the specialized industrial machinery it produces is very tight and, thanks to the knowledgespreading capability of embodied technological change, it has important implications for local development. As Forni and Paba (2002) have shown, local systems specialized in these interconnected sectors are more successful than local systems specialized in only one production.

3. TECHNOLOGICAL CHANGE AND ECONOMIC PERFORMANCE IN THE INDUSTRIAL DISTRICT: EVIDENCE FROM CASE STUDIES The Biomedical District of Mirandola As aptly pointed out by Biggiero (2002), the biomedical district of Mirandola represents a challenge to the commonly held view of industrial districts in Italy. In fact, it diﬀers from the traditional industrial districts with regard to its two main features: (i) it is specialized in high-tech activities (namely, the production of health-care products, with the exception of those for pharmaceutical use); (ii) it was started relatively recently, in 1963.

The competitive advantage of a region: Emilia-Romagna

253

This district specializes in the production of disposable sterilized products (and related equipment) for hemodialysis infusion and extracorporeal circulation in general (47 per cent of value added) – an activity in which it is the international leader – cardio-surgical devices (16.2 per cent) and sterilized disposable products for respiratory use in anesthesia and corporeal reanimation (13.2 per cent). It is localized in 15 municipal areas comprising the territory of the province of Modena. More than 50 per cent of ﬁrms (36 out of 70) and 90 per cent of employment are in the area of Mirandola, where the district originated. With respect to the industry as a whole, the Mirandola district accounts for more than 16 per cent of total employment in Italy. Approximately 50 per cent of local ﬁrms are small subcontractors, all of them with fewer than 50 employees. Among ﬁrms producing for the ﬁnal market, ‘local’ ones, which account for 70 per cent of the total number of ﬁrms of this type, all have fewer than 50 employees. Total employment in the district is 3660 employees, with total sales exceeding €500 million, of which 59 per cent derive from exports (mostly to the other EU countries and the United States). Exports are led by the excellent performance of the leading ﬁrm (the Swedish-owned Gambro-Dasco), which is expanding its market shares in non-EU European countries, and increased its total sales by 30 per cent between 1997 and 2000 (R&I 2001). (See Table 10.2.) A further peculiarity of the Mirandola district with respect to the traditional view of industrial districts as comprising small ﬁrms that developed in opposition to large ﬁrms is the presence within it of MNCs and large national companies. These companies have taken over the most important ﬁrms in the Mirandola area and specialize in the production of both disposable goods and machinery. Thus, contrary to the usual idea of FDIs as driven by the availability of credit facilities, reduced labor costs or foreign market penetration, in the case of the Mirandola biomedical district, foreign ﬁrms making acquisitions have been attracted by the possibilities of accessing locally available skills, technology and know-how. This phenomenon has resulted in a relatively high degree of seller concentration, with Table 10.2

The biomedical industry in the Mirandola district

Total sales (billion lira) % of sales from export Number of ﬁrms of which subcontractors Number of employees Source: R&I (2001).

1997

2000

775 49.8 74 39 3209

998 60.7 70 35 3660

2001 1104 57.0 71 35 3941

254

Agglomeration in Italy

the top four producers (Gambro-Dasco, Mallinckrodt, B. Braun Carex and Bioﬁl) accounting for 63 per cent of employment and 73 per cent of total sales. Among the 10 non-district ﬁrms (only four of which are domestic ﬁrms) six have between 50 and 249 employees, whereas four have more than 250. These ﬁrms account for nearly 83 per cent of total production, and employ 75 per cent of the total workforce. They entered the district after 1980, through the acquisition of incumbent local ﬁrms (all founded by Mario Veronesi together with two partners5), unable to withstand competition by large ﬁrms because of their ﬁnancial and technological weakness. Consistently with the view of leading ﬁrms put forward by Lorenzoni and his co-authors, entry by MNCs and large national companies fostered the adoption of process innovations (such as the introduction of the ﬁrst Computer Aided Design (CAD) systems) and quality control procedures, as well as the more careful selection of materials. In addition, it brought in the synergies and the superior coordination skills of the group organization, and eventually acted as a driver for innovation and growth of all ﬁrms in the district (Boari and Lipparini 1999). The quality-upgrading eﬀect resulting from the emergence of these leading ﬁrms and groups set in motion a learning process among local ﬁrms and subcontractors, which in turn made major improvements to their procedures. The Mirandola district operates as a group of companies able to take a product from design to prototype to development of specialized machinery, to production, and beyond. Dozens of highly specialized ﬁrms oﬀer services such as molding, extrusion, subcontracting, assembly, sterilization, instrument manufacture and consulting (see Lichtman 2002). Product innovation is the crucial competitive factor for ﬁrms in this industry. However, since only 43 per cent of ﬁrms in the district produce for market clients, the innovation process in the Mirandola area is mostly the result of close cooperation and interaction among ﬁrms, characterized by the presence of a hierarchical structure within which larger ﬁrms promote the achievement of higher levels of eﬃciency and competitiveness (Boari and Lipparini 1999). Thus, not only are the largest biomedical ﬁrms actively involved in innovative activities, but they also include both local independent ﬁrms and subcontractors in the overall innovation process. The resulting local system of innovation is one in which MNCs and large national companies control the strategic phases of R&D, design, production of machinery with embodied technological change, and ﬁnal control, whereas small local ﬁrms are responsible for the intermediate phase of production and handle the assembly process. Crucial in this networking process is the production of specialized machinery and devices for specialized machinery which set oﬀ a learning process involving all the players (either foreign or domestic, producing for either the ﬁnal market or

The competitive advantage of a region: Emilia-Romagna

255

subcontractors) in the biomedical district. In this process, production machinery – mostly for the assembly of plastic disposables – represents ‘dedicated assets’ in Williamson’s (1985) sense, namely resources designed for speciﬁc purposes and which cannot easily be redeployed to alternative uses. In eﬀect, the innovative process technologies employed in the Mirandola district are so speciﬁc to the production of certain disposable goods that their design and development requires close cooperation among all the ﬁrms involved in the local biomedical ﬁliere. Interaction among producers and users of machinery and capital equipment is therefore a factor favoring the creation of speciﬁc knowledge which contributes to the overall technological competitiveness of the district. It is therefore not surprising that, according to Ceris (see CNEL/Ceris-CNR 1997), the second specialization of the Mirandola district is in the production of specialized industrial machinery. To summarize, the Mirandola district displays the features of what Lipparini and Lomi (1999) called an ‘organizational community’, one in which the district’s various areas of competence constitute a sort of tacit knowledge which diﬀuses among all local players. The glue that joins everything together and enables the circulation of information among all ﬁrms in the local arena is the coordination skills brought in by the advent of exogenous forces (multinationals from other countries and large national companies) and interaction between producers and users of machinery. In this case, too, as is typical of the Emilia-Romagna districts, exchange of information among players within the focal ﬁliere and embodied technological change are the engines of knowledge dissemination at the local level. And this is consistent with the original idea of the industrial district. As implied in Marshall’s original formulation of ‘external economies’, spillovers do not stem from producers of similar products but are related to the input–output or customer–supplier relationships that arise from interaction between ﬁrms producing specialized capital equipment and machinery and ﬁrms using those devices (Forni and Paba 2002). The case of the ‘biomedical valley’ in the Emilia-Romagna region proves that it is the ‘right’ agglomeration of industries at local level that is crucial for industrial district development. The Ceramic Tile District of Sassuolo When comparing the distinctive features of two industrial clusters that dominate the global ceramic tile industry – Sassuolo in Italy and Castellón in Spain – Meyer-Stamer et al. (2001) stress that whereas it is the capital goods producers that drive technical change and innovation in the Italian district, what drives competitive advantage in the Spanish one is innovation

256

Agglomeration in Italy

in downstream activities. Meyer-Stamer et al.’s paper also contributes to the reassessment of the competitiveness of tile clusters in the developing world. Tile ﬁrms in Brazil’s leading cluster, located in Santa Catarina, beneﬁt from the ﬁerce rivalry among Italian producers (most of which are located in the Sassuolo district), among Spanish producers and between Italian and Spanish producers. Although Brazilian ﬁrms are technology followers, they are innovative in downstream activities, experimenting with concepts which are not yet used by Italian or Spanish manufacturers. It is well known that Italy has been the leader in the ceramic tile industry since the Middle Ages. Today, the industry is mostly located around the town of Sassuolo, in the province of Modena. The industry is made up of companies of various sizes, most of them SMEs, ranging from small craft enterprises producing hand-made products according to centuries-old traditions to large publicly traded corporations producing the latest in porcelain material. According to Assopiastrelle, the employers’ association for the Italian tile industry, Italy accounts for 40 per cent of the entire world trade in ceramic tiles, employing approximately 37000 people and manufacturing more than 630 million square meters of tiles annually in nearly 600 ﬁrms. The industry is characterized by a relatively low level of concentration, with the largest ﬁve ﬁrms accounting for about 17 per cent of total production (see Prometeia 2002).6 Only 29 ﬁrms have total sales exceeding €50 million (21 of which are in the Sassuolo district!), whereas 353 micro ﬁrms fall below the €2 million threshold. Italian producers of ceramic tiles are deeply integrated in international trade, with 70 per cent of total sales represented by exports to foreign markets. According to CNEL/Ceris-CNR (1997), the second specialization of the Sassuolo district is the production of specialized machinery for the ceramic tile industry. The Sassuolo district started up during the 1950s as an industrial agglomeration within which ﬁnal ﬁrms were also directly involved in the development and reﬁnement of production machinery and raw materials. However, it was only in the 1960s that, as a consequence of the specialization and division of labor among district ﬁrms, a group of specialized suppliers of machinery and capital equipment came into being (see Russo 1985). Another important organizational change in the industry occurred during the late 1980s, when the leading ﬁrm7 and the industrial group became the main forces of growth in the local system. Nowadays, this industrial cluster still maintains its leadership in the production of ceramic tiles, although its recent economic performance has been only slightly better than that of other Italian ﬁrms not located in the same area. In fact, comparison of the economic performances of two samples of ceramic tiles producers, one located in the Sassuolo district, the other outside the district, gives a somewhat controversial picture (Table 10.3),

257

The competitive advantage of a region: Emilia-Romagna

Table 10.3 The economic performances of Italian ﬁrms in the ceramic tile industry (% of total sales)

District ﬁrms (Sassuolo) Sales Net income (%) Gross operating surplus (%) ROE (after tax) Fixed investments (%) Cash ﬂow (%) Non-district ﬁrms Sales Net income (%) Gross operating surplus (%) ROE (after tax) Fixed investments (%) Cash ﬂow (%)

1995

1996

1997

1998

1999

2000

15.5 4.6 16.2

1.5 2.0 11.8

6.0 2.2 12.5

7.1 2.8 12.7

5.6 3.4 13.3

11.0 1.9 11.7

12.8 10.4 11.2

5.3 6.5 8.2

5.9 5.2 8.6

7.7 8.0 8.7

9.3 7.1 9.5

5.3 5.9 8.2

14.0 5.1 16.2

0.3 3.0 12.9

5.7 3.1 13.6

3.2 4.0 14.4

5.2 4.5 15.7

8.6 3.1 14.1

12.3 9.7 10.6

6.6 7.2 8.9

7.1 4.6 9.2

8.8 5.7 9.9

9.7 7.9 10.4

6.5 6.1 10.1

Source: Prometeia (2002).

although within a generally positive framework. Both samples exhibit favorable dynamics of total sales, with district ﬁrms performing on average better in the four ﬁnal years of observation. Conversely, non-district ﬁrms performed better than ﬁrms in the Sassuolo district in terms of net income, gross operating surplus, ROE (Return on Equity) and cash ﬂow. District ﬁrms were instead characterized by a more aggressive investment strategy which should result in a greater likelihood that they will beneﬁt from embodied technological change more than non-district ﬁrms in the following years. Nevertheless, the non-substantially diﬀerent economic performances of district and non-district ﬁrms is probably also connected to the capacity of the former to leave the narrow boundaries of the local system and become more integrated with the latter. This might be an indirect indication of the emergence of the ‘multi-located’ industrial district as a productive aggregate in which an appropriate network of suppliers, sectoral externalities, contracting and subcontracting with other ﬁrms belonging to the same or related industries, spillovers and knowledge originating from outside the local system are instrumental in determining the path and intensity of innovation activities carried out by industrial ﬁrms within a given portion of territory.

258

Agglomeration in Italy

This kind of agglomeration might have been favored by the diﬀusion of faster means of communication enabled by the ICT revolution, which provided viable alternatives to the various kinds of face-to-face communication that characterized the spatially concentrated industrial district. In this connection, the transaction cost advantages resulting from exploitation of ICT enable the relocation process of productive activities to be implemented without determining any signiﬁcant additional cost for the ﬁrm.

4. TECHNOLOGICAL CHANGE IN THE INDUSTRIAL DISTRICT: AN ECONOMETRIC ANALYSIS This section explores whether, for ﬁrms already able to realize and patent their innovations abroad, being located within an industrial district is a factor positively aﬀecting innovative capability. The IMPERO database developed at Aster (the Agency for Technological Transfer of the EmiliaRomagna region) was employed for this purpose. IMPERO contains micro-level data on patenting activity by ﬁrms located in the region, including balance sheet ﬁgures and a full range of qualitative and quantitative information. Patents, Firm Size and Firm Location Analysis of the innovative performance of ﬁrms in the Emilia-Romagna region took account of all ﬁrms with at least one patent registered with the European Patent Oﬃce (EPO). In particular, the analysis dealt with the patent activity of ﬁrms in industrial districts compared to a control sample of non-district ﬁrms that had also patented with the EPO. Their limited heuristic value notwithstanding, patents are widely employed in the studies of the output side of the innovation (for a critical survey, see Piergiovanni and Santarelli 1996). In this connection, the use of data comprising ‘highquality’ patents, such as those provided by the EPO, represents a viable alternative to the data collected through the national patent system. Nevertheless, it remains true that ﬁrms of diﬀerent sizes have a diﬀerent propensity to use patent protection and that ﬁrms in traditional industries are more likely to develop non-patentable innovations than are ﬁrms in technologically progressive industries. As a consequence, in this more than in other cases, the empirical results are likely to reﬂect the partial inadequacy of the innovation data that are employed. The period considered was 1979 to 1997, which was characterized by the increasing integration of Emilia-Romagna ﬁrms into the global economy,

The competitive advantage of a region: Emilia-Romagna

259

with a signiﬁcant process of partial relocation of manufacturing activities in least-developed and transition economies (see Barba Navaretti et al. 2002). Several ﬁrm-speciﬁc factors were taken into account when attempting to explain inter-ﬁrm diﬀerences in patenting activity: ●

● ● ● ● ● ● ●

the fact that a ﬁrm is/is not located within an industrial district (DIST). The classiﬁcation proposed in the CNEL/Ceris-CNR (1997) report, which is the one most akin to the guidelines developed by the Ministry of Industry was adopted for this purpose; ﬁrm size (total number of employees in each year during the relevant period) (SIZE); total net value of property, plant and equipment (as a proxy for the stock of total ﬁxed assets) (PPE); the dynamics of the gross operating surplus (as a measure of the economic performance of the ﬁrm) (GOS); the fact that the ﬁrm operates with one or more plants (as a proxy for the organizational structure) (MULTI); the fact that the ﬁrm is part of an industrial group (GROUP); the fact that the ﬁrm is an exporting ﬁrm (EXP); and the fact that the ﬁrm belongs to one of the categories (specialized suppliers (SSUP), supplier dominated (SDOM), scale intensive (SINT)) in Pavitt’s (1984) taxonomy.

Firms for which no balance sheet data were available and ﬁrms which exited before the end of the period were dropped from the original list of those with at least one patent with the EPO. Since complete balance sheet data were available only for the years between 1986 and 1995, the econometric analysis performed in the next subsection focuses on this period only, taking the total number of patents granted to the ﬁrm between 1979 and 1985 as the cumulated stock of patents in the base year 1986. The rationale for choosing most of the variables listed above is intuitive. First, the district variable allows one to determine whether the external economies typical of the district do aﬀect the innovative output of the ﬁrm. Accordingly, it was determined for each ﬁrm localized in a municipality within a certain industrial district whether the ﬁrm’s productive specialization was the same as that characterizing the industrial district. Second, employment size was included in the analysis in order to seek conﬁrmation for the so-called ‘second Schumpeterian hypothesis’, according to which innovative capability increases with ﬁrm size.8 Third, the value of property, plant and equipment is both a measure of ﬁrm scale and the stock of capital (including intangibles such as software) which can usefully be employed in the production process under the

260

Agglomeration in Italy

hypothesis that the larger the stock of capital the higher the innovative capability of the ﬁrm. Fourth, the gross operating surplus is a measure of the wealth produced by the ﬁrm once the variable costs have been subtracted. The idea behind the introduction of this variable is that ﬁrms achieving better economic performance are more able to raise the ﬁnancial resources needed to carry the costs connected with patent registration and protection. For this reason, the annual growth rate of the gross operating surplus was used in the econometric analysis rather than its level. Fifth, multi-plant ﬁrms are likely to employ professional managers and to possess more sophisticated organizational capabilities than is usually the case of single-plant ﬁrms. Managerial skills are likely to result in the more eﬃcient organization of the innovation process, with a greater likelihood of obtaining patentable inventions. Sixth, ﬁrms belonging to an industrial group are involved in a process of information sharing that is likely to generate positive external economies. As a consequence, also ﬁrms devoid of autonomous innovative capability may be able to extract patentable innovations from a combination of knowledge freely available within the group. Seventh, export-oriented ﬁrms have to cope with international competition. Thus, in order to obtain larger shares of foreign markets they are forced to undertake innovative activities likely to result in more patents. Eighth, for nearly 20 years Pavitt’s (1984) classiﬁcation of ﬁrms according to their attitude towards innovation has been one of the most widely used taxonomies of innovating ﬁrms. Since all but two (biomedical) of the ﬁrms in the sample used for the present study are in the scale-intensive, specialized-supplier, and supplier-dominated categories of Pavitt’s scheme, the use of dummy variables for such categories may be helpful in identifying whether belonging to one category or another aﬀects the likelihood of obtaining more patents. Results from Panel Model Estimates The availability of longitudinal data allowed estimation of a ﬁxed-eﬀects panel model. Thus, the analysis started by postulating restrictions on the parameters, namely overall homogeneity of both slopes and intercepts. Since this hypothesis was not rejected by the data, the next step was to perform a pooled regression by means of generalized least squares estimators. The functional form of the model was the following: yit *it it xit uit,

(10.1)

The competitive advantage of a region: Emilia-Romagna

261

where i denotes the ﬁrms, t the years comprising the analysis, *it is a 1 1 scalar constant representing the eﬀects of those time-invariant variables peculiar to the ith ﬁrms and it (1it, 2it, . . ., Kit ) is a 1 K vector of constants, and xit (x1it, x2it, . . ., xKit ) a 1 K vector of exogenous variables, the regressors employed in the analysis, and uit is the error term with mean zero and variance 2u. Two diﬀerent methodologies were used to deal with the panel. First, standard panel techniques were used to run model (10.2): PATit o 1DISTit 2SIZEit 3PPEit 4SIZEit 5GOSit 6MULTIit 7GROUPit 8EXPit 9SSUPit 10SDOMit 11SINTit xit i it,

(10.2)

where x is a vector of controls, which can be either time variant or ﬁrm speciﬁc. Model (10.2) assumes the existence of ﬁrm-speciﬁc eﬀects. Thus, also due to the nature of the available data, the ﬁxed-eﬀect technique is the preferred estimation for the model, with inclusion of cross-section weights. Accordingly, model (10.2) was ﬁrst estimated over the entire 10-year period between 1986 and 1995, for which data on 34 ﬁrms with patents registered with the EPO were forthcoming (Table 10.4). The results show a negative impact of location within an industrial district on patenting, and also larger ﬁrm size seems to be an impediment rather than a stimulus to patenting.9 Conversely, belonging to an industrial group, a higher value of property, plant and equipment, and the fact that the ﬁrm operates with more than one plant are all factors that positively aﬀect patenting. The negative and signiﬁcant coeﬃcients of the three dummy variables for Pavitt’s taxonomy instead suggest that patenting is an activity more typical of science-based ﬁrms than of specialized-supplier, supplier-dominated and scale-intensive ones. By breaking down the 10-year period into two subperiods, it was then possible to see whether the determinants of patenting changed from the 1980s to the 1990s (Table 10.5). In this respect, for the two subperiods data were available for a larger number of ﬁrms than in the case of the 10-year period. The value of the coeﬃcient of determination adjusted for the degree of freedom (R2 adjusted) was much higher for the estimate carried out for the 1986–90 period (0.923) than it was for the one concerning the overall 10-year period (0.373). In fact, analysis carried out for 44 ﬁrms between 1986 and 1990 showed that being located in an industrial district is a factor positively aﬀecting patenting at the ﬁrm level, along with holding more than one plant and

262

Table 10.4

Agglomeration in Italy

Panel model estimates, 1986–1995

Variables

Coeﬀ.

Dependent variable: PAT (number of patents) Intercept 4.078996*** DIST (Industrial district) 0.552322** SIZE (Number of employees) –0.001223* PPE (Property, plant and equipment) 6.09E-05*** GOS ( Gross operating surplus) 0.025410 MULTI (Multi-plant) 2.262991*** GROUP (Industrial group) 0.271824* EXP (Export) 0.092256 SSUP (Specialized suppliers) 3.362484*** SDOM (Supplier dominated) 3.231630*** SINT (Scale intensive) 4.137798***

Std. Err.

Prob.

0.935913 0.190322 0.000499 9.19E-06 0.032817 0.305794 0.144838 0.135558 0.933043 0.926493 0.954777

0.0000 0.0040 0.0148 0.0000 0.4393 0.0000 0.0614 0.4966 0.0004 0.0006 0.0000

Number of observations F-test R2 adjusted (overall)

340 21.137 0.373

Note: ***, **, * statistically signiﬁcant at 99%, 95% and 90% conﬁdence levels, respectively.

Table 10.5

Panel model estimates, 1986–1990

Variables

Coeﬀ.

Dependent variable: PAT (number of patents) Intercept 1.632256* DIST (Industrial district) 0.820574*** SIZE (Number of employees) 0.000370 PPE (Property, plant and equipment) 3.11E-06 GOS ( Gross operating surplus) 0.007564 MULTI (Multi-plant) 0.828130*** GROUP (Industrial group) 0.593138*** EXP (Export) 0.847035*** SSUP (Specialized suppliers) 0.629776 SDOM (Supplier dominated) 0.772766 SINT (Scale intensive) 1.320390* Number of observations F-test R2 adjusted (overall)

Std. Err.

Prob.

0.670751 0.129919 0.000389 8.75E-06 0.023257 0.077712 0.126037 0.085835 0.670310 0.667994 0.714658

0.0158 0.0000 0.3430 0.7229 0.7453 0.0000 0.0000 0.0000 0.3485 0.2487 0.0661 240 264.264 0.923

Note: ***, **, * statistically signiﬁcant at 99%, 95% and 90% conﬁdence levels, respectively.

The competitive advantage of a region: Emilia-Romagna

Table 10.6

263

Panel model estimates, 1991–1995

Variables

Coeﬀ.

Dependent variable: PAT (number of patents) Intercept 9.889254*** DIST (Industrial district) 0.522237*** SIZE (Number of employees) 0.000217 PPE (Property, plant and equipment) 1.90E-05** GOS ( Gross operating surplus) 0.015699 MULTI (Multi-plant) 2.753913*** GROUP (Industrial group) 0.135377 EXP (Export) 0.308499 SSUP (Specialized suppliers) 9.315202*** SDOM (Supplier dominated) 9.019210*** SINT (Scale intensive) 9.328195*** Number of observations F-test R2 adjusted (overall)

Std. Err.

Prob.

2.234583 0.127098 0.000461 6.79E-06 0.027120 0.194963 0.163996 0.161900 2.233547 2.234540 2.237762

0.0000 0.0001 0.6381 0.0055 0.5632 0.0000 0.4099 0.0578 0.0000 0.0001 0.0000 265 45.432 0.627

Note: ***, **, * statistically signiﬁcant at 99%, 95% and 90% conﬁdence levels, respectively.

belonging to a group. Conversely, exporting ﬁrms and ﬁrms belonging to scale-intensive industries exhibited a disadvantage in terms of patents, whereas the coeﬃcient for the size variable was not signiﬁcantly diﬀerent from zero. The picture changes signiﬁcantly when one focuses upon the 53 ﬁrms for which data were available in relation to the 1991–95 period (Table 10.6). The value of property, plant and equipment, the fact of possessing more than one plant, and being export oriented are all factors which enhance the innovative (patented) output of the ﬁrm. Conversely, the district dummy has a negative and statistically signiﬁcant coeﬃcient, whereas the coeﬃcients of both the group and the size variables are not signiﬁcant. Evidently, innovative activity for ﬁrms in the sample is market driven, and location within an industrial district is no longer a factor that, other things being equal, is able to enhance signiﬁcantly the innovativeness of the ﬁrm. The goodness of ﬁt of this regression is still satisfactory, with R2 adjusted 0.627. A possible interpretation of the diﬀering importance of industrial districts as drivers of innovation and technological change between the second half of the 1980s and the ﬁrst half of the 1990s is that only during the ﬁrst subperiod did ﬁrms located in industrial districts beneﬁt as regards their innovative activities from the agglomeration economies that characterize local clusters.

264

Agglomeration in Italy

This advantage was probably lost when, as a consequence of globalization, these ﬁrms became more involved in a relocation process which pushed down total production costs and made innovative capability a secondary element in the strength of ﬁrms specialized in the productions typical of Emilia-Romagna districts. In this connection, the resulting ‘multi-located’ district appears to be a new form of industrial agglomeration, one in which spatial concentration is no longer a factor shaping the competitive advantage of industrial districts.

5.

CONCLUDING REMARKS

After analysing some speciﬁc aspects of the evolution of industrial agglomerations in Emilia-Romagna, it is possible to address a crucial question concerning the future of the industrial district vis-à-vis the diﬀusion of ICTs: will small specialized ﬁrms continue to cluster together to form spatially concentrated industrial districts even when and if ICTs become a tool for the interchange of information and the communication of needs, which is as eﬃcient as the face-to-face dealings and exchange of information that typify the traditional industrial districts, or is a new form of industrial agglomeration, the ‘multi-located’ district (as deﬁned by Santarelli 1988), emerging? A possible answer to this question resides in the long-term evolution of industrial districts, in particular those in Emilia-Romagna. This evolution has been punctuated by increasing enlargements of the type and number of activities carried out by ﬁrms belonging to this industrial agglomeration. What has remained unchanged is the circulation of information and the close relationships among ﬁrms that led to increasing inter- and infra-sectoral integration (Garofoli 1987). In this connection, the emergence of specialized suppliers of capital equipment in the 1960s, the advent of leading ﬁrms belonging to industrial groups during the late 1980s, and, likely, the eﬃcient relocation of the most labor-intensive phases of the overall production process consequent upon the availability of more reliable devices for exchange of information since the 1990s, are the three crucial events in the history of industrial districts. The resulting multi-located district of the last few years is therefore nothing but a new form of industrial agglomeration in the age of globalization: whereas spatial concentration is no longer the most crucial factor enabling the prosperity of the modern district, its distinctive organizational features and the ﬂows of information that it is able to set in motion are substantially unchanged. What Harrison (1994) saw as a point of strength of MNCs – namely, the ability to relocate manufacturing throughout the world to exploit diminishing tariﬀ and transportation

The competitive advantage of a region: Emilia-Romagna

265

costs besides escaping increasing competition by low-wage countries – is now also a point of strength of industrial districts. The (spatially concentrated) industrial district is dead: long live the (multi-located) industrial district!

NOTES 1. One in which the success lies not just in the realm of ‘economics’, since broader social and institutional aspects are just as important. 2. Specialization indices, share of employment accounted for by SMEs, share of employment in the relevant industry as compared to the regional and regional averages and so on. 3. Whereas the national average is 5.9 per cent. 4. Whereas the national average is 6.2 per cent. 5. A pharmacist by training, as early as 1962 Veronesi began to prepare the ground for what was eventually to become the Mirandola district. In the course of his work with local hospitals, Veronesi saw a nascent market for plastic disposables. During the years that followed, he founded numerous companies (including Miraset, Sterilplast, Dasco, Bellco and Dideco) to supply the medical market with components for infusion, hemodialysis, oxygenation and related applications (see Lichtman 2002). 6. Of the 32 largest producers of ceramic tiles in Italy, 30 are located in Emilia-Romagna, either in the Sassuolo district or in the districts of Faenza and Imola. 7. The largest producer of ceramic tiles in the Sassuolo district is Iris Ceramica. Set up in 1961, this company attained market leadership thanks to direct control of production cycles and experience in advanced manufacturing technologies. 8. As a consequence of the lower propensity of smaller ﬁrms to undertake those R&D activities which have been shown to result in patented innovations. 9. Although the coeﬃcient of the SIZE variable is statistically signiﬁcant only at the 90 per cent conﬁdence level. In any case, this result is consistent with the empirical regularity that emerged from a number of studies concerning the independence of the ﬁrm’s innovative intensity on the ﬁrm’s size (see Klette and Kortum 2002).

REFERENCES Audretsch, D.B. and M.P. Feldman (1996), ‘R&D spillovers and the geography of innovation and production’, American Economic Review, 86 (4), 630–40. Audretsch, D.B., E. Santarelli and M. Vivarelli (1999), ‘Does startup size inﬂuence the likelihood of survival?’, in D.B. Audretsch and A.R. Thurik (eds), Innovation, Industry Evolution, and Employment, Cambridge: Cambridge University Press, pp. 280–96. Barba Navaretti, G., E. Santarelli and M. Vivarelli (2002), ‘The role of subsidies in promoting Italian joint ventures in least developed and transition economies’, Applied Economics, 34 (12), 1562–9. Becattini, G. (1979), Scienza economica e trasformazioni sociali, Florence: la Nuova Italia. Becattini, G. (1990), ‘The Marshallian industrial district as a socio-economic notion’, in F. Pyke, G. Becattini and D. Sengenberger, Industrial Districts and Inter-ﬁrm Cooperation in Italy, Geneva: ILO, pp. 35–51.

266

Agglomeration in Italy

Becchetti, L., A. de Panizza and F. Oropallo (2002), ‘Distretti industriali: identità e performance’, paper presented at the 43rd Annual Meeting of Società Italiana degli Economisti, Ferrara, 25–26 October. Biggiero, L. (2002), ‘The location of multinationals in industrial districts: knowledge transfer in biomedicals’, Journal of Technology Transfer, 27 (1), 111–22. Boari, C. and A. Lipparini (1999), ‘Networks within industrial districts: organising knowledge creation and transfer by means of moderate hierarchies’, Journal of Management and Governance, 3 (3), 339–60. Brusco, S. (1982), ‘The Emilian model: productive decentralization and social integration’, Cambridge Journal of Economics, 6 (1), 167–84. Brusco, S. (1986), ‘Small ﬁrms and industrial districts: the experience of Italy’, Economia Internazionale, 39 (2–3–4), 85–97. Censis (2001), Per una nuova mappa dello sviluppo locale – X forum delle economie locali, Rome, Censis collana editoriale note & commenti. Club dei Distretti (1999), http://www.clubdistretti.it/Distretti/mappe/ipi/ﬁg 44_ pag_178.pdf. CNEL/Ceris-CNR (1997), Innovazione, piccole imprese e distretti industriali, Rome: CNEL. Fondazione, G. Brodolini (1996), Il libro della piccola impresa, Rome: Adnkronos Libri. Forni, M. and S. Paba (2002), ‘Spillovers and the growth of local industries’, Journal of Industrial Economics, 50 (2), 151–71. Garofoli, G. (1987), ‘Il modello territoriale di sviluppo degli anni ’70 e ’80’, Note economiche, 17 (1), 156–76. Giovannetti, E. (2000), ‘Technology adoption and the emergence of regional asymmetries’, Journal of Industrial Economics, 48 (1), 71–102. Grabher, G. (ed.) (1993), The Embedded Firm: On the Socioeconomics of Industrial Networks, London: Routledge. Harrison, B. (1994), Lean and Mean: The Changing Landscape of Corporate Power in the Age of Flexibility, New York: Basic Books. Henderson, V., A. Kuncoro and M. Turner (1995), ‘Industrial development in cities’, Journal of Political Economy, 103 (7), 1067–90. ISTAT (1997), I sistemi locali del lavoro in Italia, Rome: Istituto Poligraﬁco e Zecca dello Stato. Kim, S. (1995), ‘Expansion of markets and the geographic distribution of economic activities: the trends in US regional manufacturing structure, 1860–1987’, Quarterly Journal of Economics, 65 (4), 881–908. Klette, T.J. and S. Kortum (2002), ‘Innovating ﬁrms and aggregate innovation’, NBER Working Paper No. 8819, Cambridge, MA, March. Lazerson, M. and G. Lorenzoni (1999), ‘Resisting organizational inertia: the evolution of industrial districts’, Industrial and Corporate Change, 3 (3), 361–77. Lichtman, B. (2002), ‘Regional focus: Northern Italy’, EMDM (European Medical Device Manufacturer), Medical Devicelink, www.devicelink.com/emdm/archive/ 02/09/004.html. Lipparini, A. and A. Lomi (1999), ‘Interorganizational relations in the Modena biomedical industry: a case study in local economic development’, in A. Grandori (ed.), Interﬁrm Networks. Organization and Industrial Competitiveness, London: Routledge, pp. 120–50. Lorenzoni, G. and C. Baden-Fuller (1995), ‘Creating a strategic center to manage a web of partners’, California Management Review, 37 (2), 146–63.

The competitive advantage of a region: Emilia-Romagna

267

Lorenzoni, G. and O. Ornati (1988), ‘Constellation of ﬁrms and new ventures’, Journal of Business Venturing, 3 (1), 41–57. Marshall, A. (1890), Principles of Economics, London: Macmillan. Meyer-Stamer, J., C. Maggi and S. Seibel (2001), ‘Improving upon nature: creating competitive advantage in ceramic tile clusters in Italy, Spain and Brazil’, INEF Report No. 54, Duisburg: Institut für entwicklung und Frieden der GerhardMercator-Universität Duisburg. Moussanet, M. and L. Paolazzi (eds) (1992), Gioielli, bambole, coltelli. Viaggio de Il Sole 24 Ore nei distretti poduttivi italiani, Milan: Il Sole 24 Ore. Pavitt, K. (1984), ‘Sectoral patterns of technical change: towards a taxonomy and a theory’, Research Policy, 13 (4), 343–73. Piergiovanni, R. and E. Santarelli (1996), ‘Analyzing literature-based innovation output indicators: the Italian experience’, Research Policy, 25 (5), 689–711. Piore, M.J. and Ch. Sabel (1984), The Second Industrial Divide, New York: Basic Books. Prometeia (2002), Analisi dei microsettori – Piastrelle, Bologna: Prometeia. R&I (Ricerche and Interventidi Politica Industriale) (2001), ‘Il distretto biomedicale modenese’, Modena, mimeo. Rosenberg, N. (1976), Perspectives on Technology, Cambridge: Cambridge University Press. Russo, M. (1985), ‘Technical change and the industrial district: the role of interﬁrm relations in the growth and transformation of ceramic tile production in Italy’, Research Policy, 14 (3), 329–43. Santarelli, E. (1988), ‘Distretti multi-localizzati e innovazione nell’industria italiana’, Economia Marche, 7 (2), 161–208. Santarelli, E. and A. Sterlacchini (1994), ‘Embodied technological change in supplier dominated ﬁrms. The case of Italian traditional industries’, Empirica, 20 (3), 313–27. Sforzi, F. (1985), ‘Riﬂessioni sul distretto industriale: un’ipotesi di identiﬁcazione spaziale’, in R. Innocenti (ed.), Piccola città & piccola impresa. Urbanizzazione, industrializzazione e intervento pubblico nelle aree periferiche, Milan: Franco Angeli, pp. 247–67. Williamson, O. (1985), The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting, New York: Free Press.

11. Where is the Internet? Agglomeration in space and cyberspace Emanuele Giovannetti, Karsten Neuhoﬀ and Giancarlo Spagnolo There’s no there there. (Holtzman 1994, p. 197)

1.

INTRODUCTION: AGGLOMERATION

Spatial asymmetries, geographical agglomeration, industrial districts and their morphological changes in time are commanding a growing empirical and theoretical interest among economists, to the point that Krugman (1991, p. 5) identiﬁes concentration as ‘the most striking feature of the geography of economic activity’. Fujita and Thisse (2002) describe how the competitive equilibrium paradigm is incompatible with agglomeration, or indeed with any relevant notion of space since its introduction implies zero transport costs in equilibrium; consequently spatial, imperfect competition is required to study agglomeration. Their analysis is based on two forces: ‘localized positive externalities’ (centripetal) and ‘transport costs’ (centrifugal). The interplay between these two forces produces the economic landscapes, where agglomeration of cities and industrial districts are determined. An industrial organization approach to explain these features lies in the ‘principle of minimum diﬀerentiation’ introduced by Hotelling (1929) in his model of product diﬀerentiation through location decisions. In this setting, Hotelling argued that with linear transportation costs ﬁrms tend to locate at the center of the city and supply identical products. D’Aspremont et al. (1979) proved that such an equilibrium requires that ﬁrms are located suﬃciently far apart. By considering quadratic transport costs they reestablished existence and uniqueness of a price equilibrium and reversed Hotelling’s conclusion by showing that ﬁrms tend to locate as far apart as 268

Agglomeration in space and cyberspace

269

possible from each other. This property has been called, the ‘principle of maximum diﬀerentiation’.1 Initial analysis of price competition with product diﬀerentiation considered a one-dimensional characteristic space, traditionally interpreted as geographical distance. Irmen and Thisse (1998) extended previous work on product diﬀerentiation by considering ﬁrms’ location in a multidimensional characteristic product space. Assuming quadratic transport costs, they found that if one of the commodity characteristics is more relevant than others in the consumer’s utility functions, then, in equilibrium, there is maximum diﬀerentiation along this characteristic and agglomeration along all other dimensions. In Fujita and Thisse (2002), urban and economic history are re-read along similar lines: before the Industrial Revolution there were small ﬁxed costs of production, and high transport costs, hence production was fragmented into many small units. After the Industrial Revolution, transport costs have decreased dramatically, and assuming that proximity lowers production costs, this would provide an incentive to agglomerate at a few locations. However, these beneﬁts were balanced by the increased market power due to isolation from the nearest competitors. In this trade-oﬀ, lower transport costs reinforce the incentives toward agglomeration by reducing the beneﬁts of spatial market power, and display a monotone decreasing relationship with the degree of agglomeration. In particular, Krugman (1991) explains this dynamics of geographical agglomeration as the interplay of increasing returns, labor migration and transport costs. Economic integration resulting in lower transport costs leads counterintuitively to geographical agglomeration. However, by linking the changing patterns of the core–periphery relation to diﬀerent stages of a process of gradually declining transportation and communication costs, Krugman and Venables (1995, p. 859) ﬁnd an emerging ‘U-shaped pattern of global economic change, of divergence followed by convergence’. Puga (1999) and Giovannetti (2000) also obtained a non-monotonic relation between transport costs and agglomeration. Since Marshall (1890), agglomeration has been studied as the result of three forces: a pooled labor market, greater provision of non-traded inputs, and knowledge spillovers. Glaeser et al. (1992) stressed the importance of geographic proximity in deﬁning the extent of knowledge spillovers within ﬁrms of a given industry to explain the agglomeration in cities.2 The micro foundations for agglomeration often lie in the proximity among economic agents. Proximity matters since a basic input for ﬁrm’s activities, tacit knowledge, is assumed to be transferable only through face-to-face interaction: ‘the transfer of information through modern transmission devices requires its organization according to some pre-speciﬁed patterns, and only

270

Agglomeration in Italy

formal information can be codiﬁed in this way’ (Fujita and Thisse 2002, p. 172). Is this necessarily true? A wink on a web-cam is transmitted through a codiﬁed pattern not necessarily understood by the users. The crucial issue is whether, and to which degree, knowledge is transferable over distances. If tacit knowledge is non-codiﬁable, and not transmittable, then space matters for increasing returns. This might not be universally true, since new communication technologies are transforming, and often breaking, the link between tacit and non-codiﬁable knowledge. Every image and sound, not only written text, can be transmitted and reacted upon in real time: for example, watching the Gioconda’s smile on the Internet3 does transmit, through a codiﬁed sequence of binary numbers, tacit ideas in a space, the Internet, where geographical proximity matters less than connectivity or language aﬃnity. The relevance of proximity, in geographical space, should therefore be considered a parameter reﬂecting the degree of transmissibility of the knowledge relevant to the speciﬁc activity under study. This will diﬀer across industries and will reﬂect the evolution of the diﬀerent technologies and their human–machine interface.

2. PROXIMITY IN CYBERSPACE: DEATH OF THE DISTANCE? In some sectors, in particular for digital goods, new technologies reduce transport costs to almost zero, hence diminishing the centrifugal forces. The interaction of this eﬀect with the previously discussed reduction of local externalities suggests that the impact of new communication technologies will have diﬀerent intensities and signs depending on the speciﬁc sector analysed; and that there will be plenty of non-monotonic eﬀects between a technology-driven reduction in transport costs and agglomeration. The relation between the Internet and the notion of distance as expressed, traditionally, by transport costs shows its most extreme consequences in the ‘death of distance’ attitude, claiming that the instantaneous communication made possible by the Internet leads to a collapse in space–time boundaries. This position is easily represented by the statement that advanced telecommunications are not reducing distance but they make it entirely meaningless, since the time taken to communicate over 10 000 miles or over 1 mile is virtually indistinguishable. In this framework, time – being dissociated from distance – becomes the only relevant economically scarce dimension. Mitchell4 (1998) deﬁnes this as the economy of presence rather than proximity, and analyses the profound implications on the comparative advantages of diﬀerent locations which, more than on proximity, will be based on

Agglomeration in space and cyberspace

271

the idea of connectivity. Kitchin (1998) argues that geography instead of becoming irrelevant is transformed, in particular since there is unequal spatial distribution of the bandwidth and the underlying physical network topologies, so that the analysis of the geographic distribution of the information infrastructure becomes essential to understand the agglomeration forces. However, by 2003, the bandwidth between and within developed countries no longer allowed for any meaningful diﬀerentiation, the only diﬀerentiation being the bandwidth and access charges of private consumers and small industry to the ﬁnal loop. Critical viewers of the ‘space-less’ economy often stress the unique relevance of face-to-face relations, direct human interaction, which independently of the non-direct human communication speed and cost characterizes the idea of place as diﬀerent from space. Following this view, a place has ‘insiderness’, that is, it provides an identiﬁcation for the individuals belonging to it (Relph 1976, Place and Placeness from Dodge and Kitchin (2000)). The relevance of a notion of insiderness, reintroducing a distance between places which might have ‘distance-less’ communication costs, does not necessarily link this to a speciﬁc geographical location: suppose there is insiderness in an online community; its borders will not necessarily be drawn in the geographic space. If there are places in cyberspace, these are formed by a common interest aﬃnity and language more than geographical proximity. It has been argued that moral commitment and social cohesion are primary identiﬁers of places in cyberspace and that places emerge as a consequence of the relationships that the subjects participating in them establish among themselves. The considerations described above imply that the trade-oﬀ between the centripetal and centrifugal forces deﬁning agglomeration and districts’ boundaries is not only aﬀected by the impact that information and communication technology (ICT) has on the cost-relevance of distance, but it also depends on whether ICT facilitates the emergence of places relevant for establishing cyber-local externalities among peers. If this is the case, ICT will have two opposite eﬀects on the agglomeration: (a) weakening the centrifugal forces, because of the lessening of the isolation market power due to lower transport costs; and (b) redesigning the baricentre of the centripetal forces, attracting agglomeration around a ‘virtual location’. To better understand the possibility of having virtual places becoming attractors of a district we need to further analyse the possibility that local externalities, used traditionally to explain geographical agglomeration, can work around a virtual place. Usually the nature and range of local externalities is closely linked to the language and protocol used. Online communities5 generated their own language protocols and expressions that play the same delimiting borders of traditional languages for reciprocal understanding.

272

Agglomeration in Italy

Fluency in this format is, however, only very loosely connected to geographical location. The nature of local externalities and communications protocols is of course related to the speciﬁc nature of the interaction sought and particularly related to the ability to establish, maintain and verify ‘brand reputation’ and ‘trust’. This possibility is itself linked to the evolution and diﬀusion of both: (a) public key infrastructure encryption technologies,6 such as digital signature and certiﬁcate and (b) the power of enforcement of legally binding cyber agreements.7 However, although they are necessary conditions, encryption technologies and a clear legal enforcement framework do not provide a suﬃcient condition for the establishment of trust in cyber-mediated relations because, at least, of contract incompleteness. The inability of drafting complete contracts is often dealt with through the emergence of conventions both in geographical and cyber-space interactions. A relevant research question to understand the possible emergence of virtual districts is: ‘do new technologies provide the means for the emergence of conventions necessary to facilitate trust in cyber mediated exchanges?’. Online places have been historically characterized by behavioral codes also deﬁned ‘netiquette’, the breaking of which often has disruptive consequences on the deviant’s reputation within the community. If there is competition between geographical and virtual districts, their relative competitive advantage will depend on whether the monitoring of these codes is easier through geographical proximity or via online interaction, and if the ensuing necessity of a credible retaliation of a deviant’s behavior is more easily implemented in an online connected community or in a geographically clustered one. These elements taken together should drive the agglomeration/polarization dynamics in the speciﬁc industry under study, deﬁning the shape and borders, if existing, of the geographical or cybergeographical distribution of the industry.

3. ‘RELATIONAL GOVERNANCE’ AS A CENTRIPETAL FORCE FOR DISTRICTS We wrote above that an important factor determining whether virtual districts will emerge independent of geographic proximity is the possibility of establishing, maintaining and verifying brand reputation and trust in cyberspace. In this section we shall argue that the impressive recent developments in ICT notwithstanding, there are reasons to think that geographical proximity will still oﬀer comparative advantages in terms of facilitating reputational forces.

Agglomeration in space and cyberspace

273

Analyses of agglomeration in the New Economic Geography typically abstract from the microeconomic structure of the industry, to focus on the interaction of centripetal and centrifugal forces in determining (given) ﬁrms’ location choice. However, we are interested here in industrial districts, and a crucial feature is their micro structure: small ﬁrms; vertically disintegrated production (vertical specialization); ﬂexible system of vertical and horizontal subcontracting. These features had already been noticed by Marshall (1890), and have been stressed more recently by Brusco (1982) and Becattini (1990). What hasn’t perhaps been suﬃciently stressed, is that a vertically fragmented industry relying on frequent vertical and horizontal subcontracting implies a large number of transactions which are likely to require a good governance system. Williamson (1996) distinguishes three main forms of governance for inter-ﬁrm (vertical and horizontal) transactions: explicit contracts, assets posted as hostages, and relational (or implicit) contracts. Assets posted as hostages are seldom observed and ‘complete’ explicit contracting is seldom a feasible option for ﬁrms in a district because court enforcement involves ﬁxed costs and is not economic for small transactions between small ﬁrms. Therefore transactions within districts must be governed mainly by longterm relations, where reputational forces can ensure compliance with exchange agreements (preventing ‘hold-up’; see Blonski and Spagnolo 2002). Both Brusco and Beccattini stress that interaction within a district involves forms of interﬁrm cooperation. Recent observers focused on ‘cooperative information sharing’ among competing ﬁrms in R&Dintensive industrial districts. For example, a study of Silicon Valley by Annalee Saxenian (1994) discusses how informal, cooperative social relations play a crucial role in enforcing knowledge exchanges within the computer-industry district in Silicon Valley. For the biotechnology industry, Powell (1996) and Powell et al. (1996) argue that formal arrangements merely represent the tip of the iceberg in the set of informal relations, and point out that the ‘development of cooperative routines goes beyond simply learning how to maintain a large number of ties’. But information sharing apart, all transactions between small ﬁrms in a district (any vertical and horizontal subcontracting) require a form of cooperation, of ‘relational governance’; suﬃciently strong reputational forces are needed to enforce exchange agreements where court enforcement is infeasible. We have argued that eﬀective reputational forces ensuring compliance with inter-ﬁrm exchange agreements appear crucial for the survival of any type of district. We also know that the eﬀectiveness of reputational forces may be enhanced by having transactions embedded in a community, in a social network. This may facilitate the governance of inter-ﬁrm transactions and enhance the competitive strength of the district.

274

Agglomeration in Italy

Furthermore, a community network may facilitate information transmission, and it is well known that better information transmission increases the eﬀectiveness of reputational forces within communities (Kandori 1992). However, as argued in previous sections, the impressive recent advancements in ICTs suggest that communication in cyberspace may soon become a good, if not a better substitute for community-based face-to-face information transmission mechanisms. But a district embedded in a community network, with its tissue of social relations, may also rely on social sanctions – besides the usual economic ones – as additional threats to discipline and govern inter-ﬁrm transactions (Spagnolo 1999, 2000; Annen 2001). Retaliation against deviants can take place in both social and business life, increasing incentives to behave. This form of ‘social capital governance’ is common in many districts around the world, and may constitute a substantial comparative advantage for agglomerated, embedded districts (although it may also be misused). This last consideration suggests at a new potential, diﬀerent from those previously identiﬁed in the literature: agglomerated districts should have better governance – that is, lower transaction costs – than more dispersed ones: agglomeration is a precondition for embeddedness since communities of people are geographically concentrated. Recent experiments on ‘cheap talk’ in coordination and repeated games provide an additional ground to believe that the need for eﬀective relational governance may constitute a signiﬁcant, as yet disregarded centripetal force for districts. Pre-play communication in game-theoretic experiments, ‘cheap-talk’, can be seen as a special form of face-to-face interaction/communication. However, the value of this kind of face-to-face communication does not lie in a more direct and precise transmission of information – as usually argued in the agglomeration literature – since experiments are expressly designed so that cheap talk conveys no information whatsoever (see Crawford 1998). Rather, this form of face-to-face interaction has been shown to have value in facilitating agents’ coordination and the creation and maintenance of trust, an essential ingredient of the long-term cooperative relations necessary to govern at low cost a complex, ﬂexible system of subcontracting. To conclude, this section identiﬁed a novel potential centripetal force based on the governance needs of fragmented industrial districts. Reputational forces are important for the cost-eﬀective governance of small transactions and ﬂexible informal subcontracting. Agglomeration may or may not help in terms of better information ﬂows since, as argued in previous sections, ICTs may soon eﬀectively substitute for face-to-face and community-managed information transmission. But geographical closeness may help the reputational governance of districts’ subcontracting

Agglomeration in space and cyberspace

275

systems by allowing for community embeddedness – hence for social sanctioning power to enforce exchanges – and for better trust building thanks to personal, face-to-face interaction. Note that this force will be important where explicit contracts and the court system are relatively expensive, but also when crucial aspects of inter-ﬁrm transactions are not easily monitorable and veriﬁable, in which case formal contracting is impossible even where the court system is eﬃcient. Robust experimental evidence indicates that the presence of important non-contractible aspects in a transaction immediately leads to abandoning spot markets and establishing long-term cooperative relations (Fehr et al. 2001). As discussed below, in the ‘peering’ decision between Internet Service Providers (ISPs) there are substantial aspects of the transaction that are impossible to measure or monitor, so that the peering decision may require substantial trust and informal cooperation between peering partners. This may activate the centripetal force discussed above: face-to-face meetings and social connections may facilitate the governance of peering agreements, and the former require geographical proximity.

4.

THE INTERNET

The Internet is composed of many independent networks of very diﬀerent size, located around the globe, all directly or indirectly interconnected with one another. This last feature guarantees the Internet’s most important property: universal exchange of traﬃc between all end users (universal connectivity). The industry is still rather unregulated, and networks are left completely free to decide where, how and with whom to interconnect. Lacking a really dominant network, competitive forces and positive network externalities have been suﬃcient until now to keep all the networks interconnected8 (‘potential’ regulatory intervention may also have played a role). ISPs are often rather small networks that sell Internet interconnection and related services to end users, businesses and consumers. They rely on connections to larger networks for the delivery of their customers’ packets to their destinations outside the range of the ISP’s own subscribers. The largest networks are called ‘backbones’. These own or lease national or international high-speed ﬁber optic networks and deliver packets around the world for the many smaller networks connected to them. Backbones are not fully specialized in connecting other networks; in most cases they also reach businesses and consumers directly by operating their own vertically integrated ISPs.

276

Agglomeration in Italy

Interconnection Agreements Clearly, a smooth exchange of traﬃc between networks is crucial to the functioning of the Internet. Two simple types of interconnection agreements have emerged to regulate traﬃc at exchange points between networks: transit agreements and peering agreements. In a transit agreement, a large network – the transit provider – oﬀers access to the entire Internet to a smaller customer network against the payment of a fee often related to the capacity of the connection link. In other words, in a transit arrangement one network pays another one for interconnection and becomes a ‘wholesale customer’ of the other network, able to access all end users this other network can access through its other interconnection agreements. This transaction typically involves a single connection point between the two networks. Under a peering agreement, two networks exchange the traﬃc directed to each other’s end users only. Monetary settlements between peering partners used to be excluded, although recently some networks have started charging for peering (Miller 2002). Peering can be seen as a reciprocal, non-monetary exchange relationship that often implies various forms of cooperation. Peering also implies establishing direct exchange points between the two networks, and the costs of creating and maintaining the exchange points are typically shared evenly. Peering agreements may also be multilateral, and traﬃc exchanges may take place at private peering points or at organized exchange points such as network access points (NAPs) and Internet exchange points (IXPs), specialized facilities where ISPs can connect to each other to exchange Internet traﬃc. NAPs and IXPs are typically ‘public’ Internet exchange points where a switching system is provided to enable any member to exchange traﬃc with several other members. To peer at a NAP/IXP, an ISP usually has to establish a connection and pay a membership fee, after which it can use the circuit to carry the aggregate traﬃc to all of the other members of the NAP/IXP with whom the ISP agrees to peer. This makes peering at a NAP/IXP of an ISP cheaper than establishing a direct bilateral peering exchange point which would require installing a direct connection and many commercial deals. Being a member of a NAP oﬀer further advantages such as sharing information and access to a free mutual technical help forum. Technological Aspects Three important technological features of the Internet frame the subsequent discussion. First, as yet there is no technology available to measure

Agglomeration in space and cyberspace

277

with suﬃcient precision the load placed on a network by another connected network, only the amount of traﬃc exchanged between two networks can be measured not the paths followed by each packet within the network. The second feature is ‘hot potato routing’. Carrying traﬃc within a network is costly. When two networks are connected at more exchange points, as is usually the case between large networks, each network routes all traﬃc as soon as possible into the other network, independent of whether this increases the overall length of the transmission path. This individually optimized routing principle is ineﬃcient since the traﬃc does not necessarily follow the shorter or uncongested path. However, lacking technology to implement more eﬃcient solutions, hot potato routing has become the conventional, accepted practice, implying that the costs of transmitting a message are borne primarily by the destination network. The third feature is that the speed of the connection between two end users is not so much inﬂuenced by the physical distance between them, by the length of optic ﬁber to be covered. Rather, the speed of a connection is crucially determined by the degree of congestion in the various networks the traﬃc crosses. The most congested network on the path determines most of the delay.

5.

THE PEERING DECISION

Earlier work has identiﬁed several factors and problems that may aﬀect networks’ decision whether and with whom to peer.9 A ﬁrst, rather obvious factor is size. Peering and Size Peering requires establishing bilateral traﬃc exchange points, or peering points, which entail ﬁxed and variable technological costs. It follows that a suﬃciently intense traﬃc ﬂow between the end users of the two networks is a necessary precondition for peering to be economically viable. The larger the two networks are, the more intense will be the traﬃc between their end users, therefore networks’ size is a determinant of the peering decision. In fact, almost all large backbone networks peer with one another, the traﬃc being exchanged at several interconnection points homogeneously distributed on the relevant geographical areas. Somewhat smaller networks also peer with networks of comparable size, but typically have to supplement their interconnection with transit agreements with backbone networks. The

278

Agglomeration in Italy

smallest ISPs only buy transit from a backbone, as the ﬁxed costs of additional exchange points would not be covered by gains from peering with another smaller network (the traﬃc between two small networks is likely to be very small). Peering and Symmetry Since the costs of setting up and maintaining peering points are usually shared equally by peering networks, unbalanced traﬃc implies an unbalanced distribution of gains from peering against a balanced distribution of costs, a rather unfair settlement. Such unbalanced situations have developed in some cases, and have led to the discontinuation of the peering arrangement, and to its replacement by a transit one. Recently, large networks started publishing guidelines for peering partners. WorldCom, in particular, published four criteria to agree to peering with a network, two of which being that the peering network has a geographic scope of at least 50 per cent of its own; and that traﬃc ﬂow at peering points does not exceed 1:1.5. Asymmetric ﬂows and geographical extension, together with the hot potato routing would place on the larger network a relatively larger burden in terms of traﬃc carried and relatively smaller gains. The lack of monitoring technology makes monetary compensation for these asymmetries hard to agree upon. Free Riding The other two conditions required by WorldCom from peering partners are that the volume of traﬃc exchanged at peering points is suﬃciently large (>150 Mbps), a size criteria whose rationale was discussed above; and that most of the peering network has a capacity of 622 Mbps. This last requirement appears to be an attempt to curb a well-known potential free-riding on infrastructure investments. As discussed above, a technological peculiarity of the Internet is that the quality of a connection between two end users depends crucially on the most congested network on the connection path. When two networks are peering and one of them is congested, the perceived speed of connection would not improve even if the noncongested network were to upgrade its infrastructure. And if the congested network chooses not to upgrade its infrastructure, it enjoys the full cost savings while it shares the reduced performance with all the networks it is peering with. This problem may of course induce caution in networks’ decision whether and with whom to peer. Avoiding the race to the bottom that this free-rider problem may induce can require active cooperation on the side of peering networks.

Agglomeration in space and cyberspace

6.

279

AGGLOMERATION IN PEERING?

Little is known about the potential eﬀects of ISPs’ geographical location on their peering decision, the focus of our empirical analysis. Should we expect the geographical location of ISPs/networks to inﬂuence their decision whether to peer? Of course, if two ISPs are very far away, building a connection from scratch would cost a lot, hence one would expect that distant ISPs would not peer. However, consider a situation where there is a NAP where peering is cheaper, and that there are a number of ISPs, all of which are connected to this NAP. Should we then expect the geographical location of these ISPs to matter in their choice of peering partners? Should agglomeration patterns be observed in the peering decision? The centrifugal force usually considered in the New Economic Geography, softening competition through local diﬀerentiation, would not be active in this case, since the decision to peer at the NAP is independent from the location choice of the ISPs with respect to end users. Some centripetal forces considered in the literature, such as knowledge spillovers obtained through interactions with peers, may be moderately active; and transport costs would be represented by the mile-cost of interconnection and the cost of reaching peers for joint activities and face-toface interaction. However, if we consider a population of ISPs all of whom are already connected to a given NAP, as we do in the analysis below, milecost of interconnection does not matter. As discussed above, many features of a peering agreement are not directly monitorable, not to say veriﬁable/contractible. Hence peering agreements may require a great deal of trust and informal cooperation, in which case ‘face to face’ can be important and the new centripetal force discussed in Section 7 may be active. In the next section we perform an exploratory empirical analysis of the peering decision of Italian ISPs connected with MIX, the Milan Internet Exchange, to evaluate the strength of agglomerating forces on peering.

7.

THE EMPIRICAL ANALYSIS

Introduction Italy has three major public Internet exchange points: two located in the northwest: the MIX in Milan and the TOPIX in Piedmont; and one in central Italy: the NAMEX located in Rome. While the MIX (67 members) and the NAMEX (16 members) have academic origins, the TOPIX (15 members) was launched, in 2002, by a consortium of network

280

Agglomeration in Italy

operators. Our empirical investigation focuses on the MIX,10 whose statutory aim is to facilitate the cooperation and communication between ISPs in Italy. Access to the MIX is open to any operator upon acceptance of the conditions included in its Memorandum of Understanding11 and these reﬂect elements of non-proﬁt, cooperative, behavioral patterns whose relevance to the peering decision has been discussed above. In particular bilateral peering agreements established within MIX premises must be free of charge and each member must ensure that its usage of MIX resources is not detrimental to the usage by other members. On the other side the MIX s.r.l. will adhere to the commitments to publish and update, to the sole beneﬁt of the members: the peering matrix; the bandwidth for the reference period; the traﬃc statistics for the reference period; the quality of the oﬀered service for the reference period and to provide support to the members in the set-up procedures of their equipment. Each ISP joining the exchange will pay €7746.85 una tantum, representing the joining fee for 10 years; this cost will represent the penalty fee in case of leaving the MIX s.r.l. before the end of the 10 years; an annual membership fee based on the nominal value of the bandwidth (the bandwidth the ISP uses on the MIX LAN) that each ISP has declared. These fees do not include the costs for connecting the ISPs site to their equipment within the MIX and the conﬁguration of ISP equipment within the MIX.12 The Data The data are taken from the MIX peering matrix, which provides details of the actual choice of binary peering among its members and their nominal bandwidth at the MIX. We measured geographical distance based on the driving time between each two locations obtained from the Driving Directions of the Yahoo! Maps website. Finally, we searched for each member of the MIX to see whether it was participating at other European Internet exchange points, by looking at their peering matrices. The number of additional IXPs any ISP was a member of has been used as a proxy for its level of European connectivity. The Variables The analysis aims to better understand which ISP providers peer at an IXP. Figure 11.1 (lower part) gives the peering matrix of the Milan ISP as of November 2002. Two-thirds of all pairs have signed peering agreements, mirrored in the dominant dark color. We can only observe successful peering agreements, and therefore do not have information as to whether

Agglomeration in space and cyberspace

281

Distance from Milan Model Prediction Peering unlikely Peering possible Peering very likely

Observed No peering Peering

Figure 11.1

Peering matrix

such an agreement was signed with the full support of both organizations, was barely signed, or alternatively whether the peering agreement was just not signed or not expected to be. Such situations in econometrics are described by binary choice models – explanatory variables can only explain the likelihood of a peering agreement being signed rather than determine whether it is signed. Figure 11.1 (upper part) shows the result of the model estimation using the explanatory variables – the darker the color the more likely it is that the pair of ISPs will sign a peering agreement (in model prediction). In the estimation we try to ﬁnd a model such that the observed outcome is very likely, given the explanatory variables – that is, we maximize the likelihood. Results Table 11.1 gives the coeﬃcients13 which precede the explanatory variables with one standard deviation error margin. The similar size of all coeﬃcients indicates that all explanatory variables represent a signiﬁcant eﬀect. The negative coeﬃcient for travel time between the headquarters of the ISP companies, variable B, indicates, that the companies are less likely to

282

Agglomeration in Italy

Table 11.1 Value of coeﬃcients corresponding to explanatory variables with standard error Coeﬃcients

Range

A, B, D

Travel time between HQ (B) Max travel time to XP (C) Maximum bandwidth (D) Total no. of connections at other IXPs (A)

[0,1] [0,1] [0,1] [0,17]

1.20 0.13

A, C, D

Alone

1.590.13 1.280.14 1.630.13 0.89 0.12 0.920.12 0.610.11 0.15 0.01 0.150.01 0.170.10

Explanatory value of individual variables

Loglikelihood

–850 Total no. of connections at other XPs A

–900

Travel time between HQ B

–950

Explanatory value of joint variables A, C, D A, B, D A, D B, D C, D

Max travel time to XP C –1000 Maximum bandwidth D No explanatory variable 0

Figure 11.2

–1050

Relevant explanatory variables (Probit estimation)

peer, the larger the distance between the headquarters. On the other hand, companies are more likely to peer, the bigger the bandwidth of the larger company, variable D – and therefore the more traﬃc they have justifying the eﬀort required to sign a peering agreement. Finally, companies are less likely to peer the more other IXPs in Europe they are members of, variable A, because then they presumably can exchange traﬃc at other locations. In Figure 11.2 the log-likelihood of the observed peering behavior is depicted. We can see that explanatory variables increase the likelihood, represented by a lower negative log-likelihood value, of observing the peering behavior. Combining several explanatory variables we can increase the predictive power of our model. This conﬁrms that the peering decision is inﬂuenced by several factors (or at least that the factor determining the peering decision is correlated with several of our variables).

Agglomeration in space and cyberspace

283

The hit rate is a heuristic illustration of the total predictive power of the model. Given that 66 per cent of ISP pairs do peer, an uninformed person would predict for each pair to peer and therefore predict successfully in 66 per cent of the cases. The model using explanatory variables A, B, D improves on this prediction and succeeds in 76 per cent of the cases, which is interesting for social science predictions. To come back to our original question of where the Internet is and whether it agglomerates in space or cyberspace, our initial results suggest that even in one of the most technological advanced cluster of ﬁrms, that is, the ISPs building a relevant part of (not only) Italian cyberspace, spatial agglomeration still plays a positive role in establishing peering relations. The main explanation we can provide to support this result is that proximity still facilitates the monitoring and enforcement of cooperative peering relations, allowing socially embedded punishment strategies against potential free riders.

NOTES 1. 2.

3. 4. 5. 6. 7. 8.

9. 10. 11. 12. 13.

Osborne and Pitchik (1987) proved the existence of an equilibrium in mixed strategies where, unlike Hotelling’s result, location decisions are not concentrated in the same place. Localized knowledge spillovers within endogenous growth models are suﬃcient to generate diﬀerent growth rates between geographically separated locations. In this framework, asymmetry is not driven by agglomeration of production in space, but by quality diﬀerentiation across a ﬁxed geographical setting. Leonardo Da Vinci (1503–1505) La Gioconda, Oil on wood, 77 53 cm (30 20 7/8 in); Musée du Louvre, Paris, see the WebMuseum (2002), www.ibiblio.org/wm/paint/auth/ vinci/joconde/joconde.jpg. As reported in Dodge and Kitchin (2000). Like the text messaging ones. See Maeda (2004). Which is of course linked to evolution of the transnational legal agreements concerning the electronic communication sector. Were a really dominant network to emerge, network externalities may become a problem, and the dominant network could then have an interest in refusing interconnection to other networks in order to drive them out of the market. This would destroy universal connectivity, a potential event sometimes referred to as the ‘balkanization’ of the Internet, and would require regulatory intervention (Kende 2000). See Baake and Wichmann (1999); Huston (1999a and b); Kende (2000); and Filstrup (2001). The MIX s.r.l. (Ltd) was founded in January 2000. See the web page at www.mix-it.net/Documenti/regolamento-en.htm. See MIX s.r.l. MoU. We estimate the equations using a probit model. The error margin is always far smaller than the value of the coeﬃcient. Therefore a z-statistic shows that all coeﬃcients are signiﬁcant at the 0.01 per cent level, both if they are used alone as well as when they are used in combination with other coeﬃcients to explain the peering behavior. To facilitate the interpretation of the results all distances and bandwidths have been normalized such that their maximum value is one.

284

Agglomeration in Italy

REFERENCES Annen, Kurt (2001), ‘Inclusive and exclusive social capital in the small ﬁrms sector in developing countries’, Journal of Institutional and Theoretical Economics, 157 (2), 319–30. Baake, Pio and Thorsten Wichmann (1999), ‘On the economics of Internet peering’, Netnomics, 1, 89–105. Becattini, G. (1990), ‘The Marshallian industrial district as a socio-economic notion’, in F. Pyke, G. Becattini and D. Sengenberger (eds), Industrial Districts and Inter-ﬁrm Cooperation in Italy, Geneva: ILO, pp. 37–51. Blonski, Matthias and Giancarlo Spagnolo (2002), ‘Relational contracts and property rights’ (with Matthias Blonski), CEPR Discussion Paper No. 3460, London. Brusco, S. (1982), ‘The Emilian model: productive decentralization and social integration’, Cambridge Journal of Economics, 6 (1), 167–84. Crawford, Vincent (1998), ‘A survey of experiments on communication via cheap talk’, Journal of Economic Theory, 78 (2), 286–98. d’Aspremont, C., J.J. Gabszewicz and J.-F. Thisse (1979), ‘On Hotelling’s “stability in competition” ’, Econometrica, 47, 1145–50. Dodge, M. and R. Kitchin (2000), Mapping Cyberspace, London and New York: Routledge. Fehr, Ernst, Martin Brown and Armin Falk (2001), ‘Contractual incompleteness and the nature of market interactions’, Working Paper, Institute for Empirical Research in Economics, University of Zurich. Filstrup, B. (2001), ‘Internet interconnection agreements’, Final Project Report, S1646, Information Economics, http://conta.vom.gr/conta/phges/netstrategies/ papers/32.pdf. Fujita, M. and J.F. Thisse (2002), Economics of Agglomeration: Cities, Industrial Location, and Regional Growth, Cambridge: Cambridge University Press. Giovannetti, Emanuele (2000), ‘Technology adoption and the emergence of regional asymmetries’, Journal of Industrial Economics, 48 (1), 71–102. Glaeser, E., H. Kallal, J. Scheinkman and A. Schleifer (1992), ‘Growth of cities’, Journal of Political Economy, 100, 1126–52. Holtzman, S.R. (1994), Digital Mantras: The Languages of Abstracts and Virtual Worlds, Cambridge, MA: MIT Press. Hotelling, H. (1929), ‘Stability in competition’, Economic Journal, 39, 41–57. Huston, G. (1999a), ‘Interconnection, peering and settlements: part 1’, Internet Protocol Journal, 2 (1), 2–16. Huston, G. (1999b), ‘Interconnection, peering and settlements: part 2’, Internet Protocol Journal, 2 (2), 2–23. Irmen, A. and J.-F. Thisse (1998), ‘Competition in multi-characteristics spaces: hotelling was almost right’, Journal of Economic Theory, 78 (1), January, 76–102. Kandori, Michihiro (1992), ‘Social norms and community enforcement’, Review of Economic Studies, 59, 63–80. Kende, M. (2000) ‘The digital handshake: connecting the Internet backbones’, OPP Working Paper 32, Federal Communication Commission, Washington, DC. Kitchin, Robert M. (1998), ‘Towards geographies of cyberspace’, Progress in Human Geography, 22 (3), 385–406. Krugman, P. (1991), Geography and Trade, Cambridge, MA: MIT press. Krugman, Paul and Anthony J. Venables (1995), ‘Globalization and the inequality of nations’, Quarterly Journal of Economics, 110, 857–80.

Agglomeration in space and cyberspace

285

Maeda, A. (2004), ‘PKI solutions for trusted e-commerce: survey on the de facto standards competition in PKI industries’, in M. Kagami, M.Tsuji and E. Giovannetti (eds), Information Technology Policy and the Digital Divide, Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 260–81. Marshall, A. (1890), Principles of Economics, London: Macmillan. Miller, Rich (2002), ‘The economics of peering’, Web Host Industry Review, http://www.thewhir.com/features/peering.cfm. Osborne, M.J. and C. Pitchik (1987), ‘Equilibrium in Hotelling’s model of spatial competition’, Econometrica, 55, 911–22. Powell, Walter (1996), ‘Inter-organizational collaboration in the biotechnology industry’, Journal of Institutional and Theoretical Economics, 152 (1), 197–215. Powell, Walter W., Kenneth W. Koput and Laurel Smith-Doerr (1996), ‘Interorganizational collaboration and the locus of innovation: networks of learning in biotechnology’, Administrative Science Quarterly, 41, 116–45. Puga, Diego (1999), ‘The rise and fall of regional inequalities’, European Economic Review, 43, 303–34. Saxenian, Annalee (1994), Regional Advantage: Culture and Competition in Silicon Valley and Route 128, Cambridge, MA: Harvard University Press. Spagnolo, Giancarlo (1999), ‘Social relations and cooperation in organizations’, Journal of Economic Behavior and Organization, 38 (1), 1–26. Spagnolo, Giancarlo (2000), Norme Sociali, Incentivi, ed Eﬃcienza Organizzativa: una Analisi del Proﬁt-Sharing, Bologna: CLUEB. Williamson, Oliver E. (1996), The Mechanisms of Governance, Oxford and New York: Oxford University Press.

PART III

Agglomeration in the Americas

12. The software industry in North America: human capital, international migration and foreign trade Andrew Schrank* 1.

INTRODUCTION

The software industry has traditionally been highly localized. A vastly disproportionate share of the world’s software has been developed in North America, and a vastly disproportionate share of North American software has been developed in northern California’s ‘Silicon Valley’ and around ‘Route 128’ in Massachusetts (OTA 1990, p. 27; Correa 1996, p. 174; Carmel 1997, pp. 126–7; see Saxenian (1996) on Silicon Valley and Route 128 in general). But the North American monopoly on software development has recently been challenged by producers from India, Ireland and Israel, and the ‘three Is’ have therefore been portrayed as potentially replicable models of software-led development (see Dedrick et al. 1995; OECD 2000; Arora et al. 2001; Heeks and Nicholson 2002). The software industry has two alleged developmental advantages. On the one hand, it has relatively low start-up costs, for personal computers and programming software (that is, the material inputs to software production) are decidedly less expensive than inputs to traditional manufacturing industries including textiles, electronics, petrochemicals and automobiles. On the other hand, it is ‘developmentally nutritious’ (Helleiner 1990, p. 889), for it creates valuable goods and services and engenders a number of positive externalities (that is, human resource development, economy-wide productivity increases, and so on). In fact, Peter Evans describes information and communications technology (ICT) as ‘the sector most likely to spark a twenty-ﬁrst century conspiracy in favor of development’ (Evans 1995, p. 11). And the Organization for Economic Cooperation and Development’s Committee for Information, Computer, and Communications Policy holds that the software industry ‘is, arguably, the most important segment of ICT

289

290

Agglomeration in the Americas

economically, and is to the knowledge-based economy what the steel and automobile sectors were to the industrial economy’ (OECD 1998a; see also Schware 1992; Correa 1996; Eischen 2000). Nevertheless, the prospects of high proﬁts and correspondingly improved living standards have provoked intense – and potentially self-defeating – competition among the world’s ICT producers and production sites. While The Economist has identiﬁed more than 70 diﬀerent places labeled ‘Cyber this or silicon that’, it has derided the vast majority for their ‘silicon envy’ (The Economist, February 20, 1999, p. 25). What separates the pretenders from the real thing? And what are the implications for the development of the contemporary Third World? I hope to answer both questions by examining the software industry’s response to two distinct – and seemingly contradictory – geographical imperatives: centrifugal pressures engendered by the industry’s omnipresent need for abundant supplies of low-cost, highly skilled labor; and centripetal pressures engendered by the industry’s need for command, control and communications (C3). How is the ‘inherent tension between the primarily economic need to work at a distance and the needs of organizations and individuals to operate in conditions of close proximity’ overcome (Nicholson 2001, p. 13)? I hold that ‘expatriate knowledge networks’ (Meyer 2001) populated by emigrant scientists and engineers play a key role in mediating the relationship between North American IT ﬁrms and their overseas associates and thereby mitigate – albeit by no means eliminate – the manifold impediments to oﬀshore software development in their places of origin (see Eischen 2000, pp. 53–4). In other words, I hold that a transnational community of scientists and engineers holds the proverbial key to software export growth. The chapter is divided as follows. Section 2 discusses the centrifugal pressures emanating from the software industry’s need for highly skilled programmers. Section 3 examines the centripetal pressures emanating from software production’s need for C3. In Section 4, I use multivariate techniques to account for cross-national variation in software exports, ﬁnd that emigrant scientists and engineers are the most reliable predictor of exports to the US, and thereby suggest that the North American software industry has resolved the tension between cost and communications by relying upon a growing class of transnational professionals and entrepreneurs. And, ﬁnally, in Section 5 I discuss my ﬁndings and their implications for Silicon Valley, Route 128 and their imitators.

The software industry in North America

291

2. THE NORTH AMERICAN SOFTWARE INDUSTRY: CENTRIFUGAL FORCES The Stanford Computer Industry Project (SCIP) divides the software industry into four principal segments: products, services, corporate operations and embedded software. The products segment includes business and productivity software, consumer products and applications, and games. The services segment includes custom services and consulting. The corporate operations segment includes ‘in-house’ programs designed and written to improve overall corporate eﬃciency. And the embedded software segment includes software incorporated into widely available consumer and producer goods, for example, televisions, radios, aircraft, and machine tools (SCIP 1996).1 Unfortunately, the diﬀerent segments are housed in diﬀering sectors of the economy, and reliable data on the size of each segment are therefore diﬃcult to come by. While embedded software is the largest segment overall, the products and services segments are the ‘most visible’ and arguably the most international and they are therefore at the center of our analysis (Barr and Tessler 1996, p. 2). How big are the products and services segments? According to Kyle Eischen, packaged software is currently responsible for approximately onethird of the $470 billion world software products and services industry (Eischen 2000, p. 34). The remaining two-thirds are composed of customized software, system design and software services.2 Nevertheless, the US stands at the center of the market for both products and services. According to the United Nations Conference on Trade and Development, 17 of the world’s 20 largest software companies are North American, and US ﬁrms currently account for more than 50 per cent of the world’s supply of software (UNCTAD 2002, pp. 6–9; see also Eischen 2000, p. 34). Furthermore, the North American advantage appears to be growing. While the US played host to three-quarters of the world’s top software ﬁrms in 1990, it would play host to 85 per cent of their counterparts – approximately half of them based in Silicon Valley or the Route 128 area – by the late 1990s (see Table 12.1). Furthermore, the commanding North American presence is apparently subject to at least four diﬀerent ‘lock-in’ mechanisms: ﬁrst, US domination of the rapidly growing market for packaged software; second, the US role in deﬁning and defending intellectual property rights (IPRs) in software; third, the presence of network externalities and the position of the US in the network; and, fourth, US control of the domain knowledge underpinning software development (see Steinmuller 1996). I shall brieﬂy elaborate upon each of the four mechanisms.

292

Agglomeration in the Americas

Table 12.1

Top 20 software companies in 1990 and 1998–1999 (US$)

Rank

1990 Company

1998–1999

Revenue Headquarters

Company

Revenue Headquarters

1 2

IBM Fujitsu

9842.6 1820.8

Armonk, NY Tokyo, Japan

Microsoft Oracle

3

Digital Equipment Microsoft Computer Associates Hitachi

1529.4

Littleton, MA

6206

1289.9 978.2

Redmond, WA Islandia, NY

Computer Associates SAP Compuware

956.3

Tokyo, Japan

Peoplesoft

1314

933.3

9

Oracle Corp.

695.8

10

Bull Information Hewlett Packard Novell Inc.

600.6

BMC Software Electronic Arts Cadence Design Novell

Cadence Design Adobe Systems Inc. SAS Institute Inc. SAP AG Informix Software Sun Microsystem Sybase Inc.

322.0

Munich, FRG Blue Bell, PA Redwood Shores, CA Louvciennes, FR Palo Alto, CA Provo, UT San Jose, CA San Jose, CA Cary, NC

1304

8

Siemens Nixdorf Unisys Corp.

4 5 6 7

11 12 13 14 15 16 17 18 19 20

Parametric Technology

758.3

442.3 433.1

303.7 240.2 190.4 146.1 137.9 76.7 52.4

Walldorf, FRG Menlo Park, CA Santa Clara, CA Dublin, CA Needham, MA

USA: 75%; Silicon Valley and Route 128: 45%

Parametric Technology Network Associates JD Edwards and Co. Adobe Systems SAS Institute Sybase Misys

19 747 8827

4829 1638

1222

Redmond, WA Redwood Shores, CA Islandia, NY Munich, FRG Farmington Hills, MI Pleasanton, CA Houston, TX

1216

Redwood, CA San Jose, CA

1084

Provo, UT

1018

Needham, MA

990 934

Santa Clara, CA Denver, CO

894

San Jose, CA

871

Cary, NC

867 743

Dublin, CA Evesham, UK

Autodesk

740

Baan

736

Informix Software

735

San Rafael, CA Barneveld, Netherlands Menlo Park, CA

USA: 85%; Silicon Valley and Route 128: 50%

Sources: OECD 1998a, p. 43, Table 24; UNCTAD 2002, p. 7, Table 1; and company reports. UNCTAD actually claims that ‘only two companies in the top twenty listing are non-American high-technology companies, SAP from Germany and Misys from the United Kingdom’, but UNCTAD’s list of the top 20 software ﬁrms includes Baan, a Dutch ﬁrm (UNCTAD 2002, pp. 6–7).

The software industry in North America

293

First, the US is currently responsible for almost 80 per cent of the world’s supply of packaged software, and the packaged software segment is not only disproportionately proﬁtable but is in all likelihood subject to long-term, secular growth (Barr and Tessler 1996, p. 2, footnote 3; see also OTA 1990; Carmel 1997; The Economist, May 25, 1996). After all, the world’s principal software consumers prefer labor-displacing packaged solutions to labor-intensive custom solutions, the custom software segment’s share of the global software market has therefore fallen from 100 per cent to approximately 65 per cent since the 1970s, and packaged software is expected to gain additional market share over time (Eischen 2000, p. 34). Second, the US software industry’s principal competitors are imitators, and the worldwide enforcement of international property rights will tend ‘to shift market power from imitators to innovators’ over time (Maskus 2000, p. 174). In fact, Jagdish Bhagwati has characterized both the World Trade Organization’s agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPs) and the US eﬀort to enforce more aggressive bilateral standards known as ‘TRIPs-plus’ as protectionist eﬀorts – and the US-dominated Business Software Alliance’s active crusade for IPR enforcement would appear to underscore his view (Bhagwati 1999; see also Lohr 2002; Shadlen et al. 2003). Third, the software industry is subject to network externalities, and the US happens to be at the center of the key network.3 The externalities derive from the costs of learning new programs, the frequency of ﬁle sharing among users, and the need for interoperability across a wide range of hardware and software systems. According to Michael Borrus and John Zysman, the principal North American software ﬁrms (for example, Adobe, Microsoft and Netscape) not only ‘own key technical speciﬁcations that have been accepted as de facto product standards in the market’ but have ‘created a universe of licensees who produce to the standard and add value to its use – just as applications software ﬁrms like Word Perfect, PC assemblers like Compaq, peripherals producers like Canon, or content providers like Grolliers, all produce to Microsoft’s Windows operating system standards’ (Borrus and Zysman 1997, p. 17; see also Steinmuller 1996; Egan 1998; Eischen 2002a). And, fourth, software ﬁrms tend to agglomerate in ‘regions with speciﬁc industry, social or economic domain-knowledge’ (Eischen 2002a, p. 9), and knowledge-intensive regions tend to be found in the developed world. Examples would include the hardware-related software clusters in California, Massachusetts and Texas, the multimedia software cluster in New York City, the entertainment cluster in Los Angeles, and the oil and gas-related cluster in Houston (Egan 1998).

294

Agglomeration in the Americas

The US lead cannot be taken for granted, however, for the so-called ‘software bottleneck’ has apparently created an opportunity for developers overseas. What does the bottleneck entail? While software demand responds to a highly productive, capital-intensive industrial process – that is, the manufacture of microprocessors of ever-increasing size and ability – software supply is circumscribed by a traditional, craft-based activity – that is, software design and programming – and the US is therefore, according to Ted Egan (1997), subject to ‘a chronic shortage of skilled programmers’; see also Barr and Tessler 1998; Eischen 2000; UNCTAD 2002. The data in Figure 12.1 would appear to conﬁrm the existence of a bottleneck. While hardware employment expanded by approximately 25 per cent in the 1990s, software employment more than doubled. The current economic slowdown is likely to undercut rather than eliminate the bottleneck, for the shortage of skilled labor is neither temporary nor cyclical but is, according to Egan, a necessary outgrowth of the mismatch between mass production in hardware and craft production in software (Egan 1997; see also Barr and Tessler 1998). After all, Moore’s Law holds that microprocessor capacity (that is, the physical potential for software deployment) doubles every 18 months, but software development is all but unavoidably labor intensive.4 While advocates of computer-assisted 2500

Employment (000s)

2000

1500

1000

500

Hardware Software and computer services

0 1992

1993

1994

1995

1996

1997

1998

1999

Year Source: USDOC/ESA (2002, Table A-5.1).

Figure 12.1 IT employment trends in the US, 1992–2000 (thousand persons)

2000

295

The software industry in North America

software engineering, knowledge-based programming, object-oriented programming and waterfall models – to name but a few – have promised to rationalize the software development process, their eﬀorts have so far been in vain, and software development therefore remains ‘a very human centered, almost artistic, process’ (Eischen 2000, p. 29; see also Barr and Tessler 1996, p. 8). In fact, the more successful software producers have pursued new sources of labor rather than process or productivity improvement – and have thereby ‘globalized’ the US software industry. On the one hand, they have imported temporary workers on H-1B visas, and more than half of the 331,206 available H-1Bs are therefore utilized by systems analysts and programmers (USINS 2002, Table 8, pp. 10–11). On the other hand, they have outsourced programming activity to ‘external service providers’ (ESPs) overseas, and the US market for oﬀshore software development services is therefore worth an estimated US$7 billion per year (Dalesio 2001).5 Thus, the software industry suﬀers from a seemingly paradoxical pattern of globalization: while the industry’s largest and most successful ﬁrms are North American in origin, their human resources are increasingly foreign. The short-term consequences of North America’s reliance upon foreign programming talent are visible in Figure 12.2. While the demand for software labor has outgrown the demand for hardware labor, software and hardware wages have grown more or less in tandem – suggesting the existence of an ‘exogenous’ solution to the software bottleneck. 90

Average wage (US$000)

80 70 60 50 40 30

Hardware Software

20 10 0 1992

1993

1994

1995

1996 Year

1997

1998

Source: USDOC/ESA (2002, Table A-5.2).

Figure 12.2

IT wages by sector, 1992–2000 (US$000)

1999

2000

296

Agglomeration in the Americas

Table 12.2 Fastest-growing occupations in the US, 2001–2010 (employment and change) Occupation Computer software engineer, applications Computer support specialist Computer software engineer, systems software Network and computer systems administration Network systems and data communications analysts

2000

2010

% change

380 000 506 000 317 000 229 000 119 000

760 000 996 000 601 000 416 000 211 000

100 97 90 82 77

Source: Hecker (2001, Table 3).

Do foreign workers oﬀer a long-term solution to the bottleneck? The United States Immigration and Naturalization Service (INS) suggests that temporary workers in ‘computer-related occupations’ earn a median annual income of US$58 000 – approximately US$10 000 less than a citizen or resident alien in the same position – and that overseas software developers earn signiﬁcantly less (USINS 2002; Matloﬀ 2002). But the Bureau of Labor Statistics expects the demand for software programming talent to double over the course of the next decade, and treats applications and systems software engineers as two of the ﬁve fastest-growing occupations in the US (Table 12.2). The Bureau’s projections may seem optimistic in the current economic climate; however, they cannot be ruled out, for the contemporary world recession has reduced rather than eliminated the demand for software in the core areas of consumption (see Table 12.3). In fact, we can imagine at least two distinct sources of ongoing or renewed demand for oﬀshore programming talent in the early twenty-ﬁrst century: ﬁrst, the short doubling-time in software demand; and second, the ongoing process of hardware-to-software migration. For example, Soumitra Dutta holds that ‘the amount of software code in most consumer products and systems doubles every two to three years’. What accounts for the short doubling time? According to Dutta, it is propelled not only by Moore’s Law but ‘by escalating demands placed on the functionality of software systems’ (Dutta 2001, p. 55). New systems are not only larger and more complicated than their predecessors but are expected to meet more demanding quality and performance standards (see also Barr and Tessler 1996, p. 8). Furthermore, the process of hardware-to-software migration is well underway, for the world’s principal integrated device manufacturers (IDMs) are beginning to replace hardware switches with software switches, hardware

297

37 780 26 329 772 9748 1597 81 065

1994 40 669 35 228 685 11 892 1746 95 695

1995 46 802 35 188 876 12 843 2270 104 659

1996 54 010 35 629 1031 13 692 2887 114 770

1997 65 250 44 678 1111 12 860 3137 135 411

1998 75 006 48 701 1210 15 743 3270 153 552

1999 90 969 56 766 1803 17 271 3438 181 341

2000

96 556 63 020 2059 18 947 3572 196 237

2001

14.4 13.0 15.6 10.2 13.5 13.4

CAGR (%)

Source: WITSA (2002, p. 82); calculations by authors. Note that world total is greater than the sum of regional totals; the world total includes Africa, South Asia, and the Middle East.

CAGR Compound Annual Growth Rate.

33 020 23 538 646 8740 1301 71 549

USA Western Europe Eastern Europe East Asia Latin America World

Note:

1993

Region

Table 12.3 Software spending by principal region and world total, 1993–2001 (US$m)

298

Agglomeration in the Americas

research and development (R&D) with software R&D, hardware patents with software patents, and hardware revenues with software revenues (see, for example, OECD 1998a, p. 5; Eischen 2000, pp. 23, 35, 43; UNCTAD 2002, p. 5). While computer and oﬃce equipment manufacturers such as Apple, Hewlett-Packard, IBM and Sun Microsystems have frequently been characterized as software companies disguised as hardware companies, they are not unique. A recent story in Business Week holds that ‘the “value” in telecoms is shifting from hardware to software’ as well: ‘Operators used to spend 80% of their capital budgets on hardware and just 20% on software, says Amrish Kacker, a senior consultant at researcher Analysys Consulting in Cambridge, England. Now, the software portion is on track to top 35% within ﬁve years’ (Business Week, November 4, 2002, p. 45). In fact, the overall ratio of software to hardware spending has grown from approximately 1:10 to 1:1 since the early 1970s (see WITSA 2002, p. 35). The implications of growing demand for software programming talent are unclear. While foreign workers – and the current recession – have taken pressure oﬀ of the US software labor market in the short run, their presence may threaten the industry in the long run, for lower wages inhibit training and entry into IT-related ﬁelds and thereby ‘increase the country’s reliance on foreign talent even further’ (Glanz 2001, Section 4, p. 3). The short-term consequences are already apparent in Silicon Valley, where approximately one-third of the skilled engineering labor force is composed of immigrant workers (Saxenian 1999, p. v). What are the likely long-term consequences? ‘So far’, according to Peter Freeman and William Aspray, ‘the US still dominates the software industry, perhaps because the capacity of foreign labor sources is strictly limited by the numbers of highly educated individuals and by the educational infrastructures in other countries. However, it is hard to know how long this domination will last’ (Freeman and Aspray 1999). As more and more countries including Chile, China, Mexico and Korea – try to keep their talent at home, the industry may undergo a pronounced structural and geographic shift (Glanz 2001; Brown and Kirkpatrick 2002). On the one hand, the attractions of oﬀshore software development are likely to expand as foreign programmers gain skill and experience in a growing array of leading-edge projects and applications. After all, Rita Terdiman and Tom Berg hold that the advantages of oﬀshore development have already ‘gone way beyond cost – although current economic conditions have caused cost to once again become the primary reason customers are exploring oﬀshore today. Other drivers – which will resurface once the global economy revives – are time to market, an available and ﬂexible labor pool, quality, higher productivity and a 24-hour workday for support activities’ (Terdiman and Berg 2001, p. 4). On the other hand, the population of oﬀshore development locations

The software industry in North America

299

is likely to grow, for the ‘traditional’ programming platforms – the three Is – have inspired a growing array of imitators including China, Russia, the Philippines, Mexico and a number of Caribbean locations (Terdiman and Berg 2001; Bansal et al. 2001; UNCTAD 2002). Consequently, Forrester Research, a Cambridge, Massachusetts-based consulting ﬁrm, expects the US to lose more than three million software programming jobs by 2015 (Schwartz 2003), and a number of industry sources have oﬀered similar forecasts (Brown and Kirkpatrick 2002).

3. THE NORTH AMERICAN SOFTWARE INDUSTRY: CENTRIPETAL FORCES The decline of the North American software industry is anything but a foregone conclusion, however, for the scale and scope of oﬀshore outsourcing are ultimately circumscribed by barriers to command, control and communications. After all, the process of software development requires ‘constant interaction between designers, programmers, and end-users’ and the binding constraints on productivity enhancement are therefore communications related rather than technological (Eischen 2000, pp. 29–30).6 In fact, the communications-intensive software production process tends to be characterized by diseconomies of scale and scope: The costs of C3 – and therefore the costs of production – are inversely related to the size and degree of dispersal of the software development team. The process of programming is therefore most eﬃciently performed by a single person or, failing that, in a single location. And the oﬀshore programming model is therefore a decidedly second-best alternative to onshore design and development. The broader implications are in all likelihood obvious. While the omnipresent demand for low-cost, highly skilled labor tends to foster a worldwide search for software programming talent, the need for C3 tends to inhibit – if not arrest – the process of dispersal. A number of recent examples – including the pall cast upon South Asian producers by the recent Indo-Pakistani ﬂare-up, the corresponding growth of interest in socalled nearshore (as opposed to oﬀshore) programming options, and the emergence of security concerns among critics of outsourcing in the US – serve to underscore the inherent tension between labor cost reduction and C3 enhancement (that is, transaction cost reduction) as well as the corresponding socio-political limits to oﬀshore software development more generally (Moore 2002; Ray 2002; Schwartz 2003). Thus, the question is not simply whether opportunities for oﬀshore software development will grow but where they will grow and what they will

300

Agglomeration in the Americas

look like. Will oﬀshore development remain concentrated in the three Is or expand to up-and-coming regions in East Asia, Eastern Europe and Latin America? Will oﬀshore programming activity stay conﬁned to ‘costsensitive, trailing-edge projects’ (Terdiman and Berg 2001, p. 4) or will cutting-edge activities move oﬀshore as well? And, ﬁnally, which countries will lure which types of investment and why? I hope to answer these questions, and thereby unpack the underpinnings of software export success, by exploring the correlates of North American software imports in multivariate context.

4.

DATA ANALYSIS

The United States National Science Board (NSB) has published annual data on US software imports from the 43 leading exporting nations in the 1990s (NSB 2002). While the data are bedeviled by a variety of measurement problems, and overall imports are therefore in all likelihood markedly underestimated, the OECD holds that ‘trade statistics can give an indication of the relative size and geographical distribution of cross-border sales of software goods’ (OECD 2000, p. 28; see also OECD 1998b).7 Therefore, I use the NSB data to explore the correlates – but not the precise levels – of software export success at the country level: the rate of tertiary enrollment, the density of research scientists and engineers, the depth of the national hardware infrastructure, the prevalence of English language proﬁciency, the size of the expatriate knowledge network (EKN), the degree of intellectual property protection, the level of political stability, the size of the population and gross domestic product per capita, which serves as a control for unmeasured country-level heterogeneity. The variables and data sources are described in Table 12.3, but I shall brieﬂy elaborate upon the more theoretically salient predictors before presenting my results (see Table 12.4). 1.

Skilled labor The most obvious requirement for oﬀshore software development is a ready supply of skilled programming talent. However, the sources of software programming talent are by no means obvious. The US IT sector is populated not only by university graduates with computer science and engineering degrees but by graduates of ‘nondegree programs produced by these same suppliers, distance education programs, employer-based and for-proﬁt training organizations, and self-study’ (Freeman and Aspray 1999, p. 74). Nor is the US alone. While the Indian ESPs exercise a strong hiring preference for university graduates and engineers, the majority of the ‘engineers working in the industry are, in fact, not trained in software engineering, computer science, or

301

The software industry in North America

Table 12.4

Variables and data sources

Variable

Description

Data sources

Exports

Average value of software exports to the US in millions of dollars for 1997–99 (to control for year-to-year variation) in log form GDP per capita at purchasing power parity in 1995 Population in millions in 1995 PCs per 1000 residents Gross tertiary enrollment ratio

NSB (2002)

GDP/capita Population PCs Tertiary education EKNs

Stability

Expatriate knowledge networks: the number of immigrant scientists and engineers from the host country in the US in 1995 An indicator variable which assumes the value of 1 for countries in which English is an oﬃcial language Percentage of ICT expenditure in GDP Estimated percentage of pirated software deployed in 1995 Index of political stability

Distance to US

Kilometers from the national capital to San Francisco

Engineers 1

Scientists and engineers in R&D per 1000 residents in 1990–96 Absolute number of scientists and engineers in R&D in 1990–96

English

ICT/GDP Piracy

Engineers 2

World Bank (2002) World Bank (2002) World Bank (2002) World Bank (2002) NSF (1993)

USCIA (2002)

WITSA (2002) BSA (2001) Kauﬀman et al. (1999) ‘How far is it?’, www.indo.com/ distance/ UNDP (1999) Author’s calculation based on UNDP and World Bank

related disciplines’ (Arora and Athreye 2002, p. 263). On the contrary, they are mechanical, chemical and electrical engineers who have until recently been underemployed by the slow-growing Indian economy. In other words, the software industry’s key ingredients, skilled workers, tend to have varied social and educational origins. Many are self-taught, educated informally, or trained on the job. Consequently, the institutional foundations of software success are unclear. While a large body of literature underscores the need for specialized training and tertiary IT education, others suggest that ‘with proper training, bright graduates from any ﬁeld would suﬃce’ (Arora and Athreye 2002, p. 263). Thus, I proxy the human resource base with three diﬀerent measures: the gross tertiary enrollment ratio; the number of

302

2.

3.

4.

Agglomeration in the Americas

scientists and engineers in research and development per 1000 residents; and the absolute number of scientists and engineers in R&D. Hardware infrastructure A second key underpinning of software export success is a basic hardware infrastructure. The principal oﬀshore programming platforms have traditionally featured an abundance of personal computers (PCs) and servers, a reliable power supply, and a ready supply of bandwidth. While the growing use of broadband and satellite communications, video-conferencing equipment, and sophisticated encryption technologies will likely increase the barriers to entry in the near future, the PC remains the backbone of the software industry, and I therefore proxy the national hardware infrastructure with a common and widely used measure: the number of PCs per 1000 residents. English language proﬁciency The role of the English language remains controversial in the software ﬁeld. The earliest programming platforms – the three Is – are all English-speaking countries, and a number of countries have therefore come to view lack of English proﬁciency as an all but insuperable barrier to entry. While Costa Rica has undertaken a large-scale English-as-a-second-language campaign, Chile has for the most part abandoned the North American market in favor of ‘packaged and bespoke software that meets the special needs of local Spanish-speaking industries’ (UNCTAD 2002, p. 20). Nevertheless, Taiwan, Malaysia and a number of other countries have made great strides in software programming and service provision without the beneﬁt of the English language, and the importance of English proﬁciency therefore remains unclear. I proxy English language knowledge with an indicator variable which assumes the value of 1 for every country in which English is an oﬃcial language. Expatriate knowledge networks While widespread programming skills are a necessary precursor to software export success, they are not necessarily suﬃcient, for individual skills are arguably no more useful than ‘the networks that mobilize and activate them’ (Meyer 2001, p. 101). Perhaps unsurprisingly, the relevant networks are frequently migratory. After all, the three Is are home to three of the world’s great diasporas, and their emigrant scientists and engineers have arguably played a critical role in their software export success. Richard Heeks and Brian Nicholson hold that a ‘diaspora formed the basis for contacts and then contracts that set each country’s software export market in motion. Scratch beneath the surface of all early exports and you ﬁnd a US- or Europe-based expatriate’ (2002, p. 7; see also Saxenian 1999, 2002a, 2002b). While an ideal measure of the strength of expatriate knowledge networks is anything but obvious, the extant literature

The software industry in North America

5.

303

suggests that EKNs are underpinned by ‘generic’ rather than ‘speciﬁc’ scientiﬁc and engineering skills (Khadria 2001), and I therefore use the absolute number of immigrant scientists and engineers in the US as a proxy for the level of a given country’s exposure to – or engagement with – the US scientiﬁc and technical community. Government support The literature on oﬀshore software development tends to underscore the importance of government support. But the principal components of government support are not obvious. For example, the Indian industry’s achievements have been attributed to both the ‘enlightened “hands oﬀ” policies of the government of India’ (Arora and Athreye 2002, p. 255) and, on the contrary, ‘well orchestrated’ government planning (Mir et al. 2000, p. 20). Which perspective is accurate? The answer is best found by examining the speciﬁc elements of government support. While a supportive government is expected to provide essential public goods including education, training and a reliable telecommunications infrastructure, it is also expected to provide tax incentives, low-cost capital, and/or the protection of intellectual property. The last factor is considered doubly important, for foreign investors are allegedly unlikely to entrust their technology to partners who are prone toward or capable of ‘piracy’ and ‘theft’, and local ﬁrms are said to be unlikely to innovate if they cannot capture the returns to their investment (see Teran 2001; Terdiman and Berg 2001; Heeks and Nicholson 2002).

The aforementioned factors – skilled labor, hardware infrastructure, English language proﬁciency, expatriate knowledge networks and government support – have almost invariably been characterized as vital components of oﬀshore software development. They are occasionally joined by other factors including a large domestic market for IT, political stability and spatial proximity to the US (see, for example, Terdiman and Berg 2001; Arora and Athreye 2002). But the literature on the global software trade tends to be anecdotal and I know of no systematic, comparative investigation of the correlates of software export success. I hope to ﬁll the lacuna by estimating a number of multivariate regression models. Speciﬁcally, I regress the natural logarithm of the average value of software exports from each of the 43 countries tracked by the NSB for 1997, 1998 and 1999 on ﬁve diﬀerent vectors of independent variables: a baseline model which includes measures of GDP per capita, population, PCs, tertiary education and expatriate knowledge networks (Model 1); a restricted model which explores the eﬀects of the C3 variables – EKNs, English and distance from the US – while controlling for GDP per capita and population (Model 2); a more extensive model which incorporates a number of

304

Table 12.5

Agglomeration in the Americas

Regression results

Variables

Model 1

Model 2

Model 3

GDP/cap

0.0001498** (0.0000741) 0.0005837 (0.0013001) 0.0004361 (0.0047832) 0.0147889

0.0001853*** (0.0000456) 0.0003035 (0.0014819)

0.0001699* (0.000099) 0.0005911 (0.0011627)

0.0001923*** 0.0001465*** (0.000068) (0.0000541) 0.0007398 0.0015272 (0.0010967) (0.0014726)

0.0000354*** (0.00000931)

0.0000439*** (0.0000125) 0.7151518 (0.8594462) 0.000091 (0.0001708)

0.0000346*** (0.00000814)

0.0000322*** (0.00000886)

Population PCs Tertiary education EKNs English Distance ICT/GDP

Model 4

Stability

0.0988237 (0.2092519)

Engineers 1 Engineers 2

No. of obs. R2

0.0000347*** (0.00000792)

0.1600734 (0.36869) 0.0059089 (0.0431354) 0.1512388 (0.6547594)

Piracy

Constant

Model 5

2.69928*** (0.9405544) 43 0.46

3.485141 (2.102715) 43 0.47

3.511296 (4.302355) 42 0.52

2.43339*** (1.028163) 38 0.38

0.00000329*** (0.00000127) 2.251296** (1.004859) 38 0.43

Note: ***, **, * statistically signiﬁcant at 99%, 95% and 90% conﬁdence levels, respectively.

additional control variables (Model 3); and two ﬁnal models which utilize the alternative measures of human capital (Models 4 and 5). The results are displayed in Table 12.5; unstandardized regression coeﬃcients are displayed above their parenthesized robust standard errors. Model 1 attributes the bulk of the explained variation to two variables: GDP per capita and expatriate knowledge networks. While the GDP per capita variable is highly collinear with the PCs variable (r 0.89) and would therefore appear to capture at least part of the eﬀect of the national hardware infrastructure as well as otherwise unmeasured attributes of economic development, the EKN variable is unambiguously signiﬁcant and would appear to account for varying levels of access to North American knowledge, capital and product markets. Are the substantive as well as the statistical eﬀects signiﬁcant? We can explore the substantive impact of either GDP per capita or EKNs by

The software industry in North America

305

estimating the predicted value of a 1000 unit increase in either variable when the individual predictors are set to their sample means. The calculations suggest that a US$1000 increase in GDP per capita will generate more than US$2 million worth of exports and a 1000 migrant increase in the EKN will yield approximately US$500 000 worth of exports annually. Nevertheless, the software import data are in all likelihood dramatically underestimated and we would therefore be well advised to use the remaining models to test the plausibility – and not the point estimates – of the baseline equation. While Model 1 attributed variation in software exports to GDP per capita and expatriate scientists and engineers, and therefore underscored the importance of C3, Model 2 conserves scarce degrees of freedom by dropping PCs and tertiary education and thereafter asks whether the English language and proximity to Silicon Valley are as important as – or perhaps oﬀer an alternative to – bridge-building EKNs. The results suggest that they are not, for the coeﬃcients are opposed to their predicted relationships and are in any case insigniﬁcant. Model 3 adds controls for the eﬀects of overall ICT spending, software piracy and political stability; the control variables are neither meaningful in their own right nor threats to the basic relationships established in Model 1. However, the individual control variables tend to co-vary with GDP per capita and the negative ﬁndings should therefore be interpreted with extreme caution.8 In fact, the high correlations between GDP per capita and a number of related predictors of software export success (for example, PCs, ICT spending, teledensity, intellectual property protection, stability and so on) constitute all but insurmountable obstacles to robust estimates of individual regression coeﬃcients, and I therefore conclude by re-examining a seeming anomaly from Model 1 – the negative ﬁnding on tertiary education – by incorporating two alternative measures of human capital: the number of scientists and engineers in R&D per 1000 residents and the absolute number of scientists and engineers in R&D. Unfortunately, the data are available for only 38 of the 43 countries in my sample; the results, however, are clear. While Model 4 treats scientists and engineers in R&D as a relative factor (that is, the number of scientists and engineers in R&D per 1000 residents), and ﬁnds no signiﬁcant eﬀect, Model 5 incorporates an absolute measure and implies that 1000 additional scientists and engineers in R&D will account for a statistically signiﬁcant but substantively questionable US$50 000 worth of exports per year at the sample mean. Thus, the bulk of the evidence suggests that relatively well-oﬀ countries with domestic scientiﬁc resources and access to expatriate knowledge networks hold the best prospects of conquering the North American software market. The results underscore the sociological as well as the economic

306

Agglomeration in the Americas

aspects of software development. They suggest that an eﬀective scientiﬁc and educational training apparatus is no more than one aspect of the software success story, and not necessarily the most important one. Thus, the real winners in the struggle to imitate – if not necessarily replicate – the achievements of Silicon Valley and Route 128 will be members of transnational communities who keep their feet in both worlds. What will victory entail? The evidence I have adduced provides no clearcut answer. While IT optimists wax enthusiastic about the software industry’s potential contribution to sustainable development, their critics invoke nightmarish visions of deskilling, wage compression and the emergence of software sweatshops in the developing world (see, for example, Prasad 1998). The truth, I suspect, will lie somewhere in between. After all, the universe of oﬀshore production sites already appears to be sorting itself into a hierarchy which is in many ways akin to the hierarchies which characterize more mature commodity chains, and the ‘stratiﬁcation of oﬀshore countries based on cost and skill sets’ (Robb 2002), like the stratiﬁcation of apparel, footwear and electronics manufacturers, is sure to generate disparate outcomes.

5.

DISCUSSION

The results presented in Table 12.4 are reasonably compelling. But they should not be accepted at face value, for the regression equations I have presented suﬀer from at least four distinct limitations. First, the key variables are not precisely measured. The data on software imports are in all likelihood vastly – if more or less consistently – underestimated; the data on human capital and expatriate knowledge networks proxy but do not precisely measure their underlying concepts; and the key control variables – for example, GDP per capita, PCs, ICT in GDP, stability, piracy – tend to be highly collinear. Second, the data are cross-sectional. While the software industry’s relative youth means that the data can arguably be interpreted as implicitly longitudinal (that is, they track the industry’s birth and growth over the past quarter of a century), it does not aﬀord the precision of a truly longitudinal research design and it therefore suﬀers from a number of well-known obstacles to robust causal inference. Third, the data include only countries which have achieved a minimal level of software export to the US and it therefore suﬀers from two diﬀerent varieties of selection bias. First, it excludes countries which are unsuccessful in the ﬁrst place. And, second, it excludes exporters who successfully target non-US markets.

The software industry in North America

307

And, ﬁnally, to the extent that the data on scientists and engineers in R&D oﬀer an adequate proxy for the national human resource base, they comprise a neat interpretation of the EKN coeﬃcients. Do expatriates ‘mobilize and activate’ underlying human resources? Or are they simply another manifestation of human resource abundance? The latter interpretation is certainly plausible. After all, India’s scientiﬁc diaspora has frequently been interpreted from a supply–push rather than a demand–pull perspective (see, for example, OECD 2000, p. 132). Which interpretation is superior? While a deﬁnitive answer will almost certainly require additional research, the surprisingly small correlation between my measure of EKNs (that is, the number of immigrant scientists and engineers in the US in 1995 per source country) and the absolute number of scientists and engineers in R&D per country (r 0.35) suggests that the eﬀects of domestic human resources are to a large degree mediated by the depth of EKNs and not vice versa. Furthermore, the data I have presented have at least two additional desirable features. First, they are intuitively plausible, for a number of analysts have underscored the frequently overlooked socio-cultural underpinnings of successful software development. And, second, they are broadly consistent with an abundance of case-study material, for a large and growing body of evidence suggests that expatriate scientists and migratory entrepreneurs have played a key role in IT development not only in the three Is but also in their more successful imitators (see, for example, Heeks and Nicholson 2002 on Russia, the Philippines, China; and Saxenian 1999, 2002a, 2002b on Taiwan). Finally, the evidence I have adduced has one additional desirable property. It suggests that structural features (for example, language, geography and country size) are not necessarily impediments to software export success. On the contrary, the key variables, GDP per capita notwithstanding, are subject to at least some degree of policy manipulation. Governments can, after all, devote more resources to science and education; and they can encourage the growth and fertilization of expatriate knowledge networks as well – at least within limits.

NOTES *

I would like to thank Danny Breznitz, Seán Ó Riain, Michael Piore, participants at the JETRO conference and, most importantly, Clemente Ruiz Durán for their valuable collaboration and feedback. 1. The Stanford system is one of several currently in use. The OECD (1998a) reviews the various classiﬁcation systems in use in the late 1990s. 2. The packaged software segment includes software which arrives shrink-wrapped, bundled or leased.

308

Agglomeration in the Americas

3. Network externalities exist wherever the value of participation in a network grows with the number of participants, for example, the value of a particular word-processing program is a function of the number of users as well as the functionality of the program itself. 4. See, however, Tuomi (2002) for a skeptical view of Moore’s Law. 5. A number of multinational corporations prefer dedicated ‘oﬀshore development centers’ (ODCs) or joint ventures with local ﬁrms to traditional outsourcing arrangements. But the bulk of oﬀshore development is currently conducted through arm’s-length or subcontracting arrangements 6. Fred Brooks has summarized the limits imposed by the communications-intensive nature of software production (that is, the ‘mythical-man month’ problem) by noting that a single programmer can achieve far more in a year than 12 programmers can achieve in a month (see Brooks 1995 [1974] for the classic formulation as well as Eischen 2002b). 7. The OECD cites a number of obstacles to the precise estimation of software imports and exports. First, the valuation of software is frequently based on the medium of transfer (that is, diskettes, CD-ROMs, or in the case of bundled software computer hardware) rather than the underlying content (that is, the software itself). Second, systems of valuation tend to be inconsistent over time, space and product. And, third, a good deal of software is transferred within the same ﬁrm and is therefore subject to transfer pricing (OECD 1998b, p. 4; 2000, p. 28). 8. The Pearson correlation coeﬃcients are 0.77 for ICT/GDP, – 0.80 for piracy and 0.80 for stability.

REFERENCES Arora, Ashish and Suma Athreye [2002], ‘The software industry and India’s economic development’, Information Economic and Policy, 14; 253–73. Arora, Ashish Alfonso Gambardella and Salvatore Torrisi (2001), ‘In the footsteps of Silicon Valley? Indian and Irish software in the international division of labor’, SIEPR Discussion paper No. 00–41, Stanford University. Bansal, Anjali, Noshirkaka, Manish Kejriwal and Sunish Sharma (2001), Oﬀshoring software product development to India, New York: McKinsey & Co. Barr, Avron and Shirley Tessler (1996), ‘The globalization of software R&D: the search for Talent’, Position paper for the Council of Foreign Relations Study Group on the Globalization of Industrial R&D, December 12, Stanford: Stanford Computer Industry Project (SCIP). Barr, Avron and Shirley Tessler (1998), ‘Strategies for survival: the software talent shortage’, Intelligence Magazine, March, pp. 13–17. Bhagwati, Jagdish (1999), ‘Economic freedom: prosperity and social progress’, Keynote speech delivered to the Conference on Economic Freedom and Development, Tokyo, June. Borrus Michael and John Zysman (1997), ‘Wintelism and the changing terms of global competition: prototype of the future?’, Berkeley Roundtable on the International Economy, Working Paper 96B, Berkeley: BRIE. Brooks, Fred (1995 [1974]), The Mythical Man-Month: Essays on Software Engineering, Reading, MA: Addison-Wesley. Brown, Eryn and David Kirkpatrick (2002), ‘The reverse brain drain’, Fortune, 146 (November 11), 39. Business Software Alliance (BSA) (2001), Sixth Annual BSA Global Software Piracy Study, Washington, DC. Business Week, ‘Saving Ericsson’, November 4, 2002.

The software industry in North America

309

Carmel, Erran (1997), ‘American hegemony in packaged software trade and the “culture of software” ’, The Information Society, 13, 125–42. Correa, Carlos (1996), ‘Strategies for software exports from developing countries’, World Development, 24 (1), 171–82. Dalesio, Emery (2001), ‘Firms hunt tech talent overseas’, Milwaukee Journal Sentinel, November 26, Milwaukee, Wisconsin. Dedrick, Jason, S.E. Goodman and Kenneth Kraemer (1995), ‘Little engines that could: computing in small energetic countries’, Communications of the ACM, 38 (5), 21–6. Dutta, Soumitra (2001), ‘The importance of organizational leadership for creating technology excellence’, in Global Competitiveness Report 2001–2002, World Economic Forum, Oxford: Oxford University Press. The Economist, ‘The importance of being American’, May 25, 1996. The Economist, ‘Silicon envy’, February 20, 1999. Egan, Edmund (1997), The spatial dynamics of the U.S. computer software industry’, unpublished PhD dissertation, Department of City and Regional Planning, Berkeley: University of California. Egan, Ted (1998), ‘Structural change and spatial dynamics of the US software industry’, paper presented at the Sloan Foundation Globalization Workshop, Duke University, September. Eischen, Kyle (2000), ‘Information technology: history, practice and implications for development’, Center for Global, International and Regional Studies (CGIRS) Working Papers Series WP#2000-4, Santa Cruz: University of California. Eischen, Kyle (2002a), ‘Mapping the microfoundations of informational development: linking software processes, products and industries for global trends’, Center for Global, International and Regional Studies (CGIRS) Working Papers Series WP#2002-2, Santa Cruz: University of California. Eischen, Kyle (2002b), ‘Software development: an outsider’s view’, Computer, May, pp. 36–44. Evans, Peter (1995), ‘Embedded Autonomy: States and Industrial Transformation, Princeton, NJ: Princeton University Press. Freeman, Peter and William Aspray (1999), The Supply of IT Workers in the United States, Computing Research Associates, Washington, DC. Glanz, James (2001), ‘Trolling for brains in international waters’, New York Times, April 1, section 4, p. 3. Hecker, Daniel (2001), ‘Occupational employment projections to 2010’, Monthly Labor Review, November, 57–84. Heeks, Richard and Brian Nicholson (2002), ‘Software export success factors and strategies in developing and transitional economies’, University of Manchester Institute for Development Policy and Management Working Paper No. 12. Helleiner, G.K. (1990), ‘Trade strategy in medium-term adjustment’, World Development, 18 (6), 879–97. Kauﬀman, Daniel, Art Kraay and Pablo Zoido Lobaton (1999), Aggregating Governance Indicators, Washington, DC: World Bank. Khadria, Binod (2001), ‘Shifting paradigms of globalization: the twenty-ﬁrst century transition towards generics in skilled migration from India’, International Migration, 39 (5), 45–71. Lohr, Steve (2002),. ‘The intellectual property debate takes a page from 19thcentury America’, New York Times, October 4, p. C-4.

310

Agglomeration in the Americas

Maskus, Keith (2000), Intellectual Property Rights in the Global Economy, Washington, DC: Institute for International Economics. Matloﬀ, Norman (2002), ‘Debunking the myth of a desperate software labor shortage’, Testimony to the US House Judiciary Committee, Subcommittee on Immigration, April 21, 1998. Updated September 10, 2002, http://heather.cs. ucdavis.edu/itaa.real.mtm. Meyer, Jean-Baptiste (2001), ‘Network approach versus brain drain: lessons from the diaspora’, International Migration, 39 (5), 91–110. Mir, Ali, Biju Mathew and Raza Mir (2000), ‘The codes of migration: contours of the global software labor market’, Cultural Dynamics, 12 (1), 5–33. Moore, Stephanie (2002), ‘Region at risk of war: contingencies a must for Indian outsourcing’, Giga Information Group, Cambridge, MA, June 4. National Science Board (2002), Science and Engineering Indicators 2002, Arlington, VA: National Science Foundation. National Science Foundation (1993), Scientists and Engineers Statistical Data System, http://sestat.nsf.gov/, accessed November 2002. Nicholson, Brian (2001), Global Software Outsourcing: The Solution to the IT Skills Gap, Berlin: Anglo-German Society for the Study of Industrial Society. Oﬃce of Technology Assessment (OTA) (1990), Computer Software and Intellectual Property, Washington, DC: OTA. Organization for Economic Cooperation and Development (OECD) (1998a), ‘The software sector: a statistical proﬁle for selected OECD countries’, Committee for Information, Computer, and Communications Policy, January 29, OECD, Paris. Organization for Economic Cooperation and Development (OECD) (1998b), ‘Measuring electronic commerce: international trade in software’, Committee for Information, Computer, and Communications Policy, April 30, OECD, Paris. Organization for Economic Cooperation and Development (OECD) (2000), OECD Information Technology Outlook 2000: ICTs, E-commerce, and the Information Economy, Paris: OECD. Prasad, Monica (1998), ‘International capital on “Silicon Plateau”: work and control in India’s computer industry’, Social Forces, 77 (2), 429–52. Ray, Saritha (2002), ‘India success in software is set back by war talk’, New York Times, June 6, p. W-1. Robb, Drew (2002), ‘5 top trends in oﬀshore outsourcing’, Datamation, December 17. Saxenian, Annalee (1996), Regional Advantage: Culture and Competition in Silicon Valley and Route 128, Cambridge, MA: Harvard University Press. Saxenian, Annalee (1999), Silicon Valley’s New Immigrant Entrepreneurs, San Francisco: Public Policy Institute of California. Saxenian, Annalee (2002a), ‘Brain circulation: how high-skill immigration makes everyone better oﬀ’, Brookings Review, Winter, 28–31. Saxenian, Annalee (2002b), Local and Global Networks of Immigrant Professionals in Silicon Valley, San Francisco, CA: Public Policy Institute of California. Schware, Robert (1992), ‘Software industry entry strategies for developing countries: a “walking on two legs” proposition’, World Development, 20 (2), 143–64. Schwartz, John (2003),. ‘Experts see vulnerability as outsiders code software’, New York Times, January 6, p. C-1. Shadlen, Kenneth, Andrew Schrank and Marcus Kurtz (2003), ‘The political economy of intellectual property protection: the case of software’, DESTIN Working Paper Series No. 03–40, Development Studies Institute, London: London School of Economics.

The software industry in North America

311

Stanford Computer Industry Project (SCIP) (1996), ‘The US domination of worldwide software product sales increased in 1995’, Press Brieﬁng, October 17. Steinmuller, W. Edward (1996), ‘The U.S. software industry: an analysis and interpretive history’, in David Mowery (ed.), The International Computer Software Industry: A Comparative Study of Industry Evolution and Structure, Oxford: Oxford University Press, pp. 15–52. Teran, Horacio (2001), ‘Intellectual property protection and oﬀshore software development: an analysis of the U.S. software industry’, Minnesota Intellectual Property Review, 2 (1), 1–50. Terdiman, Rita and Tom Berg (2001), ‘Oﬀshore application outsourcing’, Gartner, Inc., Stanford, Connecticut: September 24. Tuomi, Ilkka (2002), ‘The lives and death of Moore’s Law’, First Monday, 7 (11), http:www.ﬁrstmonday.org/issues/issue7_11/tuomi, accessed November 2002. United Nations Conference on Trade and Development (UNCTAD) (2002), Changing Dynamics of Global Computer Software and Services Industry: Implications for Developing Countries, New York: UNCTAD. United Nations Development Program (UNDP) (1999), Human Development Report, CD-ROM, New York: UNDP. United States Central Intelligence Agency (US CIA) (2002), CIA World Factbook 2002, www.cia.gov/cia/publications/factbook. United States Department of Commerce/Economics and Statistics Administration (USDOC/ESA) (2002), Digital Economy 2002, Washington, DC: DOC. United States Immigration and Naturalization Service (USINS) (2002), Report on Characteristics of Specialty Occupation Workers (H-1B): Fiscal Year 2001, Washington, DC: INS. World Bank (2002), World Development Indicators 2002, Washington, DC: World Bank. World Information Technology and Services Alliance (WITSA) (2002), Digital Planet 2002, Arlington, VA: WITSA.

13. Mexico: the management revolution and the emergence of the software industry Clemente Ruiz Durán* 1.

INTRODUCTION

Since the enactment of the North American Free Trade Agreement (NAFTA) in 1994, the process of integration between the US and Mexico has increased, trade has multiplied by a factor of eight, and massive ﬂows of investment have gone to Canada and Mexico from the US. Indeed, by the end of 2001, the accumulated ﬂow of US direct investment since NAFTA’s signing had reached US$65 billion. Trade and investment ﬂows have reshaped economic geography through the development of new supply chains in the electronics, automobile, auto parts and garment industries and, more recently, the software industry. Information technology (IT) has helped develop manufacturing processes far from their design centers, enabling them to be updated and modiﬁed from the headquarters in the US. The IT market has reached an estimated US$6.5 billion, of which a 57 per cent share comes from the hardware industry, 32 per cent from services and 11 per cent from the software industry. The market location of industry in 2001 shows that clusters had developed in some speciﬁc regions of Mexico, for instance in the Distrito Federal metropolitan area (which includes the federal state of Mexico), holding 48.7 per cent of the total number of ﬁrms, followed by the states of Nuevo Leon and Jalisco, with 12.1 and 4.4 per cent, respectively. But as Michael Peneder (2001) has suggested, clusters imply more than simple density; they tend to produce and rely upon positive externalities, such as knowledge spillovers between ﬁrms, specialized inputs and services from supporting industries, and a geographically pooled labor market for specialized skills. Software clusters share these positive externalities with other clusters, but the software industry diﬀers from the traditional concept of industry because it is based on knowledge rather than on machines. It requires an institutional framework that supports the existence 312

313

Source:

Select IDC (2000).

36.1 10.7 6.9 4.6 3.3 0.6 2.6 2.6 2.5 3.1 2.1 2.3 1.2 1.4 1.7 1.7

45.7 13.3 8.3 4.1 3.0 0.5 2.3 2.3 2.3 1.6 1.9 1.2 0.7 1.2 0.9 0.9

45.0 13.4 8.6 4.2 3.0 0.6 2.4 2.4 2.3 1.7 1.9 1.3 0.7 1.2 0.9 0.9

40.9 12.1 7.8 4.4 3.1 2.9 2.5 2.5 2.4 2.3 2.0 1.7 1.4 1.3 1.3 1.3

40.9 53.0 60.9 65.2 68.3 71.2 73.7 76.2 78.6 80.9 82.9 84.7 86.1 87.4 88.7 89.9

Hardware Software Services Total Cumulative Guerrero Hidalgo Yucatán Coahuila Morelos Aguascalientes Durango Colima Zacatecas Baja California Sur Tlaxcala Oaxaca Nayarit Quintana Roo Tabasco Campeche

Geographical distribution of the IT market in Mexico, 2001 (%)

Distrito Federal Nuevo León México Jalisco Veracruz Chihuahua Baja California Puebla Guanajuato Tamaulipas Sonora Michoacán Chiapas Querétaro San Luis Potosí Sinaloa

Table 13.1

1.6 1.6 1.2 1.5 1.2 1.1 1.1 3.8 0.6 0.5 0.5 0.5 0.4 0.4 0.3 0.2

0.9 0.8 0.7 1.3 0.6 0.6 0.6 2.1 0.3 0.3 0.3 0.3 0.2 0.2 0.2 0.1

0.9 0.9 0.7 1.4 0.7 0.6 0.6 2.1 0.3 0.3 0.3 0.3 0.3 0.2 0.2 0.1

1.2 1.2 0.9 0.9 0.9 0.8 0.8 0.6 0.4 0.4 0.4 0.4 0.3 0.3 0.2 0.2

91.2 92.4 93.3 94.3 95.2 96.0 96.8 97.4 97.9 98.3 98.7 99.0 99.4 99.7 99.9 100.0

Hardware Software Services Total Cumulative

314

Agglomeration in the Americas

of educational networks; connectivity and computer density are essential, and a skilled entrepreneurial network must be present. Furthermore, it helps if there is an open business environment and the public sector pushes for policies to support software industries. Mexico has developed a domestic market for software of about US$600 million, the 24th largest in the world.1 Its market is larger than those of export-oriented countries like India and Ireland. But the size of the market if compared in per capita expenditure reaches only US$6, which is low compared with the US, where the expenditure is US$340, or the Scandinavian countries, where expenditures are in the range of US$250– 300. (See Figure 13.1.)

2. HOW AND WHY? DRIVING FORCES IN THE MEXICAN SOFTWARE INDUSTRY The emergence of the software industry was a byproduct of the crisis of the 1980s. Before that time, protection allowed management to be loose, and all IT was imported from the US; most of the installed capacity came from large IBM mainframes. In this period, most of the operations were managerial, and in some cases there were speciﬁc production processes that were controlled by IT. Most of the businesses involved in the operations were large national ﬁrms and transnational subsidiaries. The practice at this time was for large mainframes to come bundled with software; there was no freestanding software market. As large producers were the ones to hold the code, there was no place for local innovation. With the introduction of personal computers (PCs) in the late 1970s, computers became accessible to other groups of businesses, the software market was increasingly decoupled from the hardware market, and the door to innovation was opened. Technology was not monopolized by one single company: Apple introduced the PC, but IBM immediately introduced its own model. So there was space to produce software in three large areas – the old mainframe technology and two platforms in the PC area. Local innovators immediately appeared, and as there was no need for large investments, middle- and upper-class engineering students who could aﬀord to buy PCs were able to begin developing software. It could be argued that the environment was open to innovation when the debt crisis surfaced. The debt crisis of the 1980s called the whole production model of the 1970s into question, with particular emphasis on its carelessness. Large sums of money from the oil booms obscured and in many ways allowed ineﬃciencies in production caused by lax management control. Reduction of activity in the 1980s (the average rate of growth of GDP per capita,

315

—1

Figure 13.1

Ln size of market

6

0

China

1

Russia

2

3

Spain

4

Portugal

Taiwan

South Africa Korea

Ln of per capita expenditure

Mexico Poland

Brazil

Israel

Austria

5

6

Denmark Norway

Sweden Switzerland

Finland Singapore Ireland

Belgium

USA

Netherlands

Canada

Germany UK France

Australia

Italy

Japan

Large market, Large per capita expenditure

Size of market and per capita expenditure in software

India

7

8

9

10

11

12

7

316

Agglomeration in the Americas

which was 3.4 per cent from 1960 to 1981, fell at an average rate of –1.2 per cent from 1982 to 1989) led businesses to become cost conscious and, in fact, encouraged a management revolution; forced to look for eﬃciency measures, managers began to use IT to control their processes. This management revolution was reinforced by additional elements: 1.

2.

3.

4.

The opening of the economy in the late 1980s: as margins collapsed, domestic prices were forced into line with global prices. This was a real crisis for businesses in Mexico, as they were used to having a large margin; adapting to the new environment implied a learning process, not only in cost management, but also in the control of large ﬂows of information. In the production processes, as management practices were upgraded with the introduction of Japanese production techniques (just in time, zero inventory, quality control, quality circles) all large businesses looked to modernize through the introduction of IT. As small- and medium-sized ﬁrms invested in computer systems, they demanded programs appropriate to the size and needs of their operations and became aware that Microsoft programs introduced with the PC revolution were for more general use than they required for their management. The introduction of electronic communication, e-mail, and the Internet opened new innovation areas.

All these factors opened fertile territory for systems engineers willing to get into the design of software to satisfy the needs of a business world that was forced to modernize. In this case, domestic market needs drove the growth of the software industry, not foreign demand as in India or Ireland. However, as there was no large driving force in the process, market developments were slow and quite decentralized; our analysis ﬁnds no orderly pattern of development as allegedly found in Bangalore, for example (Balasubramanyam and Balasubramanyam 2000). It can be argued that the Mexican software industry has followed the pattern of the management revolution, and that it can be divided into three periods (see Figure 13.2): 1. 2.

3.

The ﬁrst wave of management reform came from large businesses that were seeking IBM or similar solutions. Second-wave management appeared in the 1980s as the PC became common in medium-sized enterprises. At this time software was still focused on large-business solutions, so system engineers began to develop solutions for the mid-sized range of businesses. In the 1990s a third wave of management appeared as NAFTA was enacted. In this last wave, customized solutions became standardized

Mexico: the management revolution and the software industry

317

Figure 13.2 Management revolution and software industry development in Mexico and packaged software became more common, enabling small businesses to begin to get into the management revolution.

Measuring the Emergence of the Software Industry in Mexico The Mexican software industry grew at an average annual rate of 9 per cent between 1993 and 2001. What does that mean in terms of market size? In 1993 the number of businesses served by IT ﬁrms was 3435; by 1999 the number was reduced to 1301. This meant a large improvement, as it reduced the ratio by two-thirds. What are the segments of the industry that have driven growth? According to the Instituto Nacional de Estadística, Geografía e Informática (INEGI 2002), the industry can be subdivided into four segments: computing consultancy, information processing and elaboration, in-line consulting and software editing and design. This allows us to describe how the value chain behaves (Figure 13.3). In the four segments described by INEGI there are about 2052 ﬁrms, located in three main areas: Distrito Federal, Monterrey and Guadalajara. Absorption in these areas is 38 per cent of the total. Why have they developed in these areas? The interaction of many diﬀerent factors helped this development to take place. Entrepreneurs took advantage of a new market niche, but the market had been developing for the last two decades. Public institutions gave crucial assistance. Furthermore, the market was reorganized to become more competitive: 1.

In Monterrey, development has been associated with the large industry located in the area. The demand of local ﬁrms has been the primary

318

Figure 13.3

0

10

20

Mexican software industry value chain

Computing consultancy

Sotfware editing and design

On-line consultancy

Electronic information processing

30

50

60

Index of value added per worker

40

70

80

90

100

Mexico: the management revolution and the software industry

2.

3.

4.

319

factor driving businesspeople to invest. This has been complemented by demand from US ﬁrms, which have looked to support the development of software factories in the area, as was the case of Softtek by General Electric. Guadalajara has a total of 151 small and medium-sized enterprises (SMEs) that develop software plus IBM, which has developed the second-largest software factory in the country. The SME employees number about 540 and IBM employees in the software factory in El Salto, and Jalisco number about 500. This development has been pushed by the existence of an educational network (13 higher education institutions oﬀer BA degrees related to IT), connectivity infrastructure and an electronics cluster with large original equipment manufacturers (OEMs). In Mexico City a more complex software industry has developed to respond to the demand of local ﬁrms interacting with the largest software transnational, which has located its headquarters in the city. The emerging Aguascalientes cluster focuses on new technologies. Here, local government has applied a cluster approach to the software industry.

This chapter will describe the evolution of these software clusters and the relation it has to US ﬁrms. The main hypothesis is that this process could help develop a more solid software industry than in other regions of the world, due to the interaction of local with international demand; we develop this hypothesis by comparing Mexican software development with the developments of the industry in India and Ireland, which have been based on international demand. The software industry does not respond to the traditional, gradual pattern of industrial development; it requires maturity to develop a code and demands innovation from the outset. The environment for such development requires knowledge clusters that could enable innovation. In Mexico such clusters have been related to what we might call ‘knowledge agglomerations’, where highly skilled populations are able to innovate based on their formal training (graduate school) or learning gained in the workplace. In the last 20 years Mexico has been able to combine both sorts of outcomes. During the 1980s it developed a National Researchers’ System (Sistema Nacional de Investigadores), which today includes approximately 8000 researchers, about one-tenth of them in the engineering area, enabling innovations to be developed. Complementing this system is business training in diﬀerent areas, especially in the electronics industry, where the lack of skilled labor in the 1960s and 1970s led big transnational corporations to develop a training system inside the plants, as was the case with IBM.

320

Agglomeration in the Americas

The geography of this process has been diverse; locations have been correlated to manufacturing clusters in diﬀerent regions. The leading region has been Mexico City and its environs, which hold more than 40 per cent of the total, followed by the reindustrializing and ‘maquila’ (in bond plants) states that jointly share another 40 per cent. The oil-producing states, southern states, basic producer states and tourism states jointly hold 20 per cent of the total. It could be argued that the development of the software industry is directly linked to modern corporate organization, which demands eﬃciency. Distrito Federal: The Diﬀerent Faces of Economic Restructuring Mexico City houses the largest and oldest business community in the country. In this area it is possible to ﬁnd the most modern and sophisticated and also some of the most backward sorts of entrepreneurship. So the ‘management revolution’ encompasses diﬀerent stories and processes: of global and subsistence business, of 200-year-old and just created establishments, of the largest educational and research network of the country, and, not least, of the center of power of the country. Almost all government and private operations that take place still have to go through Mexico City, although decentralization has progressively reduced this phenomenon. Combining all the operations carried out in Mexico City, leads to a GDP that is larger than those of Ireland, Portugal or Thailand; it is like a country unto itself. Therefore, simpliﬁcation here could lead to overly rapid conclusions that do not reﬂect what is really happening in the area. This is precisely the case for ITs, as ﬂows of information generated in each sub- region of Mexico City make this business proﬁtable just by attending to the local needs of the area. This explains why more than a quarter of the software industry is located in the Mexico City metropolitan area and why it is domestically rather than export oriented. According to the Census Bureau, the number of IT businesses in the area grew from 236 in 1993 to 703 in 1998. This cluster consists of small, medium and large companies including wholly foreign owned ones such as IBM, Microsoft, Oracle and SAP, and Mexican ones such as Hildebrando, Aspel, Sistemas Dinámicos Internacionales, among others. A prosperous IT business community emerged in the city and most successful developers are today members of the Asociación Mexicana de la Industria de las Tecnologías de la Información (AMITI), which has become a learning community. Each member of the association has a diﬀerent story, but the members have joined together in an eﬀort to address problems jointly and promote the industry. In fact they have been quite successful: in February 2001 they prepared a seminar bringing people from all around the world to show the

Mexico: the management revolution and the software industry

321

beneﬁts of the software industry. The government response was immediate: a program to support industry development was announced in 2002. The IT business community in the city has a long history, but its recent growth has been associated with management revolution in the 1990s, as more and more companies began to demand more software programs to solve their new role in the restructured economy and engineers from the city began to see a business niche amid the economic turbulence.

3.

FIRST MOVERS IN THE IT REVOLUTION

The ﬁrst mover in the IT business in the area was IBM; it built its headquarters in the city center and a production site in Legaria just a few blocks from the business communities of Lomas and Polanco (the wealthiest area of the city). The history of IBM in Mexico began in the 1950s. Some basic manufacturing operations occurred in Mexico, but IBM acted mainly as a distributor of its large mainframe machines, which, as mentioned before, were sold bundled with their software. IT capacity was built in the country through this distribution mechanism, with no involvement of the local community; it was the period of pure transfer of technology (PTT). Doing business in a country like Mexico in the late 1950s and 1960s required a larger involvement for transnational corporations; its sales force required training and there were not that many engineers in the market to hire. So IBM decided to train their sales force, hiring high-school graduates who were beginning their engineering degrees to be trained in a special program. Successful trainees developed abilities, and some of them received extra training in the US through a short residence in the New York area. The in-house training system developed a workforce that was able to install the equipment, customize existing software to the client’s needs, and service the speciﬁc operation afterwards. This integrated model was developed worldwide and was not speciﬁc to Mexico. Software was provided by the hardware supplier, so there was no software development outside of the hardware vendors. The integrated model hindered the potential of software development, but it trained people who would become the seeds of future development. The workforce grew over time; senior trainees were in charge of new trainees, manuals were developed (red books), and a systematic trainee approach was developed inside the corporation. Incentives were given to the trainees as they became senior consultants, allowing them to purchase IBM options at special prices, in an attempt to promote a community spirit. Twenty years later when the company received a judicial order from the US to split its operations, IBM began a restructuring of the whole operation,

322

Agglomeration in the Americas

supporting early retirement, but in such a way that senior professionals could become independent but linked to the IBM operation to support the service of their operations. A typical case is San Martin Associates, whose owner opened a software company after he accepted the early retirement oﬀer. Some of his colleagues supported his operation, and they opened an oﬃce in the high-income area of Polanco, to support Liverpool, one of the largest retailers in Mexico. Following the IBM model they have a small group of engineers who are trained in the company. Most of them come from the National Autonomous University and the Metropolitan Autonomous University, and some are hired from the Technological University of Netzahualcoyotl (one of the poorest areas in the metropolitan area of Mexico City). Business consists mainly of interconnection operations that could be solved through patches to the original programs. Software in this case is a service operation, rather than an innovative operation, but it is well paid: wages are in the range of US$2000–3000 a month for a junior consultant and up to US$4000 a month for a senior consultant. The niche of service for installed software is a large business area, but nowadays it competes with more speciﬁc sorts of development. Economic restructuring in the 1980s led to new needs in diﬀerent areas that could be identiﬁed as follows: management, trade, communications and speciﬁc industries software. IBM’s registration of consultants involved in these sorts of operation brought in a new sort of development, which might be called the IBM network of developers. Opportunity Captured At the beginning of the 1980s, the Mexico City business community faced diﬃcult times: the fall of oil prices, economic disruption, devaluation, heretofore unthinkable foreign exchange controls, bank nationalization and the virtual collapse of economic activity. Those were hard times, but were also the roots of a diﬀerent source of entrepreneurship in the country: no more protection, no more government support. Solutions did not emerge overnight but were created from day-to-day experience and were built from the ground up. The entrepreneurship model changed: formerly, businesses’ main relation was not coordination with other businesspeople but rather a vertical relation with government to get protection or ﬁnancial support. The fall of central government support pushed businesspeople to begin a new dialogue with their equals, other businesses with problems that they were not able to solve by themselves. Nationalization was more than a government decree; it created citizen consciousness about the role of the state in the economy. In terms of management, the business community had the diﬃcult decision of learning how to do business in a diﬀerent way. Learning was crucial,

Mexico: the management revolution and the software industry

323

but it was not an easy task because the ﬁrst step was how to ask the right questions. One thing was clear: in the ﬁrst stage the question was, how to survive? The basic answer for individual entrepreneurs was to reduce costs. It was in this area that IT was fundamental to solving problems; it was required to analyse large ﬂows of information, organize databases and introduce cost accounting. This opened a niche for all of those willing to take advantage of it; developers were there with their recently acquired university degrees ready to get into business; reality also pushed them to open their own ﬁrms as employment was on the low side even for skilled workers. The ﬁrst movers were those ﬁrms designing accounting solutions. Aspel was a pioneer in the area, founded by two systems engineers from the National Autonomous University of Mexico (UNAM) 20 years ago. The company was started almost by accident while the founders were working on a freelance project for another company. The two founders were working at the time at a government oﬃce in charge of distributing housing options to workers (Infonavit). Today Aspel is the leader in accounting solutions for SMEs; it handles demand through ﬁve specialized areas: computer packages, customized software, support and training, and services. The niche market, according to Aspel, is about 400 000 companies, of which the ﬁrm serves about 56 per cent. Their strength in the domestic market allowed them to become a national company and to enter Central and South American markets: Costa Rica, Chile, El Salvador, Guatemala, Honduras, Nicaragua, Panama, the Dominican Republic and Venezuela. Aspel has its software development oﬃces in Merida, and its Aspel Desarrollos oﬃces in Mexico City. It also has about 7000 distributors all over the country and 14 training centers for its new customers. In the 1990s, as Aspel become more competitive, it designed strategic plans and developed strategic alliances with IBM, Microsoft and Teléfonos de México (Telmex). While they are currently in the low end of the market for administrative solutions for SMEs, they are looking to move upward in the market through Aspel Desarrollos. Figure 13.4 shows Aspel’s outlook on the market for its applications. Today, Aspel is a business that looks far ahead. It has begun operations to provide solutions through the Internet. For that purpose it has taken advantage of its alliance with Telmex by making an agreement to provide through application service providers (ASPs) the packaged software named Prodigy Pymes. Another successful ﬁrm that started operations in the 1980s is Hildebrando, the largest software producer in Mexico City. In 14 years, Hildebrando has grown from a three-person project to a 500 member organization. In February 2000, Warner-Lambert presented Hildebrando with its Best Quality Provider award for the ﬁfth consecutive year. Since 1998,

324

Agglomeration in the Americas

High end of the market

SAP JD Edwards QAD Solomon

Aspel market Aspel Desarrollos Package Software

Figure 13.4

Aspel and its market niche

Hildebrando has been featured yearly in Expansion magazine as one of Mexico’s top 500 ﬁrms. Revenues have grown 5.6 times since 1993, reaching US$14 million in 2001, a share of 2.3 per cent of the market. Hildebrando has opened oﬃces in Mexico City, Guadalajara and Monterrey, as well as in Miami, Florida in the US and Madrid, Spain. ITs software factory, located in Mexico City, has the capacity and infrastructure to support 450 developers in a multitude of projects. According to its website, Hildebrando ‘strives to build strong lasting business relationships with clients’, and over 60 per cent of their revenue comes from customers who have done business with them for over ﬁve years. The client portfolio of the company shows how it has been successful in granting support to a variety of businesses in their eﬀort to improve management. Hildebrando has 14 years of experience developing information systems for a variety of industries, which has allowed it to develop not only a great amount of knowledge about the technology available but also an extensive know-how regarding its application within the business environment (Table 13.2). Hildebrando has also been able to develop networks with key corporations like Teléfonos de México (Telmex). Until recently Telmex was the sole telephone provider in the country; now, although it remains the largest telephone corporation, it faces competition from other providers. Its software needs were solved by in-house supply, but this proved to be ineﬃcient, so they decided to look to spin-oﬀs associated

325

Source:

Hildebrando Executive Report (2002).

US customers in Mexico

Retailers Other industries

Telecommunications Government

Banamex Banjército Citibank Inverlat IXE Serﬁn Bancapromex Grupo Financiero Capital Banco de Mexico Bancomer Inbursa Acciones y Valores de México S.A. de C.V. CBI Bolsa Mexicana de Valores Grupo Nacional Provincial Seguros Bital Seguros BBV Probursa Seguros Serﬁn Lincoln Seguros Banamex ATT Radio Centro Avantel Telcel Iusacell Telmex Gobierno de Jalisco Instituto de Seguridad y Servicios Sociales de los Trabajadores del Estado Servicio de Administración Tributaria Gobierno de Jamaica Integradora de Servicios Operativos, S.A. de C.V. DIF Gobierno de Tabasco Secretaría de Hacienda y Crédito Público Gobierno de Nuevo León Secretaría de Gobernación Aurrera Sanborns El Palacio de Hierro Suburbia Vips Liverpool Grupo Warner Lambert México S.A. de C.V. ICA Plastiglas de México S.A. de C.V. KFC Pemex Marzam PISA Citibank Hewlett Packard Oracle Warner Lambert Group Progress Software Wal-Mart

Hildebrando pool of clients

Stockbrokers Insurance companies

Banks

Table 13.2

326

Agglomeration in the Americas Software niches in the New Mexican business environment Communications

Figure 13.5

Zentrum Billing software

Figure 13.6

Foreign trade

Sector applications

Software businesses niches

Sigma Tao Telecommunications software

Blitz Software and web pages development

Intelmex Telmex institute

Teléfonos de México software network

with local companies. Today they have a spin-oﬀ network that provides them with the required software, as shown in Figures 13.5 and 13.6. Sigma Tao is in charge of developing software for their analogical centrals; its main task is RP software under SAP accounting. Blitz software is a joint venture with Hildebrando and is developing intra and extra web pages for the corporation. Intelmex works as a LAN for Telmex; the extranet works for Grupo Carso, Telmex, Triana, Yellow Pages and Historical Center developments. It is organized as a software factory; beginning operations in 2001 with 118 persons and sales of US$1.2 million, it was expected that in 2002 its sales would reach US$8 million and there would be 160 employees. For the year 2004 it was expected that sales would total US$18 million and there would be about 300 employees. Its software focus is on connectivity through the Internet, and it has developed the Prodigy network that supports Telmex users connecting to the Internet. The number of users in this network has increased, totalling nearly 1 100 000 members by the end of 2002, in more than 2395 cities in the country. In the second group we ﬁnd businesses like SDI (Sistemas Dinámicos Internacionales, S.A. de C.V.), which have focused on foreign trade operations. SDI helps export and import businesses in its customs operations. Created by an economist who became aware of the problems in ﬁlling out customs forms, it has faced little competition. Because the software not only ﬁlls out forms but also helps the customer to access the prevailing laws of other countries and to get information about particular problems such

Mexico: the management revolution and the software industry

327

as tariﬀs for a speciﬁc product, it has become very popular among export corporations. Support to businesses is done through the Internet. SDI’s main clients are public agencies (for example, the Secretar’a de Hacienda y Crédito Público or Secretary of the Treasury) and the main export and import businesses of the country. Their main products are Dia (a customs database), Saai. M3 (software to ﬁll out customs forms), and SDINET (www.sdinet.com.mx/), a site for advising clients. In the third group we ﬁnd POLYNE de México, a demand solutions developer which focuses on the textile industry. Alfredo Buzali, its founder, had only a high school education; his parents did not allow him to continue his engineering studies, but expected him to work in his father’s textile factory. At the time, Apple computers was just opening the PC market. Buzali obtained one and began to develop applications to optimize cost operations for the area of the factory for which he was given responsibility by his father. His success allowed him to open his own company in 1981 to address the software demands of the textile industry. Today, the company is the main software developer for the textile industry and employs 25 programmers. Junior programmers earn US$1200 a month, while seniors earn around US$2000. POLYNE’s market includes about 300 clients who buy applications costing between US$12 000 and US$25 000 on average. Earnings in 2001 totaled US$600 000 with the equivalent of 24 software packages; ﬁnancing has come out of earnings, and bank ﬁnancing has helped ﬁnance a mortgage of only 20 per cent of the factory buildings. Piracy has damaged the company, however, and there has not been any protection from public institutions.

4. WITH A LITTLE HELP FROM MY FRIENDS: DIALOGUE AND AGGLOMERATION EXPERIENCE AMITI and some of the software developers in the Distrito Federal have created a dialogue project called ‘Desarrolladores’ which includes a website used to exchange information on how the software market is evolving (www.desarrolladores.com.mx/). The resulting exchange has produced a consortium under the name of Consorcio Uno, whose aim is to support the development of programs on a joint basis to achieve economies of scale. Consorcio Uno hopes to access international software services markets to promote the capabilities and competitiveness of participating companies by expanding their target markets. The main objective is to create a relationship with companies in the US searching for software services such as custom software development, outsourcing of IT services, and business

328

Agglomeration in the Americas

opportunities to develop new markets. The main aim of Consorcio Uno is to take advantage of agglomeration economies; to this end it is also promoting certiﬁcation among its members. In-house Developments: The Government Software Niche The largest demand for software comes from the government, which is estimated to account for about 55 per cent of domestic demand. Software is developed by systems engineers contracted by government agencies and also by a state-owned enterprise, Información y Servicios Tecnológicos (INFOTEC, www.infotec.gob.mx/) It was established in 1975 jointly by Consejo Nacional de Ciencia y Tecnología (CONACYT, www.conacyt. gob.mx/), and Nacional Financiera (NAFIN, www.naﬁn.gob.mx/) to address business needs in information, training, and consultancy. In 1994, a reform was introduced and its main goal became the development of specialized solutions for government agencies and the sale of some of its products directly to the domestic market. Since 1998, INFOTEC has been a self-ﬁnancing government agency; it does not get any budget support from the federal government, but it has to turn over annual proﬁts to the ﬁnance ministry. This budget formula limits its operation. It encourages a cyclical type of operation: at the beginning of the year personnel is at a minimum, growing later as projects begin to operate; average employment throughout the year is about 80 persons, 20 of them system engineers. INFOTEC’s annual income was US$13.5 million in 2001, and, surprisingly, 85 per cent of total sales were to private ﬁrms. INFOTEC’s software development focuses on generic types of technologies, analysing software developments in the rest of the world, to produce a generic solution for the Mexican market. Among the areas studied are: collaboration systems, learning systems, management systems and containment systems. INFOTEC also develops special solutions for all government institutions. INFOTEC has developed four areas of business: RTN provides technology for the transmission and exchange of information through satellite and ﬁber optic connecting institutions through the extranet; CTA develops websites for government, similar to Blitz for Telmex; SIE provides system consulting teams; and ATO is a training center similar to that of Hildebrando. Most successful software developments have been in management, marketing and solutions development. Among them are: 1. 2.

SISTEC Connects technological centers of CONACYT with private sector demand. COMPRA-NET Software that enables the federal government to manage government purchases through the Secretaría de la Contraloría.

Mexico: the management revolution and the software industry

3.

4.

329

INTERNET 2 A software project to create information superhighways to enable data transmission in higher education institutions and research institutions. Recently INFOTEC has developed a learning system for the education ministry to support learning at elementary school which provides a remedial program that allows children to solve educational lags.

Outlook for the Mexico City IT Community The Mexico City software community has allowed business to become more knowledge intensive. The main applications are customer relations management (CRM), enterprise resource planning (ERP) and accounting solutions, which have allowed businesses to improve their management capacity. SMEs usually get accounting solutions, whereas CRM and ERP are purchased by medium and large businesses. The newer types of provider relationship management (PRM) and employee relationship management (ERM) are still quite speciﬁc and of reduced utility for Mexican business (see Figure 13.7). Local government is trying to provide the required infrastructure to allow further development of the industry. In 2000 they began the construction of a ‘techno pole’ in the metropolitan area with the purpose of promoting business integration. The old rastro (slaughterhouse) of Ferreria has been transformed into a new area of high-tech development. The model chosen assumes that a higher education institution, that is, the Instituto Tecnológico de Monterrey, and a large software corporation, that is, IBM or Microsoft, will induce developers to house their project inside the techno pole.

Figure 13.7 Management revolution and software industry development in Mexico City

330

Agglomeration in the Americas

The Monterrey Software Cluster The Monterrey software cluster emerged as a response to the needs of business conglomerates (Table 13.3). Businesses have historically agglomerated in Monterrey and the surrounding state of Nuevo Leon; large national conglomerates have located their headquarters in the area for years, for instance: Alfa, a diversiﬁed manufacturing conglomerate, Hylsa, a large steel producer; FEMSA/Cervecería Moctezuma, Mexico’s largest beverage company and one of the leading beverage companies in Latin America, exporting its products to the United States, Canada, and select countries in Latin America, Europe and Asia; and CEMEX, the largest cement producer in Mexico, with plants all over the world. Nuevo León has 101 000 businesses in total, and only 256 IT businesses, with a software cluster comprising about 76 companies. Softtek is the software star of Mexico. With about 2000 employees it is by far the largest corporation in this area; its next largest competitor, Hildebrando from Mexico City, has only 500 employees. Softtek was born amid the 1980s crisis, established in 1982 by Gerardo Lopez, one of the engineers of the group working with Dinámica, along with four other partners, all of them trained in system engineering at the local Instituto Tecnológico de Monterrey. At the outset the company was supported by General Electric (GE), which at that time was looking for a partner in Mexico, trying to take advantage of highly trained, low-cost labor (the devaluation of 1982, which took the peso’s value from 26 to 150 pesos per dollar, reduced the cost to one-ﬁfth what it would be in US dollars). GE hired Softtek to write customized software; this helped the company to gain expertise not only through direct orders but also through training in the US. This enabled the company to grow, allowing it to become a ‘software factory’. Expansion was made possible through internal resources as well as through the use of stock options. These good old days came to an end in the 1990s, however, when GE decided that the Mexican market had enough potential that they should take advantage of it directly instead of through a partner. GE therefore established its own company in Mexico under the name of Ddemesis, a subsidiary that was already working in Bangalore. This pushed Softtek to look for local clients. At the time, privatization of banks was taking place, and the banks needed software developers to help them deploy CRM and ERP systems. Some industries at the time were privatized, among them Teléfonos de México. Since the large supplier of software at the time was Softtek, they decided to create a joint venture under the name of Sigma Tao. In the 1990s demand was also increased by NAFTA, allowing Softtek to take advantage of the service integration with the US, by developing

Mexico: the management revolution and the software industry

331

what can be called ‘nearshore’ services: oﬀshore services that are done in the same time zone and at a short distance from the US. Roberto Montelongo, vice president of the company, argues that Softtek has the capacity to develop software when the problem is not deﬁned or changes often, and in that sense it has a competitive advantage over companies in other parts of the world, such as India because of proximity to the US. Mexico has other advantages over India as well. These include NAFTA, whereby Mexican employees can be awarded a certain type of visa (TN) unavailable to non-NAFTA partners; the consequences of 9/11, which mean that stricter immigration measures are applied to people from the Middle East and South Asia; and the Indian/Pakistani conﬂict, which has created some risks for American ﬁrms that have connections with Indian companies. However, Mexico also has some disadvantages compared to India. Labor costs are three times cheaper in India than in Mexico; the skilled labor force is large in absolute – if not relative – size; and English is widely spoken. Furthermore, the common image of Mexican employees is one of low skill levels. To exploit the advantages that Mexico has over India, and to start targeting the US market in nearshore services, Montelongo believes that Indian companies will start establishing joint ventures with Mexican companies. Currently, Softtek has a development center in Monterrey, Nuevo Leon, where the trainees are still twice as cheap as in the US. This allows Softtek to compete in cost with foreign competitors. Most of Soﬀtek’s software developments are customized software (70 per cent of total sales), and the company also does consulting and packaging (30 per cent of total sales). Sales revenues for Softtek in 2001 were US$73 million. These revenues are distributed according to the following geographical location: US, 35 per cent; Mexico, 30 per cent; South America, 30 per cent; and Spain, 5 per cent. In the area of quality assessment, Softtek is certiﬁed as CMM-3 under the ‘capability maturity model’ developed at Carnegie-Mellon University.2 Its need to become standardized with quality practices forced changes on the organization and its software development processes. Softtek spends about 2.5 per cent of its revenues on formal training programs for its employees. In addition, it requires employees to spend one week per year in these types of courses. However, employees receive most of the training on a daily basis in the form of on-the-job training using mentors. Furthermore, Softtek promotes self-study for employees, encouraging them to take courses. As for English classes, Softtek requires daily classes of an hour for its employees. Montelongo believes that its employees are classiﬁed in English skills as follows: ﬂuent, 66 per cent; regular (good written skills), 30 per cent; trainees, 5 per cent. Among Softtek’s major

332

1 ABC Systemas 2 Ad Inﬁnitum, S.A. De C.V. 3 Administración Automatizada, S.A de C.V. (Assasoft) 4 Asesoría en Sistemas, S.A. De C.V. 5 Atecsys, S.A. De C.V. 6 Auto Capacitación, S.A. De C.V. 7 Axsis Tecnología, S.A. De C.V. 8 Barsoft, S.A. De C.V. 9 Caaspre Technologies de México 10 Capta 11 Castelec Internacional, S.A. De C.V. 12 CDSI México 13 Cevine Atlantico 14 Coactor, S.A. De C.V. 15 Computación Dinámica, S.A. De C.V. 16 Comutación en Accion, S.A. De C.V. 17 Consiss 18 Consultores en Informática, S.A. De C.V. 19 Consultores Integrados en Outsourcing, S.C. 20 Consultoría Dinámica Integral, S.A. De C.V. 21 Consultoría en Actualización de Ingeniería e Informática

Table 13.3 Monterrey software cluster

39 40 41 42 43 44 45 46 47 48 49

25 26 27 28 29 30 31 32 33 34 35 36 37 38

Das Sistemas, S.A. De C.V. Disher Technology ERP Consultores, S.A. De C.V. e-Strategia Consulting Group Expersis, S.A. De C.V. Expertec, S.A.de C.V. Getronics CP, S.A. De C.V. Grupo Antar Sistemas, S.A de C.V. Grupo IQ, S.A. De C.V. Grupo Qualita, S.A de C.V. Grupo SCANDA, S.A. De C.V. Host C.P.A.P., S.A. De C.V. Ingeniería Creativa, S.A de C.V. Integración de Soluciones Arkatos, S.A. De C.V. Internacional de Sistemas Intosh, S.A. De C.V. Inweb Internacional ITGS, S.A. De C.V. Kernel Kinae Software Kontenix Leo Computación Lozano System, S.A. De C.V. Megasistemas, S.A. De C.V. Micro Sistemas Gerenciales, S.A. 70 71 72

65 66 67 68 69

63 64

62

60 61

59

54 55 56 57 58

Promissus Questor, S.A. De C.V. Saul Cantu y Asociados, AC Servicum Sistemas Computacionales Integrales, S.A. De C.V. Sistemas y Servicios Integrales en Comercio Exterior, S.A. De C.V. (SSINCE) Softek Solución Administrátiva Total, S.A. De C.V. Solucionees Computacionales Integradas, S.A. De C.V. Soluciones Integrales en Sistemas Soluciones Profesionales en Internet Solution Ware, S.A de C.V. Soporte Externo Especializado Spirit Ingeniería, S.A. De C.V. Tarik Sistemas Técnica Computacional Avanzada, S.A. De C.V. Totaltech VESA Vital Sistemas, S.A. De C.V.

333

22 Consultoría y Asesoría en Informática, S.A. De C.V. (CAISA) 23 Cosmos Consultores, S.A. De C.V. CSI (Centro de Servicios en Informática – Fac. de Ciencias Físico 24 Mátematicas (UANL)

De C.V. (MIGESA) 50 Micros Personales, S.A. De C.V. 51 Neoris 52 Open Service, S.A. De C.V. Profesionistas Asociados en 53 Comutación, S.A. De C.V.

73 74 75 76

Wexa Working, S.A. De C.V. You-Man Talent México Zeta & Soft

334

Agglomeration in the Americas

advantages are linkages with the educational sector, through what it calls ‘internship’ programs for senior students from Monterrey Tech (ITESM) and the University of Monterrey (UDEM), who want to do their thesis work with the company and who can later work with the company when they graduate. The success of Softtek has opened the doors to business in the area, through what we might call spin-oﬀ of large corporations. In 1993, CEMEX decided to create Cemtec to manage new technology for the group. Two years later it opened its doors to outside businesses such as Dole Fresh Fruit; in 1999 it was certiﬁed as an ISO 9001 company. In the year 2000, Cemtec created strategic alliances with international ﬁrms to provide solutions for electronic businesses (e-business). In 2001, Cemtec became integrated into Neoris Corporation, which belongs to the holding company CxNetworks (www.cxnetworks.com/), which is in charge of supplying e-business solutions to the CEMEX group and also to outside demand. Its sales are about US$100 million a year, and its major niche is in CRM software for the company’s large worldwide client list. In the 1990s a larger number of software developers came into the market. One of the success stories is Kernel, which today has a group of ﬁrms addressing software demand: Kernel Corporativo, S.A. de C.V.; Consultores en Servicios Integrados en Tecnolog’a, S.C.; Kernel Enlaces Digitales, S.A. de C.V.; Kernel Conectividad Productiva, S.A. de C.V.; and E-ntelligence Consulting, S.A. de C.V. It has specialized in attending to manufacturing businesses, spending 1.12 million man-hours in development for the sector; in addition, it has given 0.81 million man-hours to distribution ﬁrms, and 0.58 million man-hours to ﬁnancial businesses. Its products focus on the management of human resources, accounting and sales, and on production control. Besides the large software businesses, there have been small developers willing to take advantage of SME demand for software. Among them is Axsis Tecnologia, whose owner is Carlos Reyes, who graduated from the Autonomous University of Nuevo Leon (UANL) in Computer Systems, and later received an MBA from ITESM. The story is the same as in the large software companies: the owner’s previous experience is in one of the corporate businesses of the Monterrey group, in this case FEMSA. In 1998, Reyes started Axsis Tecnologia, which had 25 employees and revenues of about US$500 000 in 2001. In 2005, Reyes’s objective is to have US$8 million in revenues, doubling revenues each year. Axsis has an interesting structure. It started as a hardware distributor (Prosel), and this has provided the cash ﬂow for the software company and has allowed them to establish GIT (a technological research group), which focuses on creating new knowledge and knowledge management. Outsourcing is the main source of income

Mexico: the management revolution and the software industry

335

(about 60 per cent of the total); the companies for which Axsis works are the ﬁrst-stage companies mentioned above: Softtek, Neoris (CEMEX), Open Service, Internacional de Sistemas and Kernel. Beside outsourcing, Axsis has been able to develop packaged software: Fiestasis, Medisis, Escolarsis, Integrasis (an ERP program) and Prospectasis. The most important product is Integrasis, for which it has about 25 clients. The cost of this product ranges from US$500–5000. Apart from the products, Axsis also oﬀers services, including customization, support and upgrades. From these services, Reyes hopes to form commercial associations with his clients, and hopes to create successful trademarks to reduce his dependency on outsourcing. The market that Axsis is targeting is small to medium-sized companies (‘PyMES’) from Mexico, with their most important client being FEMSA. Another interesting area of business in Monterrey is the case of Cygnus Tech, supported by the mother company Grupo Asercom. What is interesting in this case is that Cygnus is attending to the ﬁnancial sector business in Monterrey, Banorte, the only Mexican-owned bank in the country. Cygnus Tech had revenues of US$1.5 million in 2000, and of US$500 000 in 2001. However, the crisis of 2001–2002 is undermining its market and therefore it estimates that the company will go out of business in a year if earnings do not improve. New businesses in the Monterrey area are focused on promoting quality assessments for software companies, as is the case of All Soft. Its business strategy is to do quality assessments (CMM) for companies in the software industry. To do the assessments, they have alliances with a Canadian, who is qualiﬁed to do CMM assessments for companies, and a Chilean, who does consulting. All Soft’s potential clients are companies in both the software industry (software development) such as Softtek and Antar, and companies in other industries (such as manufacturing and ﬁnance) that need to certify their IT practices. The companies that All Soft is planning to target are medium to big companies that need to standardize their software development processes. The software industry has helped the Monterrey Group to maintain its competitiveness and has opened new areas for businesses, as in the case of Softtek and its links to demand in the US. Corporate businesses in the area have established their own software businesses to control their client expansion abroad, as in the case of CEMEX, and the largest group of developers has focused on the domestic market (Figure 13.8). The Jalisco Software Cluster The Jalisco case diﬀers from the other cases. Its software industry developed as a byproduct of the electronics cluster, attracted by large-scale international

336

Agglomeration in the Americas

Figure 13.8 Management revolution and software industry development in Monterrey electronics investment (with IBM and Hewlett Packard in the lead). IBM was the main driving force; 30 years ago the corporation began to develop applications. In 1982 its General Director, Francisco Alba, pushed for a change in its output mix, getting away from electric machines toward electronic manufactures. When this decision was accepted, the accompanying requirement was managerial changes that could take control over the new operations. This demanded a managerial revolution, opening the doors to software requirements in the area of managerial control. But it took 10 years to set up an in-house team to develop software; in 1992, IBM decided to send a Guadalajara group to Rochester, where they would be trained to support software development for the operating system for the AS400 computer. It was in 1997, in the midst of the manufacturing expansion, that software diversiﬁcation took place, developing software for PCs, and tools for log and OS2 development. In the same line there were software developments to connect AS400 to make it compatible. In 1999 the setting of IBM Global Services (IGS), connected them to the system where IBM Guadalajara became the last level of support (that is to say, the one that holds the code). Employment increased from 125 in 1997 to more than 500 in 2001, with revenues of US$160 million, about 25 per cent of market share. The main feature of this operation is that it is export oriented to assist the AS400 operation worldwide. Many of the engineers trained at IBM moved out to set up their own businesses, becoming small developers of solutions for the business community in Guadalajara. Hewlett Packard has also kept sophisticated production – high-end oﬃce printers – in Mexico. It even provides a glimmer of a role for Mexicans beyond manufacturing. The company’s local facility has worldwide responsibility for the design of the paper-handling parts of the company’s laser

Mexico: the management revolution and the software industry

337

printers as well as for the software that allows users to choose between paper drawers and to dictate how the output is collated and stapled. Some of this work is being outsourced to a gaggle of young, baseball-capped Mexican engineers, male and female alike, whose consulting ﬁrm, ASCI, has festooned a downtown Guadalajara mansion with cabling and computers. ASCI was started in 1993 by Sergio Fernandez, 42, who gained his PhD in engineering from the City College of New York. Today it has 44 employees, 27 of them engineers, including some who spend so much time at Hewlett Packard’s laser printer headquarters in Boise that Fernandez leased an apartment for them. With a few years’ experience, ASCI’s engineers earn about as much as they would in the US, at least outside the Bay Area. Fernandez has become a popular software developer in the US as well; he has a company in Idaho where he gets contracts to be developed in his Guadalajara plant. Among ASCI clients in Guadalajara is the city’s largest newspaper, El Informador, for which Fernandez has developed software to develop a photography database and a security system. The Guadalajara business community is known for the predominance of micro and small businesses. In this environment, software industry development was focused on solutions for this segment of industry, mainly packaged software. Computacion en Accion (Compac) is the leader in accounting packaged software in western Mexico. It was founded in 1984 by a systems engineer, Rene Mart’n Torres Fragoso, to attend to the accounting needs of SMEs. Its main product is ContaPAQ, which has an interface with other systems (one program can control up to 999 businesses), with perfect integration with Microsoft Oﬃce; for that purpose they have a strategic alliance with Microsoft. It has become quite successful, and today has about 180 000 systems installed in Mexico and Latin America. Compac has become a national software business, second only to DF’s Aspel. It has branch oﬃces in the Distrito Federal, Monterrey, Chihuahua, Tijuana, Ciudad Juarez and Merida. Distribution reaches Central and South America; it has 70 service licenses and 1500 distributors. Its software has its own database manager, with a reporting language; the attraction of the software is that accounting operations could be done on line and with open periods. This makes the solution very attractive, as it allows ﬂexibility to increase the size of activities. Compac’s products integrate management solutions for SMEs: basic accounting, wage billing, inventory control, client management, treasury control and billing operations. Its products are well known and have won diﬀerent awards: Premio Giroscopio of Conacyt (1992), Best Accounting Software Prize of Personal Computing (1993), Preferred Supplier by Infochannel (1998), Number One Accounting Program by Personal Computing Magazine (1999), and Galardón Jalisco

338

Agglomeration in the Americas

a la Exportacion (Jalisco Export Award) of medium-sized enterprises (1999). It has sales of about US$60 million (10 per cent of the market). Compac has been successful in getting into the Internet era through Microsoft bCentral, where clients could ﬁnd its products as solutions for SMEs. The Jalisco government has set up an institutional environment to support industry development: 1. 2. 3.

It has coordinated the design of a long-term strategy, deﬁning Jalisco software niches (Figure 13.9). It will create a techno pole in Zapopan where it will house programmers. Institutional support will also come from the Institute for Information Technology of Jalisco (IJALTI) to promote certiﬁcation of software producers at level CMM-3.

One of the main features in the Jalisco experience is the eﬀort of coordination of the government to set rules for the development of the software industry that marks a diﬀerence from Mexico City and Monterrey. A push for development came from the determination of the Jalisco government to promote the creation of an electronic cluster, which embedded the software project. An Asian Tiger in Mexico: The Aguascalientes Software Cluster Traditionally and up to the late 1970s, Aguascalientes’s economy had relied primarily on agriculture, complemented by some production of wine and garments. This situation has changed radically since the early 1980s, a period of fast industrialization in which Aguascalientes has experienced rates of growth in both manufacturing and exports which far exceeded the national average. This growth has largely been fueled by considerable amounts of foreign direct investment (FDI), particularly from the Japanese automobile and US electronics industries. How did this extraordinary growth come about? In 1974, the incoming state governor decided to pursue a radically diﬀerent development strategy during his tenure (1974–80) and to shift emphasis from agriculture to manufacturing. Given its limited experience in industrial promotion, the government’s ﬁrst action was to determine the main needs of industry. This was done by interviewing the owners of what were then the strong productive sectors: agriculture, winemaking, textiles, garments, railroad repairs and some metalworking. Similarly, it contacted unions to understand the main concerns of labor. The results of its initial

339

Figure 13.9

Development tools

Packed applications

Jalisco potential software niches

Traditional Computer services

e commerce Internet and Information Suppliers

Packaged Software

Software for System Infrastructure

Internet tools

Tools for life cycle

CODE

Data base

Software for consumption

Business software

Software for security

Jalisco niches

Video games

House productivity

Edutainment

Horizontal applications

Vertical applications

Netware

Utilities

Systems and Network Management Products Middleware

Operative Systems

System software

340

Agglomeration in the Americas

survey were not encouraging, for it realized that the state lacked important conditions to attract investment, particularly basic infrastructure. In 1973, Nacional Financiera (NAFIN, the federal industrial development bank) decided to support the development of medium-sized cities in 23 states as part of the national decentralization program. This program included the creation of industrial parks which provided physical infrastructure, business development services, plus a very wide array of support mechanisms such as ﬁscal incentives and project evaluation assistance. The state government took advantage of the program by creating a trust for the Industrial Park of Aguascalientes, and donated 200 hectares, 40 of which were urbanized with the support of NAFIN. The eﬀort to strengthen the necessary infrastructure and services soon resulted in new investment for the state and a broadening of the local manufacturing base. Among the key ﬁrms established in the latter part of the 1970s were three national metalworking ﬁrms and the ﬁrst automotive components ﬁrms. Since the ﬁrst industrial park was so successful in attracting industries, the trust funds have been used to build three more parks from the seed capital of the original industrial park. These years of industrial development can be considered extremely successful in at least three ways. First, the creation of important parts of the necessary infrastructure prepared the basis for incoming investment. Second, the creation of important networking institutions, such as chambers, where managers gather to exchange points of view to facilitate problem-solving and dispute resolution. Third, the image of the state changed as Aguascalientes was no longer perceived by the rest of the country as an agricultural state but started building its reputation as an industrial state. With this critical mass of suppliers and buyers present, the state became attractive for ﬁrms searching for a suitable plant location. The ﬁrst international project was Texas Instruments (TI) which started exploring options for a manufacturing plant in 1979. The government saw having the plant in the state as a priority, and worked to address the legitimate concerns of TI. For instance, a waiver was obtained on the limitations on foreign ownership (which were typical during the import-substituting industrialization era), eliminating the need to get direct approval from the president. Other key aspects in attracting a ﬁrst multinational to Aguascalientes were the stability of the local labor force, the high quality of workers, ability to perform very detailed work (resulting from the embroidery tradition of the past), and the high participation rates of women in the labor force, which is particularly important for the electronics industry. Once TI decided to settle in Aguascalientes, Xerox and Nissan soon followed. The three major multinational ﬁrms have all played an important role in the local economy by giving their employees access to a global

Mexico: the management revolution and the software industry

341

knowledge network, plus contributing with the technology which is embedded in advanced manufacturing processes. In terms of R&D capabilities, all of them develop their product and process technology in their home countries, where they tend to centralize the R&D function (though Xerox has developed some product lines locally). Even though both the government and ﬁrms have actively pursued vendor development programs, the amount of inputs provided by local ﬁrms is still quite small. This is the current challenge. Before the transformation had begun, Aguascalientes experienced a vicious cycle of low expectations, low demand for institutional change, and thus low investment and poor outcomes. Concerted private–public eﬀorts resulted in a virtuous spiral of better investment environments, institutional reforms and enhanced private dynamism. Among the key institutional features of this transformation which were responsible for the unleashing of private dynamism almost everywhere are: a focus on the improvement of the business environment, private–public partnerships to solve speciﬁc business problems, eﬀective local economic development agencies, and the presence of an eﬀective public sector entrepreneur (an extremely dynamic individual with the ability to listen to the private sector and get things done under adverse circumstances) (Figure 13.10). Foreign investment has led also to cluster formation, as transnational ﬁrms have brought with them ﬁrst-level suppliers that have begun to subcontract with some small suppliers in the area. In the software area, large companies began to develop embedded software for their equipment, as is the case of TI, Xerox and Nissan. All of them contracted engineers from the area to adapt the software for some speciﬁc uses in Mexico; this led to a learning process in the 1980s, and later in the 1990s as many of the system engineers decided to begin their own businesses. It was in the 1990s also that the government decided to promote cluster formation in diﬀerent areas: textiles and garments; electronics; furniture; the auto industry; and foreign trade. What is interesting in the experience is that they have been able to develop a methodology that consists of the formation of a civil association that will be joined by all ﬁrms that will form the cluster; in addition, the government creates an institution that will support from a technological point of view, the operation of the cluster. Besides that, they generate a manual of practices that has to be followed by the members of the cluster. This methodology was developed in the 1990s and at the beginning of the 2000s; when they took the decision to develop the IT cluster, they followed the same procedure. In this case the cluster is called ‘Innovatia’, with 24 ﬁrms that on average have ﬁve employees. Following the former experience they have decided to build a techno pole to house all ﬁrms in the cluster. Resources for this purpose will come from federal

342

Agglomeration in the Americas

Industrialization

Primary Economy Agriculture Cattle and Commerce 1970

Secondary Economy

Technology and Knowledge revolution

Third-Level Economy Clusters of products and services of high value added

Manufacturing Value Added

1975

1980

1985

1990

1995

2000

2010

2020

Development of human capital

Figure 13.10

Evolution of the Aguascalientes economy

Sector

Cluster

Technological support

Textile and garment

COCITEVA

CTV

Furniture

CONIMUEBLE

CETIMA

Auto industry

FOMAUTO

CIATEQ

Electronics

CELESA

IT

INNOVATIA

Commerce and services

CODECO

Exporting

OPEXA

Figure 13.11

CEDITI

PYMEXPORTA

Aguascalientes clusters

and state government, and part of it will also house a higher education institution to support specialization in high-tech industries; the institution in this case will be ITESM. Figure 13.11 shows all clusters developed in the state.

Mexico: the management revolution and the software industry

343

The specialization of the software cluster will be production management, general management, and support for educational institutions. In the catalog that they have prepared for the cluster presentation, 13 ﬁrms are included, as shown in Table 13.4 with the specialization speciﬁed. The main feature of the cluster is that instead of certifying individual ﬁrms in terms of quality they are working toward the certiﬁcation of the cluster. It could be argued that it is a new face of how to approach the problem of developing the market. How Transnational Supply Will Play with Local Developers Transnational corporations began the supply of IT in Mexico. Now that the market has developed, it is more attractive for them. IBM was the ﬁrst mover, which began the whole process, selling its integrated packages. Microsoft came in with the appearance of the PC and its main approach was packaged software. In the 1980s and 1990s, other ﬁrms appeared in the Mexican market, trying to take advantage of growing demand, among them SAP, Oracle and Sun Microsystems. In their eyes, the market was a tiny one but large enough to proﬁt from, with the expectation that it will grow. Management Approach of Large Suppliers The ﬁrst approach of transnationals was to sell their solutions, adapted to Mexico. In an interview with the German corporation SAP, it argued that its market niche is corporations with sales above US$300 million, and no less than US$5 million. There are about 300 Mexico clients,3 25 per cent of them with sales above US$70 million. Among the most powerful clients are the state and federal governments and corporations, that is, the Guanajuato state government; the Oﬃce of the President (which has requested SAP to create a command management system for it); and Elektra, Corbi and Gigante in the corporate area. Sales in Mexico are of about US$100 million, just a small share of global sales of around US$7.5 billion (Figure 13.12). The other large transnational corporation that supported the management revolution in Mexico was Oracle; its ﬁrst Mexican client was Pemex, supporting the beginning of the elaboration of a large database. It was not until the 1980s that Larry Ellison (one of the owners of the ﬁrm) began to develop ERP, which was immediately in demand by large Mexican enterprises. The sequence of operation database to enterprise resource planning software (ERP) was followed by consulting on line. Like the large oil company, Pemex, it has developed the medium-sized enterprise market, pushing the use of ERP, and with ASP technology it has promoted cost reduction in businesses with diﬀerent locations. Today income for Oracle comes from: ERP (70 per

344

DA Comp. S.C. Brindamos soluciones integrales ASAP, soluciones a la medida con gestión segura de información Montecristo Sistemas, soluciones informáticas integrales SCH, S.A. de C.V. Simpliﬁcando la Vida en los negocios Ingeniería Aplicada Educación, Ingeniería, Consultoría e informática GGS Internet cerca de ti ACTION SN Tecnología Decidida a ayudarte Consultoría en Desarrollo Humano. Sistema para Evaluar la Calidad del Hombre en el Nuevo Milenio

Internet solutions Opthamology consultory management Human resources evaluation

Banking accounts control; budget control Improves client management and ﬂoor control Library automation, educational software, and Internet control

Sale points and manufacturing control for the textile industry Software specialized in funds controls and dispersion for ﬁnancial institution Software for industry and commercial management; corporate Intranet and e-commerce Management and control of ﬁnancial institutions Productivity software

Real estate management; eﬀective cost estimation for industries, commerce, government and medium-sized ﬁnancial institutions; inventory management; sales points; purchases, sales payments control; management; administrative modeling; ﬁxed assets; accounting; treasury; accounting; school control; management Integrated solutions

Asesoría Administrativa y Sistemas (AAS)

Soluciones Integrales Integradas (SIA) Soluciones de Negocios para el Nuevo Milenio, en la Era Digital SIMATEX, Tecnología Creativa para tu Negocio Deloitte & Touche, ‘Contribuir a la excelencia de nuestros clientes y de nuestra gente’ Grupo INCO soluciones totales de negocio

Solutions

Aguascalientes: ﬁrms of the software cluster included in the catalog

Name of ﬁrm

Table 13.4

345

Figure 13.12

Large

Large

5% of SMEs could get SAP products

95% of medium enterprises (< US$30 m)

SAP market niche in Mexico

Small and medium business

MySAP.co m software

Industrial Sector Retail industry, consumption and manufacturing industry

Large

Large

Services

+300

+20

+70

+5

Retailers • Softtek • HP • IBM

SAP direct sales

346

Agglomeration in the Americas

cent); CRM (15 per cent) and surplus change (15 per cent). But Oracle has also developed a new market niche: developing software for state and local government introducing GRP. Today it has introduced it in the state of Mexico and in Ciudad Juarez, Chihuahua; in this area they are developing a GRP (Graphic Related Programming) for small municipalities with ASP technology. The ﬁrst governments to take advantage of that new process will be the states of Queretaro and Veracruz. Another market niche that Oracle has developed has been the educational one. In collaboration with Sun Microsystems, they have developed software called Oracle Classroom, the goal of which is to train students with tools that are already in the market. Some higher education institutions are partners of Oracle, that is, Instituto Tecnológico de Estudios Superiores de Monterrey (ITESM), Universidad Nacional Autonoma de México (UNAM), Universidad del Valle de México (UVM), Universidad Tecnológica Fidel Velazquez de Netzahualcoyotl, Universidad Tecnológica de México (UNITEC), and Universidad Autónoma de Veracruz (UAV).

5. THE NEXT STEP: THE PLATFORM FOR INTERNET OPERATIONS Today businesses are evolving to the Internet. Small and large businesses are looking for the possibility of having a web presence; solutions will be sold through the net, so the key element will be who will provide the service and what sort of platform will be used. Microsoft has begun a large operation in Mexico for that purpose; it has founded NET México, with the aim of promoting the development of the industry as shown in Figure 13.13, compatible with the government Software Development Program. Some businesses and local governments are looking for alternatives and are discussing the use of a free software approach. In any case the debate will intensify as the Internet becomes a key factor for business to develop. Mexican developers are looking for alternatives; some of them are using its capacities abroad. One example is Miguel Icaza, who started a business in Boston that he hopes could become the alternative to Microsoft’s business network. The management revolution will continue, but it is not clear how diﬀerent actors will participate, and who will take the lead in the market. One question that must be addressed is whether the market will be domestic or exported oriented. What is certain is that the Mexican market will continue to grow. Therefore, diﬀerent scenarios could be envisaged: ●

Scenario 1 The historical trend will dominate and the market will grow at a rate of 9 per cent; in this case, the market will go up from US$600 million to US$1.6 billion a year.

347

Last mile service providers

Customer system area

ASPs

Individual developers

Academic

S U P P O R T

C O M M E R C I A L

OTROS (OTHERS)

ASPs

OEMs

VARs

Retail

Systems integrators

CADs

Individual developers

S U P P O R T

C O M M E R C I A L

Figure 13.13

Mexican software industry plan

HABILITADORES (Institution which connects MNCs with government, development and commercial banks, and business associations)

Industrial developers

Academic

Systems integrators

Full ISVs

Customer system area

Domestic market

Foreign clients Exports

Sales: US$5 billion 106,000 jobs 170 exporting firms

Goal 2010

International demand

348 ●

●

Agglomeration in the Americas

Scenario 2 Export demand increases, as stated in the government’s Software Development Program, and the domestic market grows at a historical rate. In this case the market reaches US$6.6 billion per year. Scenario 3 Export demand increases, as stated in the government’s Software Development Program, and the domestic market grows at a historical rate. In this case the market reaches US$6.6 billion per year.

In any case the market growth opens a large expectation for further investment in the industry. The management revolution in Mexico is still in its initial phases, so if growth is resumed in 2003, the domestic market could register a larger rate of growth. If that is the case, the domestic market could help software producers to get competitiveness through domestic economies of scale.

NOTES * This chapter is part of a joint project with Professor Andrew Schrank (University of New Mexico) and Michael Piore (Massachusetts Institute of Technology). Visits to software businesses in Mexico and many hours of discussion with them helped to shape the ideas, and their reading of various drafts helped to improve the ﬁnal version. Analysis in this project has been possible thanks to a large group of software industry entrepreneurs who spent time providing and explaining information to build up the case studies. For particularly valuable assistance, I would like to thank Carlos Maroto, a true visionary who was at that time Director of the Asociación Mexicana de la Industria de las Tecnologías de Información (AMITI); Ricardo Zermeño, a stalwart defender of the national IT industry and a pioneer in tracking its evolution; Eugenio Godard and Braulio Laveaga, President and Director of the Camara Nacional de la Industria Electrónica y de las Tecnologías de la Información (Canieti Guadalajara); Luis Fernando Flores of the Innovatia Cluster in Aguascalientes; and Jorge Zavala, an enthusiastic software entrepreneur who is convinced of the need to bring economies of agglomeration to the industry. This chapter could not have been written without the research assistance of Bernardo Alanis and Daniel Moreno, who helped to create a large database on IT. 1. Of the 55 countries classiﬁed by Digital Planet in their 2002 Annual Report (WITSA 2002). 2. See the Software Engineering Institute’s website at Carnegie-Mellon University for a brief introduction to CMM: www.sei.cmu.edu/cmm/cmm.html. 3. SAP has about 13 000 aﬃliated businesses, so the Mexican operation is only 2 per cent of global operations.

REFERENCES Alarcón, Rafael (1999), ‘Recruitment process among foreign born engineers and scientists in Silicon Valley’, American Behavioral Scientist, 42, 1381–97. AMITI (2002), ‘Propuesta para el uso y aprovechamiento de las tecnologías de la informacióny las comunicaciones’, mimeo, Mexico City. Asociación Nacional de Universidades e Instituciones de Educación (ANIES) (1995a, 2001), Reporte Annual, Mexico.

Mexico: the management revolution and the software industry

349

ASPEL (2002), Acerca de ASPEL de México, S.A. de C.V, Mexico. Balasubramanyam, V.N. and A. Balasubramanyam (2000), ‘The software cluster in Bangalore’, in J.H. Dunning (ed.), Regions, Globalization and the KnowledgeBased Economy, Oxford: Oxford University Press. Brooks Jr, Frederick P. (1982), The Mythical Man-month Essays on Software Engineering, Reading, MA: Addison-Wesley. Carnegie-Mellon University (2002), Capability Maturity Model for Software, Pittsburgh, USA. Consejo de Ciencia y Tecnología de Jalisco (2002), ‘Programa Especial de Ciencia y Tecnología 2001–2006’, www.conacyt.mx/pecyt/. Consorcio UNO (2002), ‘The Mexican Software Services Industry’, PowerPoint presentation institucional Mexico, November. Cordoba Technology (2002), ‘Argentina I.T. Cluster’, Cordoba, Argentina. Council of Economic Advisers (2001), ‘Economic Report of the President. Innovations in Information Technology’, Washington, DC, pp. 34–55. Dunning, John (2000), Regions, Globalization, and the Knowledge-Based Economy, Oxford: Oxford University Press. Eichen, Kyle (2002a), ‘Social impact of informational production: software development as an informational practice’, Center for Global International and Regional Studies, University of California Santa Cruz WP #2002-1. Eichen, Kyle (2002b), ‘Mapping the micro foundations of informational development: linking software processes, products and industries for global trends’, Center for Global International and Regional Studies, University of California Santa Cruz WP #2002-2. Eichen, Kyle (2002c), ‘Software development: a view from the outside’, Center for Global International and Regional Studies, University of California Santa Cruz WP #2002-3. Free Software Foundation (2002), ‘Free software’, Boston, MA, USA. Hildebrando (2002), ‘Presentación corporativa’, (Executive report). Innovatia/Sedec/Gobierno del Estado de Aguascalientes (2002), ‘Catálogo de Productos Informáticos de Aguascalientes’. Instituto Nacional de Estadística, Geografía e Informática (INEGI) (2002), ‘Censos Económicos’, Aguascalientes, Mexico. Kagami, Mitsuhiro and Akifumi Kuchiki (2001), ‘Silicon Valley in the south’, Revista de Economia Pílítica (Brazilian Journal of Political Economy), 21 (4), 130–48. Lerner, Josh and Jean Tirole (2002), ‘Some simple economics of open source’, Journal of Industrial Economics, 50 (2), 197–234. López-Martínez, Roberto and Andrea Piccaluga (2001), Knowledge Flows in National Systems of Innovation, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Lubeck, Paul and Kyle Eichen (1998), ‘Silicon Islands and Silicon Valleys: rethinking Mexican regional development strategies in an era of globalization’, Center for Global International and Regional Studies, University of California Santa Cruz WP #1998-1. McBreen, Pete (2002), Software Craftsmanship. The New Imperative, Reading, MA: Addison-Wesley. Meyerhoﬀ, D., B. Laibarra, Kraan R. Van der Pouw and A. Wallet (eds) (2002), Software Quality and Software Testing in Internet Times, Berlin: Springer-Verlag. Mocorrea, Sebastián (2002), ‘Tecnologías de la Información: el nuevo escenario’, Powerpoint presentation, Mexico, 22 November.

350

Agglomeration in the Americas

Organization for Economic Cooperation and Development (OECD) (1999), Managing National Innovation Systems, Paris: OECD. Organization for Economic Cooperation and Development (OECD) (2002), Information Technology Outlook, Paris: OECD. Peneder, Michael (2001), Entrepreneurial Competition and Industrial Location, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Rosenberg, Nathan (2001), Schumpeter and the Endogeneity of Technology, London: Routledge. Robén, Camarillo Ortega (2002), Política de Fomento Económico Aguascalientes Visión 2020, Aguascalientes, Mexico: Gobierno de Aguascalientes. SAP (2001), ‘Annual Report’, Mexico. Secretaría de Economía (2002), ‘Programa para el desarrollo de la industria del software’, Mexico. Select IDC (2000), ‘Servicios de Tecnología de Información, Perspectivas del Mercado Laboral en la Industria de Tecnología en México’, Bulletin SRDB0812, Mexico. World Information Technology and Services Alliance (WITSA) (2002), Digital Planet 2002: The Global Information Economy, WITSA.

PART IV

Conclusion

14.

Conclusions Masatsugu Tsuji, Mitsuhiro Kagami and Emanuele Giovannetti

Industrial agglomerations fall into four general categories: (i) clusters where locally specialized items are produced, or local product districts; (ii) clusters where a large core ﬁrm has many subcontracting or parts makers surrounding it, or so-called industrial castle towns (jokamachi); (iii) clusters in large cities where there exist lots of basic production processes or urban-processing clusters; and (iv) government-supported industrial parks and estates often seen in developing countries. Examples of local product clusters include porcelain at Meissen, Germany, cutlery at Solingen, Germany, textiles at Lyon, France, kitchenware at Tsubame City, Japan, and the famous Italian industrial districts such as shoemakers at Riviera del Brenta. These clusters are characterized by specialized local products and agglomerations of small-scale enterprises with close community networks. Clusters specializing in information technologies (IT) are, in a sense, classiﬁed in this category. Examples are Silicon Valley and Route 128 in the US, Bangalore in India, Zhong Guan Cun in China, and newly developed Aguascalientes in Mexico. These clusters are characterized by the presence of software industries and venture capital, along with a large pool of computer-literate workers. The second category includes automotive agglomerations such as Detroit and Toyota City, large-scale capital goods industries such as iron and steel (Pittsburgh and Kitakyushu City), and chemical kombinat (Russian word meaning ‘compound’ or ‘terminal’) (Jurong Petrochemical Complex in Singapore and Yokkaichi City in Japan). These clusters are typiﬁed as one large company surrounded by many layered parts and component makers or supporting industries. Examples of the third category include Ohta Ward in Tokyo and Higashi Osaka in Osaka, where small- and medium-scale ﬁrms are specialized in basic processing such as casting, forging, molding, welding, plating, heat treatments, grinding and polishing. These areas feature a concentration of high-skilled manpower and closely interwoven specializations among small ﬁrms. 353

354

Conclusion

Examples of the fourth category include Chu-Chiang River Delta in China, Hsinchu Science Park in Taiwan, Penang (Bayan Lepas, for example) in Malaysia, Leam Chabang in Thailand, and Haiphong in Vietnam. These areas are assigned by central as well as local governments as industrial or export-processing zones and given incentives such as tax holidays and preferential loans. Thus, mainly foreign multinational corporations (MNCs) gather in these zones. Nowadays, even private companies themselves participate in the development of these industrial estates. These industrial clusters have several shared features. First, there are strong externalities derived from vertical as well as horizontal transactions through subcontracting and inter-ﬁrm collaboration, which produce ‘collective eﬃciency’. Accumulated knowledge is also shared among ﬁrms and knowledge spillovers create externalities. Second, increasing returns work in clusters as a whole because constant returns to scale and the perfect mobility of factors cannot explain spatial diﬀerences (‘world without cities’) (see Fujita and Thisse 2002, p. 6). Increasing returns to scale are an inevitable element of industrial agglomeration. Third, transportation and communication costs are also fundamentally important. If these costs are high, ﬁrms have incentives to concentrate. Fourth, knowledge is crucial in any cluster; R&D activities, new devices and skills are shared in the community. High-quality workers are also essential. Especially, software clusters need these factors. In addition, face-to-face communication is valuable when ‘tacit knowledge’ is exchanged. Thus, knowledge-based clusters will have more successful results. This implies the importance of vocational training and education. Fifth, demand is vital for the existence of industrial clusters. In particular, urban processing clusters and jokamachi-type clusters can exist so long as there is stable demand for their products. As shown later, how clusters correspond to demand shifts is critical. Lastly, the role of government, whether central or local, is also of signiﬁcance, particularly in developing countries. These clusters consist mainly of small and medium-sized enterprises, which indicates the importance of SMErelated government policy. The most crucial question is why agglomeration occurs in a given location and not in others. Our study did not clarify this point. One obviously important factor is a region’s endowment of natural resources, such as the existence of raw materials, as well as important transportation crossing points, and historical and cultural factors. Toyota City has its auto cluster because the Toyoda family came from there; Sakichi Toyoda invented the ﬁrst textile weaving machines in Japan and founded an auto-assembling company, later called the Toyota Motor Corporation. Sialkot in Pakistan is famous for its surgery-related medical products because, originally, a ﬁeld hospital was constructed in that locality in the British colonial era. Yawata

Conclusions

355

Works chose Kitakyushu City because coalﬁelds were nearby, and for the natural defenses oﬀered by the inland bay. Further investigation of the question of location, however, should focus on endogenous interactions.

CHANGING CLUSTERS IN ADVANCED COUNTRIES Industrial clusters in developed countries have recently faced three main challenges. First, catch-up by developing countries aﬀects a number of industries in advanced countries, and hence their industrial clusters. For example, traditional light industries such as textiles and food processing face severe competition from imports from developing countries. Heavy industries such as iron/steel and shipbuilding face similar competition from middle-income countries like Brazil and Korea. Even the software industry is confronted by rapid catch-up from some developing countries such as India. Second, globalization and deregulation anywhere results in increasing penetration of MNCs and relocation of their factories and oﬃces into strategic locations, thereby reforming global value chains. New ties between MNCs and traditional local product districts (‘external linkages’) are being forged in order to maximize total value chains. An open question is: does this inﬂuence local product districts negatively or positively? Furthermore, globalization produces intense complications for clusters. A typical example is the Japanese MNCs’ move into China. This creates fears of job losses for traditional industrial clusters, particularly urban-processing clusters; this is the so-called ‘hollowing-out’ of manufacturing industries. Third, the IT revolution leads to a world without distance. Using e-commerce (electronic date interchange), assemblers can purchase high-quality parts and components at lowest cost anywhere in the world. This is a fundamental change for industrial agglomerations, especially jokamachi-type clusters. Our studies and other interesting empirical evidence show that some eﬀorts are being made by clusters themselves to contend with these diﬃculties. These issues can be summarized as follows: ●

Regarding light industries, local product clusters in developed countries have shifted toward more value-added and knowledgeintensive products such as the textile and garment industries in Fukui and Lyon. Some ﬁrms in Emilia-Romagna began to form linkages with non-district ﬁrms, that is, the emergence of the ‘multilocated’ industrial districts. A number of local clusters have struggled to market new products, such as when Japan’s Tsubame City

356

●

●

Conclusion

shifted from producing kitchenware to making sports and outdoor equipment. In the case of heavy industries such as the iron and steel industry, Kitakyushu City realigned its accumulated resources and knowledge to enter completely new ‘vein industries’, including recycling of PET bottles, automobiles, electronic and oﬃce machines and ecology-oriented businesses. The hegemony of US software clusters may not be eroded in the near future, but studies suggest the possibility of the emergence of ‘nearshore’ platforms in neighboring countries such as Mexico, and Central America and the Caribbean. The case of Mexico is most likely related to its NAFTA status. In this connection, the establishment of the Free Trade Arrangement for the Americas in this decade will reinforce such tendencies in those countries.

Globalization has had a dramatic eﬀect on industrial districts. For example, the Italian shoe industry at Riviera del Brenta was ﬁrst integrated into global value chains of famous fashion brands by subcontracting production processes. This led to the abandonment of key activities such as design, sales and marketing of local traditional ﬁrms, and a weakening of the district’s power. However, in this case, the Italian shoe industry itself may ultimately become more eﬀective due to this specialization because many of the large famous brand companies are Italian. Urban-processing clusters have confronted the ‘hollowing-out’ phenomenon in Japan. They have attempted to search for more sophisticated, high value-added areas involving extremely precise and diﬃcult processing skills. In particular, they participated in new product development and design with Japanese MNCs by playing the role of seed ﬁnders or performing seedbed functions. The inﬂuence of IT on the purchasing of parts and components was enormous. Japanese automakers such as Nissan and Mazda were forced to dispose of their majority holdings into foreign hands and introduce costcutting reforms, especially by reorganizing their subcontracting systems. However, automakers relocated plants from Central Japan to outlying areas during the bubble-economy period in the early 1990s because of land congestion and labor shortages. This was made possible once the problem of distance was overcome by the use of the Internet between headquarters and local plants. Although Toyota opened a new Kyushu plant in 1992, overall concentration near its headquarters in Aichi Prefecture has been strengthened. Of all Japanese automakers, only Toyota seems to have remained immune from the eﬀects of globalization and the IT revolution and

Conclusions

357

Japan’s prolonged recession after the bubble burst, accumulating large proﬁts and running the company without debt for the last 50 years. Distribution systems for parts and components cannot be replaced by IT, but the supply chain management made possible by IT can reduce time and costs. In this context, the just-in-time system, Toyota’s well-known invention, is essential for parts suppliers to continue to concentrate in Toyota’s jokamachi.

INDUSTRIAL AGGLOMERATIONS AS AN ENGINE OF GROWTH IN DEVELOPING COUNTRIES Industrial parks in developing countries are of importance for the economic growth of those countries. East Asian countries have utilized such parks and zones to build momentum for growth. China is now exploiting such parks, utilizing foreign direct investment on a large scale. Creation of regional industrial agglomerations may help a developing country grow as a whole. In particular, the development of the software industry might be a detonator for growth. The Indian software industry suggests that certain developing countries may follow a non-traditional path to successful industrialization (‘leapfrogging’ industrialization) (see Kagami and Tsuji 2002). The success of industrial agglomerations depends on (i) infrastructure (highways, ports, electricity supply and so on); (ii) institutional frameworks (legal systems, participatory actors, coordination among actors and so on); and (iii) government support (or foreign assistance) in terms of laws, taxation and ﬁnance. As shown in this volume, the case of northern Vietnam provides good examples of where collaboration in these three areas worked well. However, if one region grows, other regions are often left behind. Regional disparity is a thorny issue as illustrated by the large income diﬀerences between the western and eastern parts of China. In conclusion, we must deepen our theoretical background on the fundamental question of how growth and location aﬀect each other (or whether regional discrepancies tend to widen or narrow over time).1

NOTE 1. More precisely, we need dynamic core–periphery theories in this respect (Fujita and Thisse 2002, p. 388).

358

Conclusion

REFERENCES Fujita, Masahisa and Jacques-François Thisse (2002), Economics of Agglomeration: Cities, Industrial Location, and Regional Growth, Cambridge: Cambridge University Press. Kagami, Mitsuhiro and Masatsugu Tsuji (2002), ‘Digital divide or digital jump: beyond the “IT” revolution’, Institute of Developing Economies (JETRO), Chiba City, Japan. Kagami, Mitsuhiro, Masatsugu Tsuji and Emanuele Giovannetti (eds) (2004), Information Technology, Policy and the Digital Divide: Lessons for Developing Countries, Cheltenham and Northampton, MA: Edward Elgar.

Index accounting solutions 323, 329, 337 accumulation paradox 67–71, 92 ACRIB (Associazione Calzaturiﬁci della Riviera del Brenta) 229, 238–9, 240, 241 Adobe Systems Inc. 292, 293 agglomeration as distinct from cluster 3 see also industrial agglomeration Aguascalientes, IT clusters in 313, 319, 338–43, 344, 353 Ahn, H. 153–4 Aichi Prefecture 27, 33 FDI in North America by automotive parts manufacturers located in 26–7 localization of automobile industry in 13, 18–22, 356 Aichi Seiko 13 Aikawa, Yoshisuke 36 airline connections 89, 120, 123, 128 Aisan Industry 30, 31 Aishin AW 30, 31, 33 Aishin Seiki 13, 30, 31 Alba, Francisco 336 Alfa 330 All Soft 335 Amata industrial zone 116, 117 American Automobile Network Exchange (ANX) 29 Amiti, M. 174 Analysys Consulting 298 ANCI 228 Andersson, T. 144 Anjo 19, 20 Annen, K. 274 Anshan Iron and Steel Group 176 Antar 335 Appalachia coalﬁelds 56 Apple Computers 298, 314, 327 application service providers (ASPs) 323, 343, 346

Arakawa Auto Body 30, 31 Arora, A. 289, 301, 303 ‘artery’ industries 39, 54, 60 Asanuma, B. 15 ASCI 337 AS400 336 Asia-Paciﬁc Import Mart (AIM) 54 Asian crisis 118, 140 Asian Development Bank 119 Asociación Mexicana de la Industria de las Tecnologías de la Información (AMITI) 320–21, 327 Aspel 320, 323, 324, 329, 337 Aspray, W. 298, 300 Assopiastrelle 256 Athreye, S. 301, 303 Audretsch, D.B. 249, 252 automobile industry, Japanese 9–32 globalization aﬀecting location of 22–8, 30–31, 32, 356–7 IT applications in 29–32, 39, 60, 356–7 in Kyushu 18, 39, 56, 357 localization in Aichi Prefecture 13, 18–22 production structure of Toyota 13–14 compared with non-hierarchical US system 14–17 production system of 10–13 Autonomous University of Nuevo Leon (UANL) 334 average capital productivity 98, 118 Axsis Tecnologia 332, 334–5 Azzalini, A. 212 B. Braun Carex 254 Baake, P. 283 backbone networks 275, 277–8 Baden-Fuller, C. 249 Bai Chay Bridge 121

359

360

Index

Balassa, B. 205, 206 Balassa Index of Revealed Comparative Advantage (RCA), applied to Italy 205–18 Balasubramanyam, A. 316 Balasubramanyam, V.N. 316 bandwidth 271, 280, 282 Bang Khadi industrial zone 105, 107, 108 Bang Pakon industrial zone 105, 107, 109, 111 Bang Phlee industrial zone 105, 106, 108 Bang Poo industrial zone 105, 106, 108 Bangalore, software cluster in 1, 140, 316, 353 Bangchan industrial zone 105, 106, 108 Bangkok distance from Hanoi 129 distance of industrial zones from 106–7 industrial zones near 104, 105, 111 Bangkok Industrial Park 111 banks 322, 330 Bansal, A. 299 Barba Navaretti, G. 259 Barr, A. 291, 294, 295, 296 barriers to entry, in luxury goods market 231 Barro, R.J. 100 Batisse, C. 174 Battese, G. 89 Batum, industrial zones in 116 Bayan Lepas industrial zone 102, 354 Becattini, G. 219, 220, 248, 273 Becchetti, L. 250 Beijing 178, 180 Beijing Economic–Technological Development Area 199 Berg, T. 298, 299, 300, 303 Bhagwati, J. 293 big-push theory 195–6 Biggiero, L. 252 Bihn Bridge 121 Binhthuan, growth rate of 119 Bintan, industrial zones in 116 Bioﬁl 254 biomedical district of Mirandola 250, 252–5 biotechnology 139, 273

Blitz Software 326 Blonski, M. 273 bluetooth 67 Board of Investment (BOI) of Thailand 105 Boari, C. 249, 254 Borrus, M. 293 Bowman, A.W. 212 brand names 230–31, 234 Brazil 50, 256 Brenta shoe district 228–9, 353 customers 229–30 implications of globalization for 241–3 local governance in Brenta 238–41 in the top brand chain 232–8, 356 broadband 64, 140 Brooks, F. 308 Brown, E. 298, 299 brownﬁeld sites, restoration of 39, 57 Brusco, S. 247, 248, 249, 273 Brynjolfsson, E. 65 BT Garment Co., Ltd 129–34 bundled software 314 Bureau of Statistics, Japan 80 business promotion policies, Korean 147–8 Business Software Alliance (BSA) 293, 301 Business Week 298 buyer-driven chains 227, 231, 237, 242 Buzali, Alfredo 327 cable communication, and population density 73, 74, 75, 78, 82 Cai, F. 174 Cairong port 120, 121, 123, 124 California 23, 27, 293 Camagni, R. 244 Canada, Toyota plants and parts suppliers in 27 Canon 119, 128, 129, 130, 135, 136, 293 capability maturity model 331 capacity building 121, 137 capital–labor ratio 214 Carmel, E. 289, 293 Carmel Ray industrial zone 114, 115 Carnegie-Mellon University 331 Castellón ceramic tile cluster 255–6

Index CCC21 (Construction of Cost Competitiveness for the 21st Century) 32 cellular phones communication using and population density 73, 76, 77, 79, 83 relationship between face-to-face meetings and 85–92 penetration rate of 64, 76, 83 CEMEX 330, 334, 335, 336 Cemtec 334 Censis 250 centripetal and centrifugal forces 2–3, 268 aﬀecting Japanese automobile industry 29, 39, 60 aﬀecting Japanese steel industry 60 aﬀecting North American software industry 291–9, 299–300 aﬀecting peering decision 279 information technology aﬀecting 270, 271 relational governance as a centripetal force for districts 272–5 ceramic tile district of Sassuolo 250, 255–8 chaebol 145, 146 chemicals industry, Italian 216 Chikuho coalﬁelds 35, 41, 56 China change in regional and industrial disparity in 175–82 Construction of the Third Front in 177, 180, 196 FDI in regions of 180, 183, 357 FDI inﬂows in 2001 119 Five Year Plans 176, 195, 196 GDP growth rate 173 industrial agglomeration in eastern area of 182–95 industrial zones in 116, 118, 354 iron ore imported from 41 Japanese investment in 23, 43, 50, 139, 355 population in 2000 181 relations with USSR 176–7 taxes in 125, 126 trade with Japan 54

361

wages in 122 western development 195–7 Zhong Guan Cun IT cluster in 353 Chiryu 20 Chrysler 32 Chu-Chiang River Delta 354 Chuo Spring 30, 31 Cikarang industrial zone 112, 114 cities as information base 72–81, 92 IT clusters in Japanese cities 68, 70, 93 spillover eﬀects and growth in 150–64 support for development of medium-sized cities in Mexico 340 City College of New York 337 CLAIR (Council of Local Authorities for International Relations) 57 clothing industry, Italian 209, 210, 216 Club dei Distretti 250 cluster as distinct from agglomeration 3 see also industrial clusters clustering policy in Aguascalientes, Mexico 319, 341–3 in China 174 importance of 5, 60, 61, 354 see also tax and lending incentives CMM assessments 335, 338 CNEL/Ceris-CNR 250–51, 255, 256, 259 coalﬁelds 35, 41, 56 Coase, R.H. 15 collective eﬃciency 5, 354 commodities, distribution of, by distance 85, 86 communication technologies, types of 73–4 complementarity between face-toface meetings and telecommunications 81–93, 149–50 and population density 74–81, 82, 83 Compac 337–8 Compaq 293 comparative advantage based on connectivity 270–71

362

Index

factor proportion theory of 214 in Italy 4–5, 205–19 complementary eﬀects, face-to-face meetings and IT 81–2, 93, 149–50 empirical evidence for 83–92 COMPRA-NET 328–9 Computer Aided Design (CAD) systems 254 computer-assisted software engineering 294–5 concurrent engineering 32 congestion 3, 39 connectivity, comparative advantage based on 270–71 Consejo Nacional de Ciencia y Tecnología (CONACYT) 328 Consorcio Uno 327–8 Consorzio Centro Veneto Calzaturiero 238 Consorzio Maestri Calzaturieri del Brenta 238 consumption goods, diversity of 149 ContaPAQ 337 Conti, G. 219 cooperative companies 13–14 cooperative information sharing 273 core competencies 231, 234, 243 core–periphery model 174, 195–7, 269 dynamic 357 corporate tax rates 99, 111, 125, 126 Corporation Recycle Tech 39 Correa, C. 289, 290 cost accounting 323 Crawford, V. 274 CRM software, see customer relations management (CRM) software CT&D Group of Taiwan 116 custom software 293, 316, 323, 327, 331 customer relations management (CRM) software 64, 329, 330, 334, 346 customs clearance 125 customs operations, software assisting 326–7 cutlery cluster in Solingen, Germany 353 CVEC 116 CxNetworks 334

Cygnus Tech 335 Czech Republic, Japanese investment in 23 D’Aspremont, C. 268 Da Nang 123 Daedeok Science Town 145–7 Daejeon Metropolitan City 145–7, 160, 165 Dahlman, C. 144 Daihatsu 13, 30, 56, 136 Dalesio, E. 295 Danang, growth rate of 119, 137 Danang export-processing zone 116 Ddemesis 330 De Benedictis, L. 207, 215, 219 de Nardis, S. 210, 215, 219 decentralization IT investment and 67–8, 92 national decentralization program in Mexico 340 of power in China 177 of production to low-wage countries 236, 237, 241, 242 dedicated assets 255 Dedrick, J. 289 defense considerations 40, 41 Deng Xiaoping 177, 196 Denso 30, 31, 33, 130, 136 Desarrolladores 327 Detroit, automotive cluster in 353 developed countries, changing clusters in 355–7 development zones 183 Dia 327 dialogue projects 327–8 digital subscriber line (DSL) 64 diseconomies of scale and scope 299 distributed networks 67 Distrito Federal, IT clusters in 312, 313, 317, 319, 320–21, 327–8 DITSNBS (Department of Industrial and Transport Statistics of National Bureau of Statistics), China 184, 186 diversiﬁcation 231 diversity, and agglomeration formation 148, 151–4 in Korea 154–8, 159, 161–6

Index division of labor among ﬁrms 182, 183, 185, 187, 248, 256 inter-regional 174 Dixit, A. 173, 182 Dodge, M. 271, 283 Dokai Bay 41 Dolan, C. 227 domain knowledge, US control of 291, 293 Dong Nai 123 Dong-Sung Cho 145 Dongfeng-Citroën Automobile Co., Ltd 187 Dragon Logistics Center 125, 130 Du Yang 174 Dutta, S. 296 dynamic scale economies in industrial districts 214, 215 e-business 329, 334 e-commerce 239, 355 e-kanban 29 East Jakarta Industrial Park 112, 114 ECLAC–World Bank 206, 210, 211, 213 ecology towns 39, 44, 54, 58 Economic Planning Agency, Japan 93 economic rents 231, 236, 237, 238, 243 Economist 290, 293 education sector Korean 142–3, 168, 169, 170 Softtek’s linkages with 331–4 software support for Mexican education sector 329, 346 see also schools; universities Egan, T. 293, 294 Eischen, K. 290, 291, 293, 294, 295, 298, 299, 308 El Informador 337 El Salto 319 electricity supplies 59, 120, 195, 357 electronics cluster in Jalisco 319, 335–6, 338 Ellison, G. 173 Ellison, Larry 343 embedded software 291, 341 embodied technological change 252, 257

363

Emilia-Romagna region, industrial districts in 5, 247–65 biomedical district of Mirandola 250, 252–5 catching-up issues facing 355 ceramic tile district of Sassuolo 250, 255–8 econometric analysis of technological change in 258–64 identiﬁcation of districts 250–51 machinery industry in 251–2, 254–5 MNCs in 253–4, 255, 264–5 employee relationship management (ERM) software 329, 330, 335, 343–6 encryption technologies 272 endogenous growth theory 71 English language proﬁciency 301, 302, 303, 304, 305, 307, 331 enterprise resource planning (ERP) software 64, 329 environmental protection, investment in 39, 53 Epifani, P. 215, 219 ERM software, see employee relationship management (ERM) software ERP software, see enterprise resource planning (ERP) software European Patent Oﬃce 248, 258, 261 Evans, P. 289 Expansion 324 expatriate knowledge networks 290, 301, 302–3, 304–5, 307 explicit contracts 273, 275 export agents 230 export-oriented ﬁrms, innovative capability of 259, 260, 263 export-processing zones (EPZs) in Indonesia 105, 112, 113, 114 in Malaysia 102, 105 in Philippines 115 in Thailand 104, 105, 111 in Vietnam 116, 117, 119, 125 external diseconomies 27, 49, 60, 249 external economies, as source of agglomeration 2–3, 268 communication technologies and 73, 275

364

Index

in industrial districts 214, 215, 248, 249, 255, 260 knowledge spillovers and 5, 37, 60, 255, 312 in software industry 291, 293, 312 external service providers (ESPs) 295 Fabiani, S. 219, 220 face-to-face interactions, see meetings (face-to-face), communication using factor proportion theory of comparative advantage 214 Factory Forestation Agreements 43, 53 Fair Trade Commission 47 fashion industry, top brand value chain in 230–32 Brenta in 232–8, 356 Fehr, E. 275 Feldman, M.P. 252 FEMSA/Cervecería Moctezuma 330, 334, 335 Fernandez, Sergio 337 ﬁliere organization 251–2, 255 Filstrup, B. 283 ﬁrm size, and innovative capability 258, 259, 261, 263 First Cavite industrial zone 114, 115 ﬂexible specialization, in industrial districts 248–9 focal ﬁrms 249 Fondazione G. Brodolini 250, 251 footwear industry, Italian Brenta shoe district 228–30, 232–43, 353, 356 comparative advantage of 210, 216, 218 and global value chains 227, 228–43, 356 foreign direct investment (FDI) in Chinese regions 180, 183, 357 eﬀect on clusters 355 expansion of Japanese (1971–1973) 47 in industrial zones 98–120, 128–34, 135, 136–7, 354 decisive factors for 121–8 investment function of 100 by MNCs in industrial districts 253–4, 255, 264–5

by MNCs in Mexico 314, 319, 320, 335–6, 338, 340–41, 343–6 motivation for 23 by Nippon Steel Corporation in developing countries 50 relocation of Japanese FDI to China 139, 355 by Toyota and its parts suppliers 22–8, 30–31, 32 foreign ownership, constraints on 99, 101–2, 114, 128, 340 Forni, M. 252, 255 Forrester Research 299 foundations 58, 59 France Brenta shoe district sales to 229 Japanese investment in 23 Lyon textile cluster in 353, 355–6 free riding 278, 283 free trade zones (FTZs) 102–3 Freeman, P. 298, 300 freight, distribution by distance 85, 86 Fuji Iron and Steel Co. 40, 43, 47 Fujita, M. 3, 97, 149, 150, 173, 174, 182, 199, 268, 269, 270, 354, 357 Fukui, textile industry in 355 Fukuoka City 36, 70, 93 Fuller, W. 89 functional downgrading 233–4, 237, 238, 242 functional upgrading 228, 234, 242 furniture industry, Italian 209, 216 Futaba Industrial 30, 31 Gambro-Dasco 253, 254 Gansu 177 Garofoli, G. 264 gas supplies 59, 195 Gasper, J. 94 Gateway Business Park 114, 115 General Electric 319, 330 General Motors (GM) 14–17, 23 General Statistical Oﬃce, Vietnam 119, 137 Gereﬃ, G. 225, 227, 231, 237–8 Germany blast furnace technologies from 41 Brenta shoe district’s sales to 229 comparative advantage of 208, 209 Meissen porcelain cluster in 353

Index printing equipment cluster in 4 Solingen cutlery cluster in 353 Gifu Prefecture 19, 20 Gini’s coeﬃcient 175 Giovannetti, E. 1, 249, 269 Glaeser, E.L. 94, 148, 150, 153, 154–5, 158, 173, 269 Glanz, J. 298 globalization 1, 4 complications for clusters posed by 355 eﬀect on Japanese automobile industry 22–8, 30–31, 32, 356–7 and industrial districts 5, 264–5, 225–43, 264–5, 356 Gnutella 67 government software niche in Mexico 328–9 government support for software development 301, 303, 304, 305, 329, 338 gravity model 87 Grolliers 293 gross industrial output value (GIOV) 175–6, 180–81, 182, 186 gross operating surplus, and innovative capability 259, 260 growth in regions, see regional growth growth sharing 17 GRP software 346 Grupo Asercom 335 Guadalajara software cluster 317, 319, 335–8, 339 Guangdong 178, 179, 180, 189–91 Guangzhou 129, 189 Gucci 231 Guerrieri, P. 219 Guizhou 177 Gyeongbuk 143, 160, 165 Gyeonggi 140, 141, 143, 160, 165 Haiphong 119–20 growth rate of 119, 137 industrial cluster in 129–34, 135, 354 transportation network 119, 120, 123, 124 wages in 121, 123 Haiphong port 98, 120, 124, 128, 134, 135 Hamaguchi, N. 97, 173, 182

365

Hanoi 119–20 growth rate of 119, 137 industrial cluster in 129, 130–31, 135, 136–7 living conditions in 123 location of 129 transportation network 119, 120, 121, 123, 124 wages in 121, 122, 123 Hanson, G.H. 173 hardware industry Korean 141–2, 143, 165 Mexican 312, 313 United States 294, 295, 296–8 Harian, B. 104 Harrison, B. 264 heavy industry clusters factors explaining 59–61 see also Kitakyushu City industrial cluster; Yawata Works Hebei 179, 180 Hecker, D. 296 Heckscher–Ohlin theorem 214, 215 Heeks, R. 289, 302, 303, 307 Heilongiang 178, 180 Hekinan 19, 20 Helleiner, G.K. 289 Henan 177, 179, 180 Henderson, V. 252 Herﬁndahl concentration index 74–7 Herschman–Herﬁndahl Index (HHI) 140–41 Hewlett-Packard (HP) 150, 292, 298, 336–7 Hibikinada Hub Port Initiative 39, 54 hierarchical production system 10, 11, 12–13 compared with non-hierarchical system 14–17 international division of labor and 28 reason for 29 Higashi-Mikawa region 19, 20 Higashi Osaka 353 Higashida 41, 42, 43, 55 highways 98, 137, 357 national highway Route 5 (Vietnam) 98, 118–21, 124, 128, 134, 135 national highway Route 18 (Vietnam) 120, 121, 124

366

Index

national highway Route 10 (Vietnam) 121, 124 Tokyo Metropolitan Highway 46 Hikari Works 40, 43, 50 Hildebrando 320, 323–6, 329, 330 Hino Motors 13 Hiroshima City, IT cluster in 70 Hitachi Metals 36 Hitt, L. 65 Ho Chi Minh City distance form Hanoi 129 growth rate of 119, 137 wages in 122, 123 Hokkaido 18 Holtzman, S.R. 268 Honda 23, 31, 33, 56, 128, 135, 136 Hong Kong 54, 129–34 Honshu 36, 39 Hosei Brake 30, 31 hospitals 120, 123 hot potato routing 277, 278 Hotelling, H. 268 hotels 123 housing 120, 123 Houston, oil and gas-related cluster in 293 Hsinchu Science Park 354 Hu, D. 174 Hubei 177, 179, 180, 187 human resources 121, 137, 149 Humphrey, J. 227, 228 Hunan 177 Hungyen 119–20, 137 Huston, G. 283 Hylsa 330 Iapadre, L. 219 IBM 298, 292, 314, 319, 320, 321–2, 323, 329, 336, 343 Icaza, Miguel 346 Ikeda, Hayato 46 Illinois, Toyota plants and parts suppliers in 23, 27 Imagawa, T. 94, 149 immigrant workers 241, 295–6, 298, 331 imperfect competition 3, 182, 268 IMPERO database 258 implicit contracts 273 import tariﬀs 99, 126, 244 Inazawa 19

increasing returns, as source of agglomeration 3, 5, 149, 182, 226, 269, 354 dynamic scale economies in industrial districts 214, 215 in Kitakyushu City 37, 60 Toyota parts suppliers 29 incubation business 147 India FDI inﬂows in 2001 119 industrial clusters in 1, 97, 140, 316, 353 Japanese investment in 23 software development for US market 289, 331 software industry and growth in 357 Indiana, Toyota plants and parts suppliers in 27 Indonesia corporate tax rate in 125 export processing zones in 105, 112–14 industrial clusters in 97 industrial zones in 98, 105, 111–14 Japanese investment in 23, 111, 139 taxes in 126 wages in 121, 122 industrial agglomeration deﬁnition of 174, 182–3 mechanism of agglomeration formation 148–9 proximity and 268–70 see also industrial clusters industrial castle towns (jokamachi) 61, 353, 354, 355, 357 industrial clusters advantages of 226 centripetal and centrifugal forces aﬀecting, see centripetal and centrifugal forces changing clusters in advanced countries 355–7 common features of 5, 354 and economic growth 1, 4, 357 examples of 1, 4, 353–4 factors explaining heavy industry clusters 59–61 reasons for geographical location of 354–5 types of 353

Index industrial districts in Italy Brenta shoe district 228–30, 232–43, 353, 356 and comparative advantage 4–5, 215–17, 218 deﬁnitions of 4, 248–9 in Emilia-Romagna region, see Emilia-Romagna region, industrial districts in governance of 273–5 identiﬁcation of 250–51 ‘multi-located’ 257, 264–5, 356 SMEs in 247, 249, 250, 253, 256 types of 244 and value chains 5, 225–43, 356 Industrial Estate Authority of Thailand 104–5, 106–10, 134 industrial groups, membership of, and innovative capability 259, 260, 261–3 industrial parks 5, 340, 357 industrial policies 46, 60; see also tax and lending incentives Industrial Revolution 63–4, 269 industrial zones (IZs) characteristics of 99 in China 116, 118, 354 development by private sector 98, 99, 103, 105–18, 134, 354 development by quasi-public sector 102–5, 106–10, 134 and economic growth 97–8, 100–102, 134 in Indonesia 98, 105, 111–14 in Malaysia 98, 99, 102–4, 354 in Myanmar 116, 118 in Philippines 98, 111, 114, 115 as quasi-public goods 97–8, 99, 105, 134 successful players for industrial agglomeration 98, 134 in Taiwan 354 in Thailand 98, 99, 104–5, 106–10, 111, 134, 354 in Vietnam 97, 98, 354 development by quasi-public sector 114–18 northern Vietnam 118–34, 135, 137 Información y Servicios Tecnológicos (INFOTEC) 328, 329

367

information disembodied and embodied 71–2 production system and required amount of 15–16 information costs 15 information spillovers 2, 3, 71–2; see also knowledge spillovers information technology growth of regional technology depending on improvements in 153–4 IT cluster in Bangalore 1, 140, 316, 353 IT cluster in Silicon valley 1, 5, 68, 71, 140, 150, 273, 289, 353 IT clusters as local product clusters 353 IT clusters in Japanese cities 68, 70, 93 IT industry in Korea characteristics of software sector 161–8 factors promoting growth of software sector and industrial agglomeration 144–8, 168–70 IT, R&D and the education sector 141–4, 168, 169, 170 IT market in Mexico 312, 313 IT revolution 63–4 accumulation paradox under 67–71, 92 face-to-face interactions and 81–93, 355 increasing role of density under 71–81 productivity paradox under 64–7, 92 in Japanese automobile industry 29–32, 39, 60, 356–7 knowledge creation, agglomeration and IT 149–50 and ‘multi-located’ industrial districts 258, 264 see also hardware industry; Internet; IT service industry; software industry Innovatia 341, 342 insiderness 271 Institute for Information Technology of Jalisco (IJALTI) 338

368

Index

Institute of Precision Molding (Penang Skills Development Center) 102 institutions capital productivity enhanced by building 118 industrial zones facilitating 99, 125 reforms to attract FDI 125–8, 134, 137, 341, 357 for software development 312–14 Instituto Nacional de Estadística Geograﬁa e Informática (INEGI) 317 Instituto Tecnológico de Estudios Superiores de Monterrey (ITESM) 329, 334, 342, 346 instrumental variables (IV) approach 89 intangible activities 243 Integrasis 335 integrated device manufacturers (IDMs) 296–8 intellectual property rights (IPRs) in software 291, 293, 301, 303, 304, 305 Intelmex 326 interconnection agreements 276 interlocking directorate 12, 13 intermediate goods, as source of agglomeration 149, 182 in cities in East Asia 97 machinery industry in China 183, 185–94 Internacional de Sistemas 332, 335 international fairs 238, 239 International Monetary Fund (IMF) 46 Internet 1 clustering forces in the Internet industry 5, 268–83 agglomerating forces on peering decisions 279–83 composition of Internet 275 interconnection agreements 276 peering decision 277–9 proximity and agglomeration 268–70 proximity in cyberspace 270–72 relational governance as a centripetal force for districts 272–5 technological aspects 276–7

economic performance driven by 65 information diﬀusion and 71, 356 penetration by region in Korea 160 platform for Internet operations in Mexico 346–8 spread of 4, 64, 67, 316, 326 support to businesses through 323, 327, 338 tacit knowledge transferred via 270 Internet exchange points (IXPs) 276 internet service providers (ISPs) 275 peering agreements 275, 276, 280–82 peering decision 277–8 agglomerating forces on 279–83 transit agreements 276, 278 INTERNET 2 software project 329 internship programs 334 inventories 26 investment licenses 125 IPI (Istituto per la promozione industriale) 220 IRC 19 Ireland, software industry in 289 Iris Ceramica 265 Irmen, A. 269 iron ore 41 iron town cluster, see Kitakyushu City industrial cluster; Yawata Works Ishihara, K. 198 ISO 9001 certiﬁcation 334 Israel, software industry in 289 ISTAT 206, 215, 216, 218, 229, 250 IT, see information technology IT service industry Korean 141–2, 143 Mexican 312, 313 Italy Brenta shoe district in 228–30, 232–43, 353, 356 governance of industrial districts in 273–5 industrial districts and value chains in 5, 225–43, 356 industrial districts in EmiliaRomagna region, see EmiliaRomagna region, industrial districts in peering decision of ISPs in 279–83 specialization and comparative advantage in 4–5, 205–19

Index Itochu Corporation 105–11, 114, 116, 117 Itoh, M. 93 Jaﬀe, A. 93 Jakarta 122 Jalisco electronics cluster in 319, 335–6, 338 IT clusters in 312, 313, 319, 335–8, 339 Japan automobile industry in, see automobile industry, Japanese cities behaving as information base in 72–81 comparative advantage of 208, 209 crude steel production 47, 48–9 economic growth rates 46, 47 evolution of major steel companies in 45 exports to US 48–9 FDI by 22–8, 30–31, 32, 47, 50, 355 in industrial zones 103, 105–11, 114, 116, 119, 120, 123, 125, 128, 129, 130, 131, 132, 133, 135, 136–7 in Mexico 338, 340–41 iron town cluster in, see Kitakyushu City industrial cluster; Yawata Works IT boom in 63–4 IT clusters in cities 68, 70, 93 labor productivity and IT investment in 65, 92 population changes in 68, 69 production technologies originating in 316 relationship between face-to-face meetings and telecommunications in 83–93 tariﬀ on imported shoes 244 textile industry in 21–2, 355 trade friction/imbalances 47–9 Tsubame City kitchenware cluster in 353, 356 Japan Auto Parts Industries Association and Auto Trade Journal 19, 31 Japan Bank for International Cooperation 125, 127

369

Japan Electronics and Information Technology Industries Association 93 Japan Steel Corporation 41–6 Japanese International Development Agency (JAIDO) 114 Japanese trading corporations, industrial zones established by 98, 105–18, 119, 134, 137 jewellery industry, Italian 216 JFE (Kawasaki Steel Co. and NKK Corporation) 45 Jiang Zeming 195 Jiangsu 178, 179, 180 Jin, X. 174 Johor 102, 103 joint ventures automobile industry 23 Brenta shoe district participating in 239 diﬃculty of ﬁnding partner for 102 external purchase of parts by 188, 189 industrial zones 105, 114, 116 software industry 326, 328, 330, 331 steel industry 44, 50 theme parks 54–5 jokamachi 61, 353, 354, 355, 357 juko-chodai 59 Jurong Group of Singapore 114, 115, 116, 117, 353 just-in-time system 17, 21, 29, 316, 357; see also Kanban method Kagami, M. 1, 46, 357 Kanban method 13, 14, 17, 18, 19, 26; see also just-in-time system Kandori, M, 274 Kanto Jidosha 13 Kaplinsky, R. 225, 227, 231 Karawan industrial zone 112, 114 Kariya 18, 19, 20 Kariya group 21–2 Kasumigaseki Building 46 Kato, H. 174, 195, 198 Kauﬀman, D. 301 Kawakami, M. 97 Kawasaki Steel Corporation 50 Kedah State Economic Development Corporation 102

370

Index

keiretsu 67 Kende, M. 283 Kennedy, L. 97 Kentucky, Toyota plants and parts suppliers in 23, 27 Kernel 332, 334, 335, 336 kernel density 212–14 Khadria, B. 303 Khrushchev, Nikita 177 Kim, S. 173, 252 Kimitsu Works 40, 43, 47, 50 Kirkpatrick, D. 298, 299 Kitakyushu Academic Research Promotion City 39, 53 Kitakyushu City, publications by 35, 39, 53, 54, 183 Kitakyushu City Industrial Census 38 Kitakyushu City industrial cluster automotive industry attracted to 39, 56 compared with Pittsburgh, USA 56–9 as industrial castle town 353 iron and coal nexus cluster 35–7 reasons for cluster formation 59–61, 355 restructuring towards new businesses 37–9 revival following decline of iron and steel industry 53–6, 356 workforce living in 36, 38 see also Yawata Works Kitakyushu Eco-town project 39, 54 Kitakyushu International Airport 53 Kitakyushu International Distribution (KID) Center 54 Kitakyushu International TechnoCooperative Association (KITA) 37–9 Kitakyushu Port and Harbor Bureau 54 Kitakyushu Technocenter 39, 53 kitchenware cluster at Tsubame City 353, 356 Kitchin, R.M. 271, 283 Klette, T.J. 265 Knorringa, P. 244 knowledge agglomerations 319 knowledge-based industries 144–8 knowledge-based programming 295

knowledge management, concepts required for 64–5 knowledge spillovers to bridge endogenous growth and agglomeration 71–2 clustering concept focusing on 3, 5, 312, 354 factors aﬀecting 148–9 IT sector in Korea 161–70 regional growth model taking account of 150–53 regression analysis on growth in regions of Korea 154–64 results of previous studies 153–4 geographic proximity and 269–70 IT and 81–92, 93, 153–4, 161–8, 312 in Kitakyushu City 37, 60 and peering decision 279 Kobe Steel Ltd (KOBELCO) 44, 45, 50 Koda 19 Koizumi, Jun-ichiro 64 Kojima, R. 176, 198 Korea characteristics of software sector in 161–8 distribution of industries by region 140–41 factors promoting growth of software sector and industrial agglomeration in 144–8, 168–70 IT, R&D and the education sector in 141–4, 168, 169, 170 Japanese investment in 43, 50 regression analysis on growth in regions of 154–64 trade with Japan 54 Korean Advanced Institute of Science (KAIS) 145 Korean Advanced Institute of Science and Technology (KAIST) 145, 146 Korean Graduate School of Management (KGSM) 146 Korean Industry Promotion Agency (KIPA) 147 Korean Institute of Science and Technology (KIST) 145 Korean Securities Dealers Automated Quotation (KOSDAQ) 148 Korean Software Industry Association (KOSA) 143

Index Korean War (1950–53) 42, 46 Kortum, S. 265 Krugman, P. 3, 97, 173, 174, 182, 199, 268, 269 Kuchiki, A. 3, 99 Kulim industrial zone 102 Kunming 129 Kuroda, A. 174 Kyoeikai parts manufacturers 13 Kyohokai parts manufacturers FDI in North America 27 location of headquarters and factories in Japan 18, 19 ties with Toyota 13–14, 21 Kyosan Denki 30, 31 Kyushu 36 automobile industry in 18, 39, 56, 356 labor migration 269 labor productivity, and IT investment 65–7, 92 Laguna Techno Park 114, 115 language and protocol, online 271–2 Lartkrabang industrial zone 105, 108 laser printers 336–7 Lat Krabang Industrial Estate 111 Latin America and the Caribbean, Japanese investment in 24, 25 Lazerson, M. 249 Leam Chabang 354 ‘leapfrogging’ industrialization 357 leather industry, Italian 209, 216, 231 legally binding cyber agreements 272 Lehmann, E.L. 212 lending incentives 46, 60, 128, 303, 357 Liaoning 178, 180 licensed manufacturing warehouse (LMW) 103 licenses 237, 293, 337 Lichtman, B. 254, 265 Light Industry and Science Park 114, 115 Linh Trung export-processing zone 116 linkages external to cluster 227–8, 238, 239–41, 243, 355, 356 fostering agglomeration 2–3, 60, 182 artiﬁcially produced 195–6 in Brenta shoe district 238–41, 243

371

forward and backward 59, 195–6, 226, 241, 243 in machinery industry in China 174, 183, 195 Lipparini, A. 249, 254, 255 local competition, and agglomeration formation 148, 151–4 in Korea 154–8, 159, 161–6, 168–9 local content requirements 26, 128 local product clusters 353, 355–6 Lohr, S. 293 Lombardia region 251 Lomi, A. 255 Lon Bing EPZ/IZ 116, 117 long-term implicit contracts 15, 16 Lopez, Gerardo 330 Lorenzoni, G. 249 Los Angeles, entertainment cluster in 293 LVMH 231 Lyon, textile cluster in 353, 355–6 M. Thai industrial zone 105, 107, 108 machinery industry, Chinese 183–95 importance for productivity in other industries 183 industrial disparity indices by sector 183–5 inevitability of agglomeration in 195 intermediate goods transactions of 183, 185–9 case study 189–94 inter-regional input–output (IO) tables 187, 190, 191 supply dependence by area 189, 192, 193 share in mining and manufacturing industries 185, 186 machinery industry, Italian comparative advantage of 209, 210, 214, 216 in industrial districts 249, 251–2, 254–5, 256 Maeda, A. 283 mail, communication using distribution by distance 85, 86 and population density 73, 74–7, 79, 83 Malaysia export-processing zones in 105

372

Index

Five Year Plan 104 free trade zones in 102–3 industrial zones in 98, 99, 102–4, 354 Japanese investment in 23, 50, 139 State Economic Development Corporations of 102–4 Mallinckrodt 254 Manila 114, 115, 122 manufacturing industry change in regional industrial disparity in China 175–82 ‘hollowing out’ of 355, 356 industrial disparity indices by sector 182, 184 share of machinery industry output value in 185, 186 Map Ta Phut industrial zone 105, 106–7, 109 Marche region 238, 248 market power 269 marketing function 234, 356 Markusen, A. 244 Marshall, A. 248, 269, 273 Marshallian trinity 3, 37, 60 Marubeni Corporation 105–11, 112, 114, 115 Marukawa, T. 174, 177, 198 Maskus, K. 293 Massachusetts, software cluster in 5, 289, 291, 292, 293, 353 Masscorp of Malaysia 116 Matloﬀ, N. 296 maximum diﬀerentiation, principle of 268–9 Mazda 30, 32, 56, 356 mechanical engineering industry, see machinery industry, Chinese; machinery industry, Italian meetings (face-to-face), communication using and population density 73, 74, 76, 77, 79 relationship between telecommunications and 81–2, 92–3, 149–50, 164, 271, 274–5, 355 empirical evidence for 83–92 Meiji government 40 Meissen porcelain cluster 353

mergers and acquisitions 93, 231, 233, 254 Metropolitan Autonomous University 322 Mexico debt crisis in 314 eﬀect of NAFTA on 312, 316–17, 330–31, 356 growth of GDP per capita on 314–15 IT market in 312, 313 management revolution and emergence of software industry in 5, 312–48, 356 Aguascalientes software cluster 313, 319, 338–43, 344, 353 dialogue and agglomeration experience 327–46 Distrito Federal 312, 313, 317, 319, 320–21, 327–8 driving forces in Mexican software industry 314–21 ﬁrst movers in the IT revolution 321–7 geographical distribution of software market 313 government software niche 328–9 in-house training 321, 322, 336 Jalisco software cluster 313, 319, 335–8, 339 measuring emergence of software industry 317–20 Monterrey software cluster 317–19, 330–35, 336 platform for Internet operations 346–8 size of software market 314, 315, 317, 346, 348 Software Development Program 346, 347, 348 value chain of software industry 317, 318 National Researchers’ System in 319 opening of economy 316 Mexico City, IT community in 320–29; see also Distrito Federal, IT clusters in Meyer, J.-B. 290, 302 Meyer-Stamer, J. 255–6

Index Microsoft 292, 293, 316, 320, 323, 329, 337, 343, 346 Microsoft bCentral 338 Microsoft Oﬃce 337 Microsoft Windows 293 Mie Prefecture 19, 20 Miller, R. 276 Minamata disease 49 minimum diﬀerentiation, principle of 268 mining industry, Chinese change in regional industrial disparity 175–82 industrial disparity indices by sector 184 share of machinery industry output value in 185, 186 Ministry of Information and Communication (MIC), Korea 144, 147, 148, 160, 170 Ministry of Internal Aﬀairs and Communication, Japan 69, 93 Ministry of Land, Infrastructure and Transport, Japan 68, 70 Mir, A. 303 Mirondola, biomedical district of 250, 252–5 Missouri, Toyota plants and parts suppliers in 27 Mitsubishi Corporation 31, 105–11, 114, 115, 116, 117, 118, 130 Mitsui & Co., Ltd 105–11, 112, 114, 115, 116, 117, 118 Mitsui Hightech 36 MIX (Milan Internet Exchange), peering decision of ISPs connected with 279–83 Miyazawa, Kiichi 125 Miyoshi 19 MM2100 EPZ/IZ 112, 114 Modena 253, 256 Modiano, P. 219 Monden, Y. 13 monodukuri 58 monopolistic competition 3, 173 Montelongo, Roberto 331 Monterrey software cluster 317–19, 330–35, 336 Moore, S. 299 Moore’s Law 294, 296

373

Mori, T. 97, 173 motorcycle industry 129 Moussanet, M. 250 ‘multi-located’ industrial districts 257–8, 264–5, 356 multinational corporations (MNCs) eﬀect on clusters 355 in industrial districts 253–4, 255, 264–5 in industrial zones 354 in Mexico 314, 319, 320, 335–6, 338, 340–41, 343–6 multi-plant ﬁrms, innovative capability of 259, 260, 261, 263 Murakami, N. 183, 185, 187, 188 Murphy, K.M. 199 musical instruments industry, Italian 216 Myanmar, industrial zones in 116, 118 Mytelka, L. 227 mythical-man month 308 Nacional Financiera (NAFIN) 328, 340 NAFTA, see North American Free Trade Agreement (NAFTA) Nagoya IT cluster in 68, 70 population changes in 68, 69 Toyota parts suppliers in 19, 20 Nagoya Port 18 NAMEX 279 NASA 55 National Autonomous University of Mexico (UNAM) 322, 323, 346 National Bureau of Statistics of China 181 National Science Board (NSB) 300, 301 National Science Foundation (NSF) 301 national security 299 nationalization of banks 322 natural resources 23, 37, 354 nearshore services 331, 356 Neoris Corporation 333, 334, 335 NET México 346 netiquette 272 Netscape 293 network access points (NAPs) 276, 279

374

Index

network externalities 248, 249, 275, 291, 293 New Duong Bridge 121, 124 new economic geography 2, 173–4, 214, 273, 279 New United Motor Manufacturing, Inc. (NUMMI) 23 New York City, multimedia software cluster in 293 newspapers, communication using, and population density 73, 74, 75, 78, 79, 82 Ng, L.F.Y. 174 Nicholson, B. 289, 290, 302, 303, 307 Nihon Keizai Shimbun 139 Ningxia 177 Nippon Denso 13 Nippon Steel Corporation 40, 41, 43, 44, 45, 47, 48–9, 50, 60 Nishi-Mikawa 18, 19, 20 Nishikimi, K. 97, 173 Nishio 19, 20 Nishiwaki, T. 33 Nishizawa, M. 174 Nissan Motor Co. 23, 30, 32, 36, 39, 56, 340–41, 356 Nisshin Steel Co. 44, 45, 50 Nissho Iwai Corporation 111, 112, 114, 115 NKK Corporation 50 Noi Bai EPZ/IZ 116, 117 Noibai International Airport 120, 128 Nomura Haiphong industrial zone 116, 117, 119–20, 125, 135 tenant companies 129–34 Nomura Securities Co., Ltd 116, 117, 119 non-codiﬁable knowledge 270 non-metallic mineral manufacturing industry, Italian 216 non-proﬁt organizations (NPOs) 58–9, 60, 67 North American Free Trade Agreement (NAFATA) 312, 316, 330, 331, 356 Nuevo Leon, IT clusters in 312, 313, 317–19, 330–35, 336 object-oriented programming 295 OECD, see Organization for Economic

Cooperation and Development (OECD) Oﬃce of Technology Assessment (OTA) 289, 293 Oﬃce Xpress Manufacturing Co., Ltd 129, 132 oﬃcial development assistance (ODA) for transportation infrastructure 98, 121, 135, 137 Ohara, M. 198 Ohbu 19 Ohio, Toyota plants and parts suppliers in 27 Ohno, T. 13 Ohta Ward, Tokyo 353 oil shocks 37, 43, 47, 59 Okamoto, N. 187, 190 Okazaki 19 one-stop service 125 Onida, F. 219 Onoe, E. 174 Open Service 333, 335 Oracle 320, 329, 343–6 Oracle Classroom 346 Organization for Economic Cooperation and Development (OECD) 46, 206, 207, 208, 219, 223–4, 289–90, 292, 298, 300, 307 organizational changes 65–7 organizational communities 255 origin–destination (OD) matrix 74–6, 85, 87, 89 Ornati, O. 249 Osaka communication use in 78, 79 Higashi Osaka in 353 IT cluster in 68, 70 population changes in 68, 69 Osborne, M.J. 283 Otsuka, K. 173, 198 Owari group 21–2 Owari region 19, 20 Paba, S. 252, 255 Paciﬁc War (1941–45) 42 packaged software 291, 293, 317, 323, 324, 331, 335, 339 Padoan, P.C. 215 Pambianco 231 paper industry, Italian 216

Index Parker Processing VN Co. 129, 130, 136 parts suppliers, Toyota’s, see primary parts manufacturers, Toyota’s; secondary parts manufacturers, Toyota’s; tertiary parts manufacturers, Toyota’s patents hardware and software 298 information ﬂows from 72 inter-ﬁrm diﬀerences in patenting activity 248, 258–64 path dependency 18, 249 Pavitt, K. 259, 260 Pearl River Delta 191 peering agreements 275, 276, 280–82 peering decision 277–9 agglomerating forces on 279–83 peering matrix 280–81 Pemex 343 Penang 102, 103, 354 Penang Skills Development Center 102 Penang State Economic Development Corporation 102 Peneder, M. 312 People’s Republic of China, see China Perak, industrial zones in 102 personal computers (PCs) domestic shipment of 64 ownership per 1000 residents and software export success 301, 302, 303, 304 software developed for 314, 316, 317, 327, 336 Phan Van Khai 125 Philippines export-processing zones in 115 industrial zones in 98, 111, 114, 115 Japanese investment in 23, 111, 114, 139 taxes in 126 wages in 121, 122 Piergiovanni, R. 258 pillar industries, development of 174, 196, 197 Piore, M.J. 249 Pitchik, C. 283 Pittsburgh, iron and steel agglomeration in 56–9, 353 Plaza Accords 43, 104

375

Pohang Steel 43, 50 Polanco 321, 322 pollution 37, 49, 53, 56, 57 POLYNE de México 327, 329 population density, communication technologies and 74–81, 82, 83 porcelain cluster in Meissen, Germany 353 port facilities 40, 41, 59, 80, 98, 137, 357 Cairong port 120, 121, 123, 124 Haiphong port 98, 120, 124, 128, 134, 135 Hibikinada Hub Port Initiative 39, 54 Kitakyushu 54, 58 Porter, M. 4 Portugal, Japanese investment in 23 pottery and china industry, Italian 209 Powell, W. 273 PPND (Pittsburgh Partnership for Neighborhood Development) 58 Prada 231 Prai industrial zone 102 Prasad, M. 306 price negotiations, in Japanese automobile industry 16, 21 primary parts manufacturers, Toyota’s 10–12 automotive assemblers supplied by 28, 30–31 FDI in North America 26–7 location of headquarters and factories in Japan 18, 19 ties with Toyota 13–14, 21 principal–agent model 15, 16 principle of maximum diﬀerentiation 268–9 principle of minimum diﬀerentiation 268 printing equipment clusters 4 private companies external purchase of parts by 188, 189 industrial zones established by 98, 99, 103, 105–18, 134, 354 private–public partnerships 341 privatization 330 PRM software, see provider relationship management (PRM) software

376

Index

probability density function (PDF) 210, 212–14 process upgrading 228 Prodigy network 326 Prodigy Pymes 323 producer-driven chains 227 product design 227, 232–3, 242, 356 product diﬀerentiation, price competition with 268–9 product innovation 254–5 product upgrading 228, 236 production eﬃciency 16–17 production function 100–101, 150 production process 227 upgrading 228 productivity paradox 64–7, 92 Prometeia 256, 257 property, plant and equipment, value of, and innovative capability 259–60, 261, 263 property rights 125 intellectual (IPRs) in software 291, 293, 301, 303, 304, 305 provider relationship management (PRM) software 329 Puga, D. 269 pure transfer of technology (PTT) 321 Pyke, F. 244 Qingdao 176 Qinghai 177, 195 Qiu, B. 174 quality assessments 335, 338 quality circles 316 quality control 254, 316 quality management (QM) 17, 21 quasi-hierarchies 227–8, 237, 242 R&D, see research and development (R&D) R&I (Ricerche and Interventidi Politica Industriale) 253 R&T Manufacturing VN Co., Ltd 129, 132 Rabellotti, R. 226, 238, 244 railways fares 89 planned projects in China 195 proximity to 68, 80 raw materials, proximity to 40, 41, 59

Ray, S. 299 Readman, J. 227 recycling 39, 44, 54, 56, 58, 59–60, 356 regional growth 148–64 basic model of growth in regions 150–53 educational and R&D institutions and 168, 169, 170 IT sector in Korea 161–70 knowledge creation, agglomeration and IT 149–50 mechanism of agglomeration formation 148–9 regression analysis on growth in regions of Korea 154–64 results of previous studies 153–4 regional specialization, and agglomeration formation 148–9, 151–4 in Korea 154–8, 159, 161–9 related parts manufacturers 10 relational contracts 273 relational governance 272–5 Renaissance Project 57 rents, economic 231, 236, 237, 238, 243 reputation 239, 272–5 research and development (R&D) controlled by MNCs 254 facilities 3 and growth of IT sector in Korea 168, 169, 170 joint activities 12, 13, 19 location of MNCs’ 341 national research laboratories 145 scientists and engineers working in 301–2, 304, 305, 307 shared in industrial clusters 354 software 298 workers employed in Korean R&D sector 142–3 restaurants 123, 137 restricted or prohibited industries 128 Reyes, Carlos 334, 335 risk sharing 17 roads, see highways Robb, D. 306 robotics, industrial clustering in 4 Romania, outsourcing to 236, 237, 241, 242 Rosenberg, N. 252

Index Rosenstein-Rodan, P.N. 195 Rossi, S. 219 Route 128 software cluster 5, 289, 291, 292, 353 Russo, M. 256 Russo-Japanese War 41 Saai. M3 327 Sabah, industrial zones in 104 Sabel, Ch. 249 Sakai Works 40, 43, 50 Sakakibara, M. 145 Sakata, I. 147 Sala-i-Martin, X. 100 Salim Group of Indonesia 116, 117 San Martin Associates 322 Sango 30, 31 Sankyu Co. 36 Santa Catarina ceramic tile cluster 256 Santarelli, E. 252, 258, 264 Santomas VN Co. 129, 130 SAP 320, 329, 343, 345 Sapporo City, IT cluster in 70, 93 Sarawak, industrial zones in 104 Sassuolo, ceramic tile district of 250, 255–8 satellite television, communications using, and population density 73, 74, 75, 78, 82 Saxenian, Annalee 273, 289, 298, 302, 307 scale-intensive ﬁrms, innovative capability of 259, 260, 261, 263 Schmitz, H. 226, 227, 228, 244, 245 schools 120, 123, 329, 346 Schware, R. 290 Schwartz, J. 299 SCM software, see supply chain management (SCM) software SDINET 327 search costs 74 Sebu 122 second Schumpeterian hypothesis 259 Second World War 41, 68, 214, 228, 247 secondary parts manufacturers, Toyota’s 12 location of factories 18, 20 ties with Toyota 14 Seki, M. 174

377

Selangor 102, 103 Select IDC 313 Sendai City, IT cluster in 70, 93 Seoul concentration of industries in 140, 141 concentration of software industry in 143–4, 164, 165, 170 growth of hardware and software industries in 165 number of workers in IT, R&D and education in 143 tele-density and Internet penetration in 160 sewage processing plants 120 Sforzi, F. 250 Shaanxi 177, 196 Shabondama Soap 36 Shadlen, K. 293 Shandong 178, 179, 180 Shanghai intra-China parts purchasing from 189 joint venture with Brenta shoe manufacturers 239 manufacturing industry based in 176 share of production base 178, 179, 180 wages in 122 Shanghai Baoshan Steel 43, 44, 50 Shanxi 177, 178 Shenzhen 122 Shibuya Bit Valley 1, 68 Shizuoka Prefecture 20 shopping facilities 123, 137 Shukurou 61 Sialkot medical products cluster 354–5 Sichuan 177 Sigma Tao 326, 330 Silicon Valley 1, 5, 68, 71, 140, 150, 273, 289, 291, 298, 305, 353 Silverman, B.W. 212 Singapore 139, 353 SISTEC 328 Sistema Nacional de Investigadores 319 Sistemas Dinámicos Internacionales (SDI) 320, 326–7 slope production method 46 Small- and Medium-scale Enterprise Agency 4

378

Index

small and medium-sized enterprises (SMEs) in Aichi Prefecture 21–2 government policy towards 354 in industrial districts 247, 249, 250, 253, 256 software developed by 319 software development for 316, 317, 323, 329, 334–5, 336, 337–8 SMBA 147, 148 Smithonian monetary system 47 ‘smoke towns’ 56, 57 social networks 273–4 social sanctions 274, 275, 283 Softtek 319, 330–34, 335, 336 software industry developmental advantages of 289–90 emergence in Mexico 5, 312–48, 356 Aguascalientes software cluster 313, 319, 338–43, 344, 353 dialogue and agglomeration experience 327–46 Distrito Federal 312, 313, 317, 319, 320–21, 327–8 driving forces in Mexican software industry 314–21 ﬁrst movers in the IT revolution 321–7 geographical distribution of software market 313 government software niche 328–9 in-house training 321, 322, 336 Jalisco software cluster 313, 319, 335–8, 339 measuring emergence of software industry 317–20 Monterrey software cluster 317–19, 330–35, 336 platform for Internet operations 346–8 size of software market 314, 315, 317, 346, 348 Software Development Program 346, 347, 348 value chain of software industry 317, 318 Indian ‘leapfrogging’ industrialization 357 software cluster in Bangalore 1, 140, 316, 353

software development for US market 289, 331 institutional framework required for 312–14 Korean characteristics of 161–8 concentration in Seoul 143–4, 164, 165, 170 factors promoting growth and industrial agglomeration 144–8, 168–70 number of establishments and workers in 141–2, 143 skilled labor necessary for 2, 294–9, 300–302, 303, 304, 305, 319, 331 spending by principal region and world total 297 United States 5, 289–307 centrifugal forces aﬀecting 291–9 centripetal forces aﬀecting 299–300 challenges to North American monopoly 289, 356 correlates of North American software imports 300–307 hardware-to-software migration 296–8 labor shortages 294–9 ‘lock-in’ mechanisms 291–3 in Pittsburgh 57 principal segments 291 top software ﬁrms located in US 291, 292 wages 295, 296, 298 see also Silicon Valley Sogoshosha 98, 134 SOHO (small oﬃces, home oﬃces) 67 Solingen cutlery cluster 353 Solow, R. 65 Solow paradox, see productivity paradox Sonobe, T. 173, 198 ‘Space World’ 55 Spagnolo, G. 273, 274 Spain Castellón ceramic tile cluster in 255–6 comparative advantage of 208, 209 spatial economics 2, 173–4, 182

Index Special Free Zone for International Distribution 54 specialized parts manufacturers 10–12 specialized suppliers, innovative capability of 259, 260, 261 spot markets 275 Stalin, Joseph 177 standards, software 293, 296 Stanford Computer Industry Project (SCIP) 291 State Economic Development Corporations of Malaysia 102–4 state-owned enterprises (SOEs) 176, 187, 188, 189, 328–9 steel industry cluster, see Kitakyushu City industrial cluster; Yawata Works Steinmuller, W.E. 291, 293 Sterlacchini, A. 252 Stiglitz, J.E. 173, 182 stock holdings 12, 13 strategic alliances 323, 337 subcontractors externalities derived from 354 to high fashion companies 232–8 in industrial districts 253, 254–5, 273, 274–5 Japanese automobile industry 12, 356 see also primary parts manufacturers, Toyota’s; secondary parts manufacturers, Toyota’s; tertiary parts manufacturers, Toyota’s substitution eﬀects, face-to-face meetings and IT 81, 84, 85, 89, 92 Suehiro, A. 110 Sueyoshi, Koichi 53 Sumitomo Bakelite 128, 130, 136 Sumitomo Coil Center 129 Sumitomo Corporation 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 128, 129 Sumitomo Metal Industries Ltd 44, 45, 50, 136 Sun Microsystems 292, 298, 343, 346 supplier dominated ﬁrms, innovative capability of 259, 260, 261 supply chain management (SCM) software 64

379

Suranaree industrial zones 105, 107, 108, 109 surplus change 346 Sustainable Pittsburgh 58 Suzuki 31 tacit knowledge 5, 144, 255, 269–70, 354 Taiwan industrial clusters in 97 industrial zones in 354 Japanese investment in 23 trade with Japan 54 Takahama 19, 20 Takashimaya Nippatsu 30, 31 Tamberi, M. 207, 215, 219 Tan Tuan export processing zone 116 tax and lending incentives 357 in industrial parks in Mexico 340 in industrial zones 98, 99, 111, 125–8, 134, 137 in Japan 46, 60 in Korea 148 for software development 303 techno poles 329, 338, 341 Technological University of Netzahualcoyotl 322, 346 Technoparks 39, 53 Teheran Valley 144, 170 telecommunications relationship between face-to-face meetings and 81–2, 92–3, 149–50, 164, 271, 274–5, 355 empirical evidence for 83–92 software to hardware spending ratio 298 tele-density and regional growth in Korea 158–64, 166–8, 169 see also cellular phones; telephone, communication using telecommuting 67 Teléfonos de Mexico (Telmex) 323 privatization of 330 software network of 324–6, 329 telephone, communication using and population density 73, 75, 77, 79, 83 relationship between face-to-face meetings and 85–92, 149–50 telephone circuits 120

380

Index

television, communication using, and population density 73, 74, 75, 78, 79, 82 Tennessee, Toyota plants and parts suppliers in 27 Teran, H. 303 Terdiman, R. 298, 299, 300, 303 Terengganu, industrial zones in 103 tertiary education, and software export success 300–301, 303, 304, 305 tertiary parts manufacturers, Toyota’s 12, 14 Tesser, S. 291, 294, 295, 296 Texas, hardware-related software cluster in 293 Texas Instruments (TI) 340, 341 textile industry external purchase of parts in Chinese 187, 188 French 353, 355–6 Italian 209, 216 Japanese 21–2, 355 software developed for 327 Thailand export-processing zones in 104, 105, 111 Industrial Estate Authority of 104–5, 106–10, 134 industrial zones in 98, 99, 104–5, 106–10, 111, 134, 354 Japanese investment in 23, 111, 139 taxes in 126 trade with Japan 54 wages in 121, 122, 123 Thang Long industrial park 116, 117, 119–20, 122, 125, 135 tenant companies 128, 129, 130–31 theme parks 54–5 thick markets 3, 60 Thisse, J.-F. 3, 173, 199, 268, 269, 270, 354, 357 Tianjin external purchase of parts to total external purchase ratio in 187, 189 manufacturing industry based in 176 share of production base 178, 180 Tobata Foundry 36 Tobata site 40, 42, 43, 44, 46, 50, 55, 56 Tohoku 18

Tokai 19 Tokaido Shin-kansen (bullet train) 46 Tokyo communication use in 78, 79 IT cluster in 68, 70 Ohta Ward in 353 population changes in 68, 69 Tokyo Metropolitan Highway 46 TOPIX 279–80 Torii, K. 103 total quality management (TQM) 17, 21 TOTO Ltd 36, 128, 130, 136 Towa Real Estate 13 township and village enterprises 188 toy industry, Italian 216 Toyoda, Sakichi 354 Toyota Boshoku 13 Toyota Central Research Institute 13 Toyota City 18, 19, 20, 353, 354 Toyota Gosei 13, 30, 31 Toyota Group parts manufacturers 13, 17 automotive assemblers supplied by 28, 30–31 FDI in North America 27 location of factories and headquarters in Japan 18, 19 Toyota Jido Shokki 13 Toyota Koki 13 Toyota Motor Corporation 4 FDI by Toyota and its parts suppliers 22–7 and dispersion of location 28, 30–31, 32 in industrial zones in Vietnam 128, 136 IT applications in 29–32 location of factories and parts suppliers in Japan 18–22, 39, 56, 356–7 origin of 354 production structure of 13–14 compared with non-hierarchical US system 14–17 Toyota Motor Manufacturing, Illinois, Inc. (TMMI) 23, 26 Toyota Motor Manufacturing, Kentucky, Inc. (TMMK) 23, 26 Toyota Shatai 13

Index Toyota Tsusho 13 Trade-Related Aspects of Intellectual Property Rights (TRIPs) agreement 293 Training Center for Human Resource Development (Malaysia) 102 transaction costs, reduction of 15, 258, 274, 299 transfer pricing 308 transit agreements 276, 278 transport density 97 transportation, relationship between telecommunications and demand for 83–92 transportation costs agglomeration to save 2, 3, 5, 149, 182, 183, 197, 354 in Japanese automobile industry 39 in Japanese steel industry 60 and demand for face-to-face interaction 85, 89, 279 iceberg transport 173–4 Japanese parts manufactured in US 26 linear 268 methods of reducing 125 quadratic 268–9 technology-driven reduction in 270, 271 transportation infrastructure industrial agglomeration assisted by 59, 80, 98, 99, 121–3, 124, 134, 135, 137 see also airline connections; highways; port facilities; railways Traü, F. 215 trigger-price system 49 TRIPs agreement 293 TRIPs-plus 293 trust 82, 272, 274, 275, 279 Tsubame City kitchenware cluster 353, 355–6 Tsuda Industries 30, 31 Tsuji, M. 1, 3, 14, 33, 357 Tuan, C. 174 Tuomi, I. 308 Tuscany 248 two-sector priority production method 46

381

Uchida, T. 198 Udon Thani industrial zone 104 Ueki, Y. 139, 140 United Kingdom Brenta shoe district’s sales to 229 comparative advantage of 208, 209 Industrial Revolution in 63–4 Japanese investment in 23 United Nations Conference on Trade and Development (UNCTAD) 291, 292, 294, 298, 299, 302 United Nations Development Program (UNDP) 301 United States Brenta shoe district’s sales to 230 comparative advantage of 208, 209 FDI by Toyota and parts manufacturers in 23–7 FDI in Mexico 312 Japanese steel exports to 49 labor productivity and IT investment in 65–7 production structure of automobile industry in 14–17 software industry in, see software industry United States Central Intelligence Agency (USCIA) 301 United States Department of Commerce/Economics and Statistics Administration (USDOC/ESA) 294, 295 United States Immigration and Naturalization Service (USINS) 295, 296 universal connectivity 275, 283 Universidad Autónoma de Veracruz (UAV) 346 Universidad del Valle de México (UVM) 346 Universidad Nacional Autonoma de México (UNAM) 322, 323, 346 Universidad Tecnológica de México (UNITEC) 346 Universidad Tecnológica Fidel Velazquez de Netzahualcoyotl 322, 346

382

Index

universities government-funded R&D in 145 industry linkages with 121, 140, 146–7, 334, 346 oﬀering IT degrees 319 software engineers graduating from 322, 323, 334, 337 in urban centers 93, 120, 150 University of Monterrey (UDEM) 334 Uno, K. 196 upgrading strategies 228 Urata, S. 97 urban collective-owned enterprises 187, 188, 189 Urban Redevelopment Agency of Pittsburgh (URAP) 58 US Air Force 150 US Bureau of Labor Statistics 66, 296 US Steel Corporation 42, 56, 57 USSR, relations with China 176–7 utility function 100 value chains globalization, industrial districts and 5, 225–43 Brenta in the top brand chain 232–8, 356 Brenta shoe district 228–9 Brenta’s customers 229–30 implications of globalization 241–3 literature on 226–8 local governance in Brenta 238–41 top brand value chain 230–32 governance or coordination of activities in 227–8, 237 Mexican software industry 317, 318 ‘vein’ industries 39, 54, 60, 61, 356 Venables, A.J. 269 vendor development programs 341 Veneto 248 venture capital 148, 353 venture companies 146–7, 170 Veronesi, Mario 254 Vietnam export-processing zones in 116, 117, 119 FDI inﬂows in 2001 119 growth rates in 118–19, 137

industrial clusters in Northern Vietnam 128–34, 135, 136–7 industrial zones in 97, 98, 354 development by quasi-public sector 114–18 northern Vietnam 118–34, 135, 137 institutional reforms in 125–8, 134 Japanese investment in 23, 98, 116, 119, 120, 123, 125, 128, 129, 130, 131, 132, 133, 135, 136–7 taxes in 125, 126, 134, 137 wages in 121, 122, 123 Vietnam Singapore Industrial Park (VSIP) 116, 117 Vietphong Garment & Textile Co., Ltd 129, 132 Vinh Phuc industrial zone 128 visa application 128 vocational training and education 354 Volex Cable Assembly 129, 130 Vollrath, T.L. 219 voluntary export restraints (VERs) 43, 49 VSSB (Malaysia) 116, 117 wages in hardware industry 295 in software industry 295, 296, 298, 322, 327, 331, 337 in Vietnam, compared with other Asian countries 121, 122, 123 WARP, see Worldwide Automotive Realtime Purchasing System (WARP) water supplies 59, 120 waterfall models 295 Wei, H. 174 Weijland, H. 97 Wen, M. 174 Wichmann, T. 283 Williamson, O.E. 15, 255, 273 willingness to pay 236 Word Perfect 293 World Bank 97, 301 World Information Technology and Services Alliance (WITSA) 297, 298, 301, 348 World Trade Organization 293 WorldCom 278

Index Worldwide Automotive Realtime Purchasing System (WARP) 29–32 Wuhan 187, 188 Xerox 340–41 Xi’an 196 Xu, X. 174 Yamada, K. 99 Yamasha 128, 136 Yantze River Delta 180, 181, 191 Yaskawa Electric Co. 36 Yawata Iron and Steel Co. 40, 43, 47 Yawata Works 40 choice of location for 354–5 coalﬁelds owned by 35 downsizing in Kitakyushu 37, 39

383

history 39–53 investment in environmental protection 39 new businesses set up by 54–6, 356 Yohai, V. 220 Yokkaichi City 353 Yokohama City, IT cluster in 70 Yoon, Myoung-hun 146, 147 Yue, X. 173 Yunnan 177 zaibatsu 41 Zenrin Co. 36 zero inventory 316 Zhejiang 178, 179, 180 Zhong Guan Cun IT cluster 353 Zhu Xiwei 174 Zysman, J. 293